Methods for cancer detection and monitoring by means of personalized detection of circulating tumor DNA

Abstract

To provide methods for detecting single nucleotide variants in breast cancer, bladder cancer, or colorectal cancer; and to provide additional methods and compositions, such as reaction mixtures and solid supports comprising clonal populations of nucleic acids.SOLUTION: For example, provided here is a method for monitoring and detection of early relapse or metastasis of breast cancer, bladder cancer, or colorectal cancer, comprising :generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from a sample of blood or urine or a fraction thereof from a patient who has been treated for a breast cancer, bladder cancer, or colorectal cancer, each amplicon of the set of amplicons spanning at least one single nucleotide variant locus of a set of patient-specific single nucleotide variant loci associated with the breast cancer, bladder cancer, or colorectal cancer; and determining the sequence of at least a segment of each amplicon of the set of amplicons that comprises a patient-specific single nucleotide variant locus, detection of one or more patient-specific single nucleotide variants being indicative of early relapse or metastasis of breast cancer, bladder cancer, or colorectal cancer.SELECTED DRAWING: None

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Application No. 62 / 657,727 filed April 14, 2018, U.S. Provisional Application No. 62 / 669,330 filed May 9, 2018, U.S. Provisional Application No. 62 / 693,843 filed July 3, 2018, U.S. Provisional Application No. 62 / 715,143 filed August 6, 2018, U.S. Provisional Application No. 62 / 746,210 filed October 16, 2018, U.S. Provisional Application No. 62 / 777,973 filed December 11, 2018, and U.S. Provisional Application No. 62 / 804,566 filed February 12, 2019. Each of these above-cited applications is incorporated herein by reference in its entirety. [Background technology]

[0002] Detecting early recurrence or metastasis of cancer has traditionally relied on diagnostic imaging and tissue biopsy. Tumor tissue biopsy is invasive and carries the risk of potentially contributing to metastasis or surgical complications, while diagnostic imaging-based detection is not sensitive enough to detect early recurrence or metastasis. Better, less invasive methods are needed to detect cancer recurrence or metastasis. Summary of the Invention [Problem to be solved by the invention]

[0003] One aspect of the invention described herein is a method of monitoring and detecting early recurrence or metastasis of cancer (e.g., breast cancer, bladder cancer, or colon cancer), comprising generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from a blood or urine sample, or a fraction thereof, from a patient treated for cancer (e.g., breast cancer, bladder cancer, or colon cancer), wherein each amplicon in the set of amplicons is a single nucleotide variant locus of at least one of a set of patient-specific single nucleotide variant loci associated with the cancer (e.g., breast cancer, bladder cancer, or colon cancer). and determining the sequence of at least one segment of each amplicon of a set of amplicons that comprise a patient-specific single nucleotide variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants is indicative of early recurrence or metastasis of cancer (e.g., breast cancer, bladder cancer, or colon cancer). [Means for solving the problem]

[0004] In addition to breast cancer, bladder cancer, and colorectal cancer, the methods described herein can be used to treat other types of cancer, such as acute lymphoblastic leukemia, acute myeloid leukemia, adrenocortical carcinoma, AIDS-related cancer, AIDS-related lymphoma, anal cancer, appendix cancer, astrocytoma, atypical teratoma / rhabdoid tumor, basal cell carcinoma, brain stem glioma, brain tumors (brain stem glioma, central nervous system atypical teratoma / rhabdoid tumor, central nervous system embryonal tumor, astrocytoma, craniopharyngioma, ependymoblastoma, ependymoma, medulloblastoma, medulloepithelioma, moderately differentiated pineal parenchymal tumor, supratentorial primitive neuroectodermal tumor, and pineal bud tumor). including bronchial tumors, Burkitt's lymphoma, carcinoma of unknown primary site, carcinoid tumor, carcinoma of unknown primary site, atypical teratoma / rhabdoid tumor of the central nervous system, central nervous system embryonal tumor, cervical cancer, childhood cancer, chordoma, chronic lymphocytic leukemia, chronic myeloid leukemia, chronic myeloproliferative disorder, colon cancer, craniopharyngioma, cutaneous T-cell lymphoma, pancreatic endocrine islet cell tumor, endometrial cancer, ependymomas, ependymomas, esophageal cancer, nasal neuroblastoma, Ewing's sarcoma, extracranial germ cell tumor, extragonadal germ cell tumor, extrahepatic bile duct cancer, gallbladder cancer, gastric cancer cancer), gastrointestinal carcinoid tumor, gastrointestinal stromal cell tumor, gastrointestinal stromal tumor (GIST), gestational trophoblastic tumor, glioma, hairy cell leukemia, head and neck cancer, cardiac cancer, Hodgkin's lymphoma, hypopharyngeal cancer, intraocular melanoma, islet cell tumor, Kaposi's sarcoma, kidney cancer, Langerhans cell histiocytosis, laryngeal cancer, lip cancer, liver cancer, malignant fibrous histiocytoma / bone cancer, medulloblastoma, medulloepithelioma, melanoma, Merkel cell carcinoma, Merkel cell skin carcinoma, mesothelioma, metastatic squamous cell carcinoma of the neck of unknown primary origin, mouth cancer, multiple endocrine neoplasia syndrome, multiple myeloma, multiple myeloma / plasmacytoma, mycosis fungoides, myelodysplastic syndrome, myeloproliferative neoplasm, nasal cancer, nasopharyngeal carcinoma, neuroblastoma, non-Hodgkin's lymphoma, non-melanoma skin cancer, non-small cell lung cancer, oral cancercancer), oral cancer, oropharyngeal cancer, osteosarcoma, other brain and spinal cord tumors, ovarian cancer, epithelial ovarian cancer, ovarian germ cell tumor, ovarian low malignant potential tumor, pancreatic cancer, papillomatosis, sinonasal cancer, parathyroid cancer, pelvic cancer, penile cancer, pharyngeal cancer, moderately differentiated pineal parenchymal tumor, pineoblastoma, pituitary tumor, plasmacytoma / multiple myeloma, pleuropulmonary blastoma, primary central nervous system (CNS) lymphoma, primary hepatocellular carcinoma, prostate cancer, rectal cancer, kidney cancer, renal cell (kidney) cancer, renal cell carcinoma, airway cancer, retinoblastoma, rhabdomyosarcoma, salivary gland cancer, Sézary syndrome, small cell lung cancer, small intestine cancer, soft tissue sarcoma, squamous cell carcinoma, cervical squamous cell carcinoma, gastric cancer It can also be used to monitor and detect early recurrence or metastasis of uterine cancer, supratentorial primitive neuroectodermal tumor, T-cell lymphoma, testicular cancer, throat cancer, thymic carcinoma, thymoma, thyroid cancer, transitional cell carcinoma, transitional cell carcinoma of the renal pelvis and ureter, trophoblastic tumor, ureteral cancer, urinary tract cancer, uterine cancer, uterine sarcoma, vaginal cancer, vulvar cancer, Waldenstrom's macroglobulinemia, or Wilms' tumor.

[0005] In some embodiments, prior to determining the sequence of at least one segment of each amplicon of the set of amplicons for the blood or urine sample or fraction thereof, nucleic acid is isolated from the patient's tumor, and somatic mutations are identified in the tumor for the set of patient-specific single nucleotide variant loci, and single nucleotide variants.

[0006] In some embodiments, the methods involve longitudinally collecting and sequencing blood or urine samples from the patient.

[0007] In some embodiments, at least two or at least five SNVs are detected, and the presence of at least two or at least five SNVs is indicative of early recurrence or metastasis of breast, bladder or colon cancer.

[0008] In some embodiments, the breast cancer, bladder cancer, or colon cancer is stage 1 or stage 2 breast cancer, bladder cancer, or colon cancer. In some embodiments, the breast cancer, bladder cancer, or colon cancer is stage 3 or stage 4 breast cancer, bladder cancer, or colon cancer.

[0009] In some embodiments, the individual has been treated with surgery prior to isolation of the blood or urine sample.

[0010] In some embodiments, the individual has been treated with chemotherapy prior to isolation of the blood or urine sample.

[0011] In some embodiments, the individual has been treated with an adjuvant or neoadjuvant prior to isolation of the blood or urine sample.

[0012] In some embodiments, the individual has been treated with radiation therapy prior to isolation of the blood or urine sample.

[0013] In some embodiments, the method further comprises administering to the individual a compound known to be specifically effective in treating breast, bladder, or colon cancers that have one or more of the determined single nucleotide variants.

[0014] In some embodiments, the method further comprises determining a variant allele frequency for each of the single nucleotide variants from the sequencing.

[0015] In some embodiments, a treatment regimen for breast, bladder, or colon cancer is identified based on determining variant allele frequencies.

[0016] In some embodiments, the method further comprises administering to the individual a compound known to be specifically effective in treating breast, bladder, or colon cancer in which one of the single nucleotide variants has a variable allele frequency greater than at least half of the other determined single nucleotide variants.

[0017] In some embodiments, the sequence is determined by high-throughput DNA sequencing of multiple single nucleotide variant loci.

[0018] In some embodiments, the method further comprises detecting clonal single nucleotide variants in breast cancer, bladder cancer, or colon cancer by determining a variant allele frequency for each SNV locus based on the sequences of multiple copies of the set of amplicons, wherein a relatively high allele frequency of the plurality of single nucleotide variant loci compared to other single nucleotide variants is indicative of a clonal single nucleotide variant in breast cancer, bladder cancer, or colon cancer.

[0019] In some embodiments, the method further comprises administering a compound to the individual that targets one or more clonal single nucleotide variants but not other single nucleotide variants.

[0020] In some embodiments, a variant allele frequency of greater than 1.0% is indicative of a clonal single nucleotide variant.

[0021] In some embodiments, the method further includes forming an amplification reaction mixture by combining a polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library created from the sample, and a set of primers that each bind within 150 base pairs of the single nucleotide variant locus or a set of primer pairs that each span a region of 160 base pairs or less that includes the single nucleotide variant locus, and subjecting the amplification reaction mixture to amplification conditions to generate a set of amplicons.

[0022] In some embodiments, determining whether a single nucleotide variant is present in the sample comprises identifying a confidence value for each allele call at each of the set of single nucleotide variant loci based at least in part on the read depth for the loci.

[0023] In some embodiments, a single nucleotide variant call is made if the confidence value for the presence of the single nucleotide variant is greater than 90%.

[0024] In some embodiments, a single nucleotide variant call is made if the confidence value for the existence of the single nucleotide variant is greater than 95%.

[0025] In some embodiments, the set of single nucleotide variant loci includes all of the single nucleotide variant loci identified in the TCGA and COSMIC datasets for breast, bladder, or colorectal cancer.

[0026] In some embodiments, the set of single nucleotide variance sites includes all of the single nucleotide variance sites identified in the TCGA and COSMIC datasets for breast cancer, bladder cancer, or colon cancer.

[0027] In some embodiments, the method is performed at a depth of read for a set of at least 1,000 single nucleotide variant loci.

[0028] In some embodiments, the set of single nucleotide variant loci comprises between 25 and 1000 single nucleotide variant loci known to be associated with breast cancer, bladder cancer, or colon cancer.

[0029] In some embodiments, the efficiency and error rate per cycle is determined for each amplification reaction of a multiplex amplification reaction of single nucleotide variant loci, and the efficiency and error rate are used to determine whether a single nucleotide variant at a set of single variant loci is present in a sample.

[0030] In some embodiments, the amplification reaction is a PCR reaction and the annealing temperature is 1-15° C. higher than the melting temperature of at least 50% of the primers in the set of primers.

[0031] In some embodiments, the amplification reaction is a PCR reaction, and the length of the annealing step in the PCR reaction is 15 to 120 minutes.

[0032] In some embodiments, the amplification reaction is a PCR reaction, and the length of the annealing step in the PCR reaction is 15 to 120 minutes.

[0033] In some embodiments, the primer concentration in the amplification reaction is 1 to 10 nM.

[0034] In some embodiments, the primers in the primer set are designed to minimize primer-dimer formation.

[0035] In some embodiments, the amplification reaction is a PCR reaction, the annealing temperature is 1-15°C higher than the melting temperature of at least 50% of the primers in the primer set, the length of the annealing step in the PCR reaction is 15-120 minutes, the primer concentration in the amplification reaction is 1-10 nM, and the primers in the primer set are designed to minimize primer-dimer formation.

[0036] In some embodiments, the multiplex amplification reaction is performed under limiting primer conditions.

[0037] Another aspect of the invention described herein relates to a composition comprising a circulating tumor nucleic acid fragment comprising a universal adaptor, wherein the circulating tumor nucleic acid was derived from breast cancer, bladder cancer, or colon cancer.

[0038] In some embodiments, the circulating tumor nucleic acid was derived from a blood or urine sample, or a fragment thereof, of an individual with breast, bladder, or colon cancer.

[0039] Another aspect of the invention described herein relates to a composition comprising a solid support comprising a clonal population of a plurality of nucleic acids, wherein the clonal population comprises amplicons generated from a sample of circulating free nucleic acid, and the circulating tumor nucleic acid was from breast cancer, bladder cancer, or colon cancer.

[0040] In some embodiments, the circulating free nucleic acids were derived from blood or urine samples, or fragments thereof, of individuals with breast, bladder, or colon cancer.

[0041] In some embodiments, the nucleic acid fragments in the different clonal populations contain the same universal adaptors.

[0042] In some embodiments, a clonal collection of nucleic acids is derived from nucleic acid fragments from a set of samples from two or more individuals.

[0043] In some embodiments, the nucleic acid fragment comprises one of a set of molecular barcodes corresponding to a sample in a set of samples.

[0044] A further aspect of the invention described herein is a method for monitoring and detecting early recurrence or metastasis of breast, bladder or colon cancer, comprising: selecting a set of at least 8 or 16 patient-specific single nucleotide variant loci based on somatic mutations identified in tumor samples from patients diagnosed with breast, bladder or colon cancer; collecting one or more blood or urine samples longitudinally from the patients after the patients have been treated with surgery, first-line chemotherapy and / or adjuvant therapy; and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof to generate a set of amplicons, wherein each amplicon in the set of amplicons is selected from the set of amplicons. wherein each amplicon of the set of amplicons comprises a single nucleotide variant locus spanning at least one single nucleotide variant locus of a set of patient-specific single nucleotide variant loci associated with breast cancer, bladder cancer, or colon cancer; and sequencing at least one segment of each amplicon of the set of amplicons comprising the patient-specific single nucleotide variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colon cancer.

[0045] A further aspect of the invention described herein is a method for treating breast cancer, bladder cancer or colon cancer, comprising treating a patient diagnosed with breast cancer, bladder cancer or colon cancer with surgery, first-line chemotherapy and / or adjuvant therapy; collecting one or more blood or urine samples longitudinally from the patient; and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast cancer, bladder cancer or colon cancer and is selected based on somatic mutations identified in the patient's tumor sample, and wherein the amplicon is a single nucleotide variant of at least one of a set of at least 8 or 16 patient-specific single nucleotide variant loci. determining the sequence of at least one segment of each amplicon of a set of amplicons comprising patient-specific single nucleotide variant loci, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colon cancer; and administering to an individual a compound known to be effective in treating breast cancer, bladder cancer, or colon cancer having one or more of the single nucleotide variants detected from the blood or urine sample.

[0046] A further aspect of the invention described herein is a method of monitoring or predicting response to treatment for breast, bladder or colon cancer, comprising longitudinally collecting one or more blood or urine samples from patients undergoing treatment for breast, bladder or colon cancer, and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast, bladder or colon cancer and is selected based on somatic mutations identified in the patient's tumor sample, and wherein the amplicon comprises at least 8 or 16 patient-specific single nucleotide sequences. and determining the sequence of at least one segment of each amplicon of the set of amplicons that comprises the patient-specific single nucleotide variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific single nucleotide variants from the blood or urine sample is indicative of a poor response to treatment for breast cancer, bladder cancer, or colorectal cancer.

[0047] In some embodiments, the methods described herein involve detecting ctDNA in the plasma of a breast cancer patient before therapy and / or during neoadjuvant therapy (e.g., after cycle 1, cycle 2, cycle 3, cycle 4, etc.). In some embodiments, a treatment plan is defined based on ctDNA concentration determination (e.g., presence / absence) and the rate of decline during neoadjuvant therapy.

[0048] In some embodiments, the methods described herein involve assessing the presence and levels of ctDNA for each cancer patient (i.e., targeting mutations actually present in the tumor). In some embodiments, the methods described herein involve detecting 2 or more, 4 or more, 10 or more, 16 or more, 32 or more, 50 or more, 64 or more, or 100 or more mutations actually present in the patient's tumor(s).

[0049] According to some embodiments of the invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients who may have metastatic recurrence (e.g., after neoadjuvant therapy and surgery) have detectable ctDNA at baseline.

[0050] According to some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients who may have metastatic recurrence (e.g., after neoadjuvant therapy and surgery) have detectable ctDNA after cycle 1 of neoadjuvant therapy.

[0051] According to some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients who may have metastatic recurrence (e.g., after neoadjuvant therapy and surgery) have detectable ctDNA after cycle 2 of neoadjuvant therapy.

[0052] According to some embodiments of the invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients who may have metastatic recurrence (e.g., after neoadjuvant therapy and surgery) have detectable ctDNA after neoadjuvant therapy and before surgery.

[0053] According to some embodiments of the present invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients who may have metastatic recurrence (e.g., after neoadjuvant therapy and surgery) have detectable ctDNA after surgery.

[0054] According to some embodiments of the invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients with detectable ctDNA (e.g., after surgery) may have metastatic disease without further treatment relapse (e.g., after neoadjuvant therapy and surgery).

[0055] According to some embodiments of the invention, at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or about 100% of patients experiencing an increase in ctDNA between baseline and cycle 1 or cycle 2, etc., may have a metastatic recurrence after surgery if no further treatment is administered.

[0056] In some embodiments, the methods described herein include detecting the occurrence, recurrence, or metastasis of specific subtypes of cancer, including specific subtypes of breast cancer. In some embodiments, the methods described herein include detecting the occurrence, recurrence, or metastasis of HR+ / HER2- tumors, including HR+ / HER2- breast cancers (e.g., hormone receptor-positive ERα+ and / or PR+). HR+ tumors are typically less aggressive and have a better prognosis, with a 5-year survival rate of greater than 90%.

[0057] In some embodiments, the methods described herein include detecting the occurrence, recurrence, or metastasis of HER2+ tumors, including HER2+ breast cancer (human epidermal growth factor receptor 2 positive), which are generally more invasive, have a worse prognosis, and are more likely to recur and metastasize than HR+ / HER2- breast cancers.

[0058] In some embodiments, the methods described herein include detecting the occurrence, recurrence, or metastasis of HR- / HER2- breast cancer, including HR- / HER2- breast cancer (TNBC, i.e., triple-negative BC). Triple-negative breast cancer (TNBC) does not express ERα, PR, or HER2. These tumors tend to be the most aggressive and have the worst prognosis of all breast cancer subtypes.

[0059] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in at least 75%, at least 80%, at least 85%, at least 90%, or at least 95% of patients with early recurrence or metastasis of cancer.

[0060] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in at least 80%, at least 85%, at least 90%, at least 95%, or at least 98% of patients with early recurrence or metastasis of HER2+ breast cancer.

[0061] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in at least 80%, at least 85%, at least 90%, at least 95%, or at least 98% of patients with early recurrence or metastasis of triple-negative breast cancer.

[0062] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in at least 75%, at least 80%, at least 85%, at least 90%, or at least 95% of patients with early recurrence or metastasis of HR+ / HER2- breast cancer.

[0063] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in patients with early cancer recurrence or metastasis at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before clinical cancer recurrence or metastasis detectable by imaging, and / or at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before elevated CA15-3 levels.

[0064] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in patients with early recurrence or metastasis of HER2+ breast cancer at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before clinical recurrence or metastasis of HER2+ breast cancer detectable by imaging, and / or at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before elevated CA15-3 levels.

[0065] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in patients with early triple-negative breast cancer recurrence or metastasis at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before clinical recurrence or metastasis of triple-negative breast cancer detectable by imaging, and / or at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before elevated CA15-3 levels.

[0066] In some embodiments, the methods described herein are capable of detecting patient-specific single nucleotide variants in patients with early recurrence or metastasis of HR+ / HER2- breast cancer at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before clinical recurrence or metastasis of HR+ / HER2- breast cancer detectable by imaging, and / or at least 100 days, at least 150 days, at least 200 days, at least 250 days, or at least 300 days before elevated CA15-3 levels.

[0067] In some embodiments, the methods described herein do not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% of patients who do not have early recurrence or metastasis of cancer.

[0068] In some embodiments, the methods described herein do not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% of patients who do not have early recurrence or metastasis of HER2+ breast cancer.

[0069] In some embodiments, the methods described herein do not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% of patients who do not have early recurrence or metastasis of triple-negative breast cancer.

[0070] In some embodiments, the methods described herein do not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% of patients who do not have early recurrence or metastasis of HR+ / HER2- breast cancer.

[0071] In some embodiments, the methods described herein have a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% in detecting early recurrence or metastasis of cancer when two or more patient-specific single nucleotide variants are detected above a predetermined confidence threshold (e.g., 0.95, 0.96, 0.97, 0.98, or 0.99).

[0072] In some embodiments, the methods described herein have a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% in detecting early recurrence or metastasis of HER2+ breast cancer when two or more patient-specific single nucleotide variants are detected above a predetermined confidence threshold (e.g., 0.95, 0.96, 0.97, 0.98, or 0.99).

[0073] In some embodiments, the methods described herein have a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% in detecting early recurrence or metastasis of triple-negative breast cancer when two or more patient-specific single nucleotide variants are detected above a predetermined confidence threshold (e.g., 0.95, 0.96, 0.97, 0.98, or 0.99).

[0074] In some embodiments, the methods described herein have a specificity of at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% in detecting early recurrence or metastasis of HR+ / HER2- breast cancer when two or more patient-specific single nucleotide variants are detected above a predetermined confidence threshold (e.g., 0.95, 0.96, 0.97, 0.98, or 0.99).

[0075] In some embodiments, the methods described herein detect patient-specific single nucleotide variants in at least 75%, at least 80%, at least 85%, at least 90%, or at least 95% of patients with early recurrence or metastasis of muscle-invasive bladder cancer (MIBC).

[0076] In some embodiments, the methods described herein detect patient-specific single nucleotide variants in patients with early cancer recurrence or metastasis at least 100 days, at least 150 days, at least 200 days, or at least 250 days before clinical recurrence or metastasis of MIBC detectable by imaging.

[0077] In some embodiments, the methods described herein do not detect patient-specific single nucleotide variants in at least 95%, at least 98%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% of patients who do not have early recurrence or metastasis of MIBC.

[0078] In some embodiments, the methods described herein have a specificity of at least 95%, at least 98%, at least 99%, at least 99%, at least 99.5%, at least 99.8%, or at least 99.9% in detecting early recurrence or metastasis of MIBC when two or more patient-specific single nucleotide variants are detected above a predetermined confidence threshold (e.g., 0.95, 0.96, 0.97, 0.98, or 0.99).

[0079] In addition to, or instead of, single nucleotide variants, the methods described herein may also be based on the detection of other genomic variants (e.g., indels, multi-nucleotide variants and / or gene fusions).

[0080] Accordingly, a further aspect of the invention described herein is a method for monitoring and detecting early recurrence or metastasis of breast, bladder or colon cancer, comprising selecting a plurality of genomic variant loci (e.g., SNVs, indels, multi-nucleotide variants and gene fusions) based on somatic mutations identified in a tumor sample from a patient diagnosed with breast, bladder or colon cancer; collecting one or more blood or urine samples longitudinally from the patient after the patient has been treated with surgery, first-line chemotherapy and / or adjuvant therapy; and generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein the amplicons are generating a set of amplicons, each amplicon spanning at least one genomic variant locus of a set of patient-specific genomic variant loci associated with breast cancer, bladder cancer, or colorectal cancer; and sequencing at least one segment of each amplicon of the set of amplicons that comprises a patient-specific genomic variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer.

[0081] A further aspect of the invention described herein is a method for treating breast cancer, bladder cancer, or colon cancer, comprising treating a patient diagnosed with breast cancer, bladder cancer, or colon cancer with surgery, first-line chemotherapy, and / or adjuvant therapy; collecting one or more blood or urine samples from the patient longitudinally; and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample, or a fraction thereof, to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast cancer, bladder cancer, or colon cancer and is associated with at least one genomic variant locus (e.g., a genomic variant locus) of a set of at least 8 or 16 patient-specific genomic variant loci, selected based on somatic mutations identified in the patient's tumor sample. , SNVs, indels, multi-nucleotide variants, and gene fusions), determining the sequence of at least one segment of each amplicon of a set of amplicons comprising patient-specific genomic variant loci, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colon cancer, and administering to an individual having one or more of the genomic variants detected from the blood or urine sample a compound known to be effective in treating breast cancer, bladder cancer, or colon cancer.

[0082] A further aspect of the invention described herein is a method of monitoring or predicting response to treatment for breast, bladder or colon cancer, comprising longitudinally collecting one or more blood or urine samples from a patient undergoing treatment for breast, bladder or colon cancer, and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast, bladder or colon cancer and is selected based on somatic mutations identified in the patient's tumor sample, and wherein the multiplex amplification reaction generates a set of amplicons for at least 8 or 16 patient-specific genomic variant loci associated with breast, bladder or colon cancer, the at least 8 or 16 patient-specific genomic variant loci selected based on somatic mutations identified in the patient's tumor sample. and determining the sequence of at least one segment of each amplicon of the set of amplicons that comprises the patient-specific genomic variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from the blood or urine sample is indicative of a poor response to treatment for breast cancer, bladder cancer, or colorectal cancer.

[0083] In addition to, or instead of, patient-specific genomic variants, the methods described herein may also be based on the detection of mutations associated with recurrent cancer (e.g., hotspot cancer mutations, drug resistance markers, cancer panel mutations), which recur in many cancer patients.

[0084] Accordingly, a further aspect of the invention described herein is a method for monitoring and detecting early recurrence or metastasis of breast, bladder or colon cancer, comprising selecting mutations associated with multiple recurrent cancers; collecting one or more blood or urine samples longitudinally from patients after the patients have been treated with surgery, first-line chemotherapy and / or adjuvant therapy; and performing a multiplex amplification reaction on nucleic acid isolated from each blood or urine sample or fraction thereof to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast cancer, bladder cancer or colon cancer. and determining the sequence of at least one segment of each amplicon of the set of amplicons comprising a mutation associated with recurrent cancer, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) mutations associated with recurrent cancer from a blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colon cancer.

[0085] A further aspect of the invention described herein is a method for treating breast cancer, bladder cancer, or colon cancer, comprising treating a patient diagnosed with breast cancer, bladder cancer, or colon cancer with surgery, first-line chemotherapy, and / or adjuvant therapy; collecting one or more blood or urine samples longitudinally from the patient; and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample, or a fraction thereof, to generate a set of amplicons, wherein each amplicon in the set of amplicons is a mutation associated with at least one recurrent cancer (e.g., a hotspot cancer mutation, a drug resistance marker, a marker mutation ... , cancer panel mutations), determining the sequence of at least one segment of each amplicon of the set of amplicons comprising mutations associated with recurrent cancer, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) mutations associated with recurrent cancer from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colon cancer, and administering to an individual having one or more of the mutations associated with recurrent cancer detected from the blood or urine sample a compound known to be effective in treating breast cancer, bladder cancer, or colon cancer.

[0086] A further aspect of the invention described herein is a method of monitoring or predicting response to treatment of breast cancer, bladder cancer or colon cancer, comprising longitudinally collecting one or more blood or urine samples from a patient undergoing treatment for breast cancer, bladder cancer or colon cancer, and performing a multiplex amplification reaction on nucleic acid isolated from each blood or urine sample or fraction thereof to generate a set of amplicons, wherein each amplicon in the set of amplicons contains at least one recurrent mutation of a set of at least 8 or 16 recurrent mutations associated with breast cancer, bladder cancer or colon cancer. and determining the sequence of at least one segment of each amplicon of the set of amplicons that includes mutations associated with recurrent cancer, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) mutations associated with recurrent cancer from the blood or urine sample is indicative of a poor response to treatment for breast cancer, bladder cancer, or colon cancer.

[0087] In addition to, or instead of, initially identifying somatic mutations from tumor samples of patients diagnosed with breast, bladder, or colon cancer, the methods described herein may also be based on identifying somatic mutations from other biological samples of the patient, such as blood, serum, plasma, urine, hair, tears, saliva, skin, fingernails, feces, bile, lymph, cervical mucus, or semen.

[0088] Accordingly, a further aspect of the invention described herein is a method for monitoring and detecting early recurrence or metastasis of breast, bladder, or colon cancer, comprising selecting multiple genomic variant loci (e.g., SNVs, indels, multi-nucleotide variants, and gene fusions) based on somatic mutations identified in a biological sample (e.g., blood, serum, plasma, urine, hair, tears, saliva, skin, fingernails, feces, bile, lymph, cervical mucus, or semen) containing cancer-associated mutations from a patient diagnosed with breast, bladder, or colon cancer; collecting one or more blood or urine samples longitudinally from the patient after the patient has been treated with surgery, first-line chemotherapy, and / or adjuvant therapy; and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample, or fractions thereof. generating a set of amplicons by performing a sequence analysis of at least one segment of each amplicon of the set of amplicons, wherein each amplicon of the set of amplicons spans at least one genomic variant locus among patient-specific genomic variant loci associated with breast cancer, bladder cancer, or colorectal cancer; and determining the sequence of at least one segment of each amplicon of the set of amplicons that comprises a patient-specific genomic variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colorectal cancer.

[0089] A further aspect of the invention described herein is a method for treating breast cancer, bladder cancer, or colon cancer, comprising treating a patient diagnosed with breast cancer, bladder cancer, or colon cancer with surgery, first-line chemotherapy, and / or adjuvant therapy; longitudinally collecting one or more blood or urine samples from the patient; and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample, or a fraction thereof, to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast cancer, bladder cancer, or colon cancer and comprises at least 8 or 16 patient-specific genomic variant loci selected based on somatic mutations identified in the patient's biological sample (e.g., blood, serum, plasma, urine, hair, tears, saliva, skin, fingernail, feces, bile, lymph, cervical mucus, or semen) that contain a cancer-associated mutation. wherein the genomic variants span at least one genomic variant locus (e.g., SNVs, indels, multi-nucleotide variants, and gene fusions) from a set of amplicons; determining the sequence of at least one segment of each amplicon of the set of amplicons that comprises the patient-specific genomic variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from the blood or urine sample is indicative of early recurrence or metastasis of breast cancer, bladder cancer, or colon cancer; and administering to the individual a compound known to be effective in treating breast cancer, bladder cancer, or colon cancer that has one or more of the genomic variants detected from the blood or urine sample.

[0090] A further aspect of the invention described herein is a method of monitoring or predicting response to treatment for breast cancer, bladder cancer, or colon cancer, comprising longitudinally collecting one or more blood or urine samples from a patient undergoing treatment for breast cancer, bladder cancer, or colon cancer, and performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample, or a fraction thereof, to generate a set of amplicons, wherein each amplicon in the set of amplicons is associated with breast cancer, bladder cancer, or colon cancer and selected based on a somatic mutation identified in the patient's biological sample (e.g., blood, serum, plasma, urine, hair, tears, saliva, skin, fingernail, feces, bile, lymph, cervical mucus, or semen) that contains a cancer-associated mutation. generating amplicons spanning at least one genomic variant locus (e.g., SNVs, indels, multi-nucleotide variants, and gene fusions) of a set of at least 8 or 16 patient-specific genomic variant loci; and determining the sequence of at least one segment of each amplicon of the set of amplicons comprising the patient-specific genomic variant locus, wherein detection of one or more (or two or more, or three or more, or four or more, or five or more, or six or more, or seven or more, or eight or more, or nine or more, or ten or more) patient-specific genomic variants from the blood or urine sample is indicative of a poor response to treatment for breast cancer, bladder cancer, or colorectal cancer.

[0091] Other embodiments, features, and advantages of the disclosed invention will be apparent from the following detailed description and claims. [Brief explanation of the drawings]

[0092] The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

[0093] The presently disclosed embodiments will be further described with reference to the accompanying drawings, wherein like structure is referred to by like numerals throughout the several views. The drawings shown are not necessarily to scale, emphasis instead generally being placed upon illustrating the principles of the presently disclosed embodiments.

[0094] [Figure 1] FIG. 1 is a workflow diagram. [Figure 2] Upper panel: number of SNVs per sample, lower panel: working assays sorted by driver category. [Figure 3] Measured cfDNA concentrations. Each data point represents a plasma sample. [Figure 4] Samples showing good correlation between previously determined tissue VAF measurements (x-axis) and the present ones using mPCR-NGS (y-axis). Each sample is represented by a separate box, and VAF data points are color-coded by tissue subsection. [Figure 5] Samples showing poor correlation between previously determined tissue VAF measurements (x-axis) and the present ones using mPCR-NGS (y-axis). Each sample is represented by a separate box, and VAF data points are color-coded by tissue subsection. [Figure 6A-B] Depth of read histograms as a function of calls made. Top: The assay did not detect the expected plasma SNVs. Bottom: The assay detected the expected plasma SNVs. [Figure 7] Number of SNVs detected in plasma by histological type. [Figure 8] SNV detection in plasma (left) and samples (right) by tumor stage. [Figure 9] Plasma VAF as a function of tumor stage and SNV clonality. [Figure 10] Number of SNVs detected in plasma from each sample as a function of cfDNA input amount. [Figure 11]Plasma VAF as a function of mean tumor VAF. The mean tumor VAF was calculated across all tumor subsections analyzed from each tumor. [Figure 12] The clonal fraction (red vs. blue) and mutant variant allele frequency (MutVAF) of each detected SNV are shown. All SNVs detected from each sample are placed in a single column, and the sample is categorized by tumor stage (pTNM stage). Samples in which no SNVs are detected are included. The clonal fraction is defined as the ratio between the number of tumor subsections in which the SNV was observed and the total number of subsections analyzed from that tumor. [Figure 13] The clonal status (blue for clones, red for subclones) and mutant variant allele frequency (MutVAF) of each detected SNV are shown. All SNVs detected in each sample are placed in a single column, and the sample is categorized by tumor stage (pTNM stage). Samples without detected SNVs are included. Clonal status was determined by PyCloneCluster using whole-exome sequencing data from tumor tissue. [Figure 14] The clonal status (blue for clones, red for subclones) and mutant variant allele frequency (MutVAF) of each detected SNV are shown. The upper panel shows only clonal SNVs, and the lower panel shows only subclonal SNVs. All SNVs detected from each sample are placed in a single column, and the sample is categorized by tumor stage (pTNM stage). Samples without detected SNVs are included. Clonal status was determined by PyCloneCluster using whole-exome sequencing data from tumor tissue. [Figure 15] The number of SNVs detected in plasma is shown as a function of histological type and tumor size. Histological type and tumor stage were determined by pathology reports. Each data point is color-coded by size, with red indicating the largest tumor size and blue indicating the smallest tumor size. [Figure 16]1 is a table of cfDNA analysis showing DNA concentration, genome copy equivalents in library preparations, plasma hemolysis grade and cDNA profile in all samples. [Figure 17] 1 is a table of SNVs detected in plasma for each sample. [Figure 18] 1 is a table of additional SNVs detected in plasma. [Figure 19] An example of a detection assay and its background allele fraction for plasma samples at relapse (LTX103). [Figure 20A-B] Schematic diagram of clinical and molecular protocols. [Figure 21] Exam overview. [Figure 22] Patient summary of 36 months of surveillance and plasma collection. [Figure 23A-B] Risk of recurrence stratified by postoperative ctDNA status. [Figure 24A-B] Risk of post-therapy recurrence stratified by postoperative ctDNA status. [Figure 25] Efficacy of adjuvant therapy in preventing recurrence. [Figure 26A-B] Time to release based on radiology and ctDNA. [Figure 27A-D] Early detection of recurrence and prediction of treatment response. [Figure 28] Schematic diagram of clinical sample collection. [Figure 29] Plasma sequencing QC. [Figure 30A-F] Early recurrence detection. [Figure 31] Recurrence-free survival and ctDNA status at diagnosis and after cystectomy. [Figure 32A-B] Neoadjuvant treatment response. [Figure 33] Signatera (RUO) process. [Figure 34] Plasma sequencing QC. [Figure 35] Sensitivity of single SNV detection. [Figure 36]Predicted inputs vs. observed VAF using Signatera (RUO). [Figure 37] Patient summary for the breast cancer study in Example 6. [Figure 38A-H]

[0033] Figure 38A is a table of information regarding the samples analyzed in the study of Example 6. Figure 38A is part 1 of the table. Figure 38B is a continuation of the table. Figure 38C is a continuation of the table. Figure 38D is a continuation of the table. Figure 38E is a continuation of the table. Figure 38F is a continuation of the table. Figure 38G is a continuation of the table. Figure 38H is a continuation of the table. [Figure 39] Patient demographics for the breast cancer study of Example 6. Raw WES data (including driver variants for 35 patients) for 50 patients were received. 218 plasma samples at various time points (between 1 and 8) were received. 108 additional extracted DNA samples were received. Recurrence status was also collected. Blood samples were taken after adjuvant therapy with a time interval of 6 months. [Figure 40] Summary of WES analysis and pool design for breast cancer study in Example 6. Pool A is based on the Signatera method. Pool B contains 25 patients, indicated by an asterisk in the box plot. 19 patients in Pool B had low tumor purity. 6 patients had very early-stage HER2- tumors. Pool B contains driver variants. [Figure 41] Plasma samples for breast cancer studies in Example 6. The median plasma volume was 4 mL. The median DNA input was 26 ng. The median DNA input was lower than the CRC and MIBC samples (45 ng and 66 ng, respectively). [Figure 42] Sequencing quality control showing median process error rate per type and median assay depth of reads for the breast cancer study in Example 6. A total of 326 plasma sequencing samples were processed. The variant call FP rate is estimated to be 0.28%. [Figure 43]Rerun of plasma samples for breast cancer study in Example 6. 319 sequenced samples, including 214 unique plasma samples, represented 49 patients. [Figure 44] Results from Pool A for the breast cancer study in Example 6. Of 49 patients, 11 were baseline positive. 3 were at only one timepoint. The remaining 8 patients remained positive throughout. Pool B and the driver yielded similar results. Driver information: 16 recurrence samples have a driver mutation. 11 recurrence samples have had at least one assay with the driver. [Figure 45] Summary of 16 patients in whom ctDNA was detected. [Figure 46] Graphical representation of the data corresponding to patient CD047 (TNBC) in Figure 38. [Figure 47] Graphical representation of the data corresponding to patient CD033 (TNBC) in Figure 38. [Figure 48] Graphical representation of the data corresponding to patient CD037 (HER2+) in Figure 38. [Figure 49] Graphical representation of the data corresponding to patient CD040 (HER2+) in Figure 38. [Figure 50] Graphical representation of the data corresponding to patient CD048 (HER2-) in Figure 38. [Figure 51] Graphical representation of the data corresponding to patient CD005 (HER2-) in Figure 38. [Figure 52] Graphical representation of the data corresponding to patient CD036 (HER2-) in Figure 38. [Figure 53] Graphical representation of the data corresponding to patient CD044 (HER2-) in Figure 38. [Figure 54] Graphical representation of the data corresponding to patient CD049 in FIG. 38. [Figure 55] Graphical representation of the data corresponding to patient CD029 in FIG. 38. [Figure 56]Graphical representation of the data corresponding to patient CD026 in FIG. [Figure 57] Pictorial representation of data corresponding to patient CD017 in FIG. 38. [Figure 58] Graphical representation of data corresponding to patient CD031 in Figure 38. HW: SHC2, PKD1, COLEC12. [Figure 59] Graphical representation of the data corresponding to patient CD025 in Figure 38. In this patient, ctDNA was observed at two consecutive time points for FGF9 mutations. This patient is likely to experience relapse in the near future. [Figure 60] Patient Recruitment and Clinical Sample Collection. For the 49 women with BC monitored in this study, collected tumor tissue and serial plasma samples were analyzed in a blinded manner using the Signatera™ RUO workflow. Exome alterations were determined by paired-end sequencing of FFPE tumor tissue specimens and matched normal DNA. Patient-specific panels containing 16 somatic mutations identified from WES were designed. Plasma samples were processed using these corresponding custom panels. 208 samples were analyzed for ctDNA detection. [Figure 61A-C] Overview of ctDNA analysis and summary of results. (A) Summary of treatment regimens for each patient (n=49) along with results from consecutive plasma samples analyzed (n=208). (B) Summary table showing total patients, number of recurrences, percentage detected by ctDNA analysis, and median lead time in days for each breast cancer subtype. (C) Comparison of molecular and clinical recurrences colored by breast cancer subtypes HR+, HER2+, and TNBC using a paired Wilcoxon signed-rank test (p-value <0.001). [Figure 62A-B]CtDNA detection in serial plasma samples predicts recurrence-free survival. (A) Detection of ctDNA in any follow-up plasma sample after surgery predicts recurrence-free survival [HR: 35.84 (7.9626-161.32)] p-value <0.001. (B) Detection of ctDNA in the first post-surgery plasma sample predicts recurrence-free survival [HR: 11.784 (4.2784-32.457)]. Data are from n=49 patients, p-value <0.001. [Figure 63] (A-E) Plasma levels of ctDNA across multiple plasma timepoints for five breast cancer patients (one per panel). Whole-exome sequencing of primary tumors and matched normals identified patient-specific somatic mutations. Using the analytically validated Signatera™ RUO workflow, each patient-specific assay was designed to target 16 somatic SNV and indel variants using massively parallel sequencing (median depth per target >100,000x). The mean VAF is indicated by the dark blue circle, and the solid line represents the mean VAF profile over time. Lead time is calculated by the difference between clinical and molecular recurrence. CA15-3 levels are graphed over time, with baseline levels indicated by light blue shading. (F) Summary of VAF and target counts detected at molecular and clinical recurrence for all ctDNA-positive samples, except for patients with only one timepoint. [Figure 64A-C] Signatera variant selection strategy for 49 patient-specific panels. (Top) Tumor tissue VAF distribution in the patient's custom panel. Different colors represent different subtypes: HER2- (dark blue), triple negative (orange), and HER2+ (green). (Center) Number of putative clonal and subclonal variants in the patient's custom panel. The median number of clonal variants in the 49 custom panels is 13 out of 16. (Bottom) Number of putative clonal and subclonal variants in the patient's WES data. [Figure 65](A-L) Plasma levels of ctDNA across multiple plasma timepoints for 12 breast cancer patients (11 relapsed and 1 non-relapsed). Whole-exome sequencing of primary tumors and matched normals identified patient-specific somatic mutations. Using the analytically validated Signatera™ workflow, each patient-specific assay was designed to target 16 somatic SNV and indel variants using massively parallel sequencing (median depth per target >100,000x). The mean VAF is indicated by the dark blue circle, and the solid line represents the mean VAF profile over time. Lead time is calculated by the difference between clinical and molecular recurrence. CA15-3 levels are graphed over time, with baseline levels indicated by light blue shading. [Figure 66] Distribution of VAF and mutant counts. In total, 251 targets were detected in the ctDNA-positive plasma samples. The VAF of detected targets ranged from 0.01% to 64%, with a median of 0.82%. The observed mutant VAF and total DNA molecule count for each sample were used to calculate the number of tumor molecules present in the patient's plasma sample. The number of mutant molecules detected in the 251 positive targets ranged from 1 to 6,500 mutant molecules, with a median of 39 molecules. [Fig. 67A-D]Signatera Quality Control Process: Quality control was performed at every step of the workflow. Of a total of 215 plasma samples, 208 passed our sample QC process, and of the 784 unique assays designed, 767 passed our QC (corresponding to a total of 3,237 assays passed out of 3,328 across all samples). A) cfDNA extracted per mL. cfDNA extracted from each plasma sample was quantified using the Quant-iT High Sensitivity dsDNA Assay Kit. Samples with quantified cfDNA amounts less than 5 ng were flagged as "WARNING." cfDNA extracted per mL ranged from 1 to 21.4 ng, with a median of 4.7 ng. B) DNA input amount for library preparation. Up to 66 ng of cfDNA from each plasma sample was used as input to the library preparation protocol. Library DNA input amounts ranged from 1 to 66 ng, with a median of 25.02 ng. The purified libraries underwent QC before proceeding to the next step. C) Sequencing coverage. Assays with coverage less than 5000x were excluded from analysis. Samples with fewer than eight passing assays subsequently failed sequencing coverage QC. The median read depth for assays that passed coverage QC was 110,000x. D) Sample concordance. To track sample integrity, SNP tracers were used to measure concordance between patient samples. A genotyping concordance score was calculated for each plasma sample compared with its corresponding matched normal genotyping data. Samples were considered to be from the same patient if at least 85% of their SNPs had identical genotypes. Six plasma samples identified as permuted were excluded from ctDNA analysis. [Figure 68A-B]Assay validation results. (A) Single-target detection sensitivity. Approximately 60% analytical sensitivity was achieved with Signatera for mutation detection in approximately 0.03% spike-in tumor DNA. (B) Estimated sample-level sensitivity of Signatera when at least two mutations are detected from a set of 16 target variants. [Figure 69] After screening and recruitment, patients were followed up with six monthly blood samples. HER2 status was determined by immunohistochemistry and fluorescence in situ hybridization assays. If either assay was positive, the patient was considered to have HER2-positive cancer. NACT: neoadjuvant chemotherapy; ACT: adjuvant chemotherapy. [Figure 70] FIG. 10 is a workflow diagram of the muscle-invasive bladder cancer study in Example 9. [Fig. 71A-G] Summary of patients for muscle-invasive bladder cancer in Example 9. Figure 71A shows the rate of synonymous and nonsynonymous mutations called from WES. One patient's tumor was highly mutated, with a mutational burden of 126 mutations / Mb, and showed a POLD1 mutation, which has previously been associated with hypermutators (Campbell, BB et al., Comprehensive Analysis of Hypermutation in Human Cancer. Cell 171, 1042-1056. e10 (2017)). Figure 71B shows the relative contribution of mutational signatures associated with bladder cancer. Figure 71C shows mutations in genes frequently mutated in bladder cancer (TCGA) (Robertson, AG et al., Comprehensive Molecular Characterization of Muscle-Invasive Bladder Cancer. Cell 171,540-556.e25(2017)). Figure 71D shows deleterious mutations in DNA damage response (DDR)-related genes that were mutated in more than 5% of 68 samples. Figure 71E shows the total number of deleterious DDR mutations. Figure 71F shows clinical and histopathological characteristics. Figure 71G shows the summarized ctDNA status. [Figure 72]FIG. 10 outlines the clinical protocol and sampling plan for the muscle-invasive bladder cancer study in Example 9. [Figure 73] Diagram outlining the Signatera™ workflow. [Figure 74] Longitudinal display of ctDNA results for all analyzed samples corresponding to the muscle-invasive bladder cancer study in Example 9. Patients are divided into three groups based on ctDNA status. The upper panel shows patients who were ctDNA positive before and after cystectomy (CX), the middle panel shows patients who were ctDNA positive only before CX, and the lower panel shows patients who were ctDNA negative. Horizontal lines represent each patient's disease course, circles represent ctDNA status, and red circles indicate samples with at least two positive assays. Treatment and imaging information is shown for each patient. [Figure 75A-E] Graphical representation of the prognostic value of ctDNA detection for muscle-invasive bladder cancer studies in Example 9. Kaplan-Meier survival analysis showing the probability of recurrence-free survival (RFS) and overall survival (OS) stratified by ctDNA status before chemotherapy (Figure 75A), before cystectomy (CX) (Figure 75B), and after cystectomy (CX) (Figure 75C). Figure 75D shows the relationship between pre-chemotherapy, pre-cystectomy, and post-cystectomy disease recurrence and ctDNA status and pre-cystectomy disease recurrence and lymph node status. Figure 75E shows the relationship between pre-cystectomy (CX) ctDNA status and pathology status at cystectomy (CX). Statistical significance was assessed using the Wilcoxon rank-sum test for continuous variables and Fisher's exact test for categorical variables. [Figure 76] Graph showing ctDNA changes in individual disease courses for muscle-invasive bladder cancer study in Example 9. Figure 76 shows a display of detailed disease courses, applied treatments, and associated longitudinal ctDNA analysis from selected patients. ctDNA status, applied treatments, and imaging results are presented according to the legend. A favorable lead time for ctDNA-based recurrence detection is demonstrated. [Figure 77]Graph showing the time difference between molecular recurrence (ctDNA positive) and clinical recurrence (radiographic imaging positive) for the muscle-invasive bladder cancer study in Example 9. P values ​​were calculated using the paired Wilcoxon signed-rank test. [Fig. 78A-H] Graphs showing predictive markers of chemotherapy response for the muscle-invasive bladder cancer study in Example 9. Figure 78A shows the relationship between disease recurrence and response to chemotherapy. Figure 78B shows the relative contribution of Signature 5 for all patients stratified by response to chemotherapy and ERCC2 mutation status, respectively. Figure 78C shows the fraction of patients responding to treatment in relation to ERCC2 mutation status. Figure 78D. RNA subtype diagram - new diagram. Figure 78E shows the association between ctDNA and response to chemotherapy for patients who are ctDNA negative throughout the entire disease course, patients whose ctDNA levels drop to zero, and patients whose ctDNA levels remain positive. Figure 78F shows ctDNA levels for all patients with detectable ctDNA before, during, and after chemotherapy. Patients are grouped by response to chemotherapy and recurrence status is shown. [Fig. 79A-D] Graph showing the total number of identified mutations per patient in relation to ERCC2 status or number of disruptive DNA damage response (DDR) mutations for the muscle-invasive bladder cancer study in Example 9. [Figure 80] Graph showing genomic heterogeneity between primary tumors and metastatic recurrences for muscle-invasive bladder cancer study in Example 9. Whole exome sequencing (WES) data of primary tumors was compared with ctDNA. WES data from plasma samples with high ctDNA variant allele frequency (VAF) were detected at the time of metastatic recurrence. Genomic locations with mutations identified in either plasma or tumor exome data were examined for base counts. The identified and resulting allele frequencies in plasma or tumor exome data are shown. Individual mutations are color-coded according to the statistical probability (strength) of the mutation call. Venn diagrams represent the number of mutations identified exclusively in tumors, plasma, or both. [Figure 81] Graph showing variant allele frequency (VAF%) at various days relative to cystectomy (CX) from 8 patients from the muscle-invasive bladder cancer study in Example 9. [Figure 82] Graph showing ctDNA levels in plasma from 10 patients previously analyzed by ddPCR compared to ultra-deep sequencing for muscle-invasive bladder cancer testing in Example 9. [Fig. 83A-E] Graphs showing clinical, histopathological, and molecular parameters for all 125 patients. Figure 83A shows the relative contribution of the five most relevant colorectal cancer-associated mutational signatures. Figure 83B shows the rates of synonymous and nonsynonymous mutations called from WES. Figure 83C shows a graph showing mutations in genes frequently mutated in colorectal cancer (TCGA) {Cancer Genome Atlas, 2012 #52}. Figure 83D shows clinical and histopathological characteristics. Figure 83E shows a graph summarizing pre- and post-operative ctDNA status. [Figure 84] Diagram showing patient enrollment, sample collection, and definition of patient subgroups used to address the defined clinical question. Abbreviations: ctDNA, circulating tumor DNA; CT scan, computed tomography scan; post-op, post-surgery; TTR, time to recurrence. [Fig. 85A-C]A graph showing quality control (QC) testing of the workflow for whole-exome sequencing of patient samples. 793 of 795 plasma samples (99%) passed the sample QC process. 194 samples (from 70 patients) were run with SNP tracers to test for concordance between the plasma sample and its corresponding tissue biopsy. All 194 plasma samples passed concordance QC. Figure 85A shows the DNA input amount for library preparation. Up to 66 ng of cell-free DNA (cfDNA) from each plasma sample was used as input to the library preparation protocol. Library DNA input amounts ranged from 1 to 66 ng, with a median of 45.66. Purified libraries were subjected to quality control before proceeding to the next step. One sample failed library preparation QC. Figure 85B shows sequencing coverage. Assays with coverage less than 5000x were excluded from analysis. Subsequently, samples with fewer than eight passing assays failed sequencing coverage QC. One sample failed the sequencing coverage requirement. The median read depth for assays that passed coverage QC was 105,000-fold. Figure 85C shows the sequencing error rates measured in all plasma samples. The average transition error rate was 5e-5, and the average transversion error rate was 8e-6. [Figure 86] The results and kinetics of circulating tumor DNA (ctDNA) for each individual patient are shown. [Fig. 87A-F]Figure 87A shows the ctDNA status before surgery (pre-op), 30 days after surgery, and during adjuvant chemotherapy (ACT). Figure 87A shows pre-op detection of ctDNA. Figure 87B shows the recurrence rate. Figure 87C shows Kaplan-Meier estimates of TTR for 94 stage I-III patients stratified by ctDNA status 30 days after surgery. Figure 87D shows the effect of ACT on ctDNA-positive patients, as assessed by recurrence rate and long-term ctDNA status. Figure 87E shows the recurrence rate stratified by ctDNA status at the first visit after ACT. Figure 87F shows Kaplan-Meier estimates of TTR for 58 ACT-treated patients stratified by ctDNA status and first visit after ACT. [Figure 88] Postoperative detection of carcinoembryonic antigen (CEA) in 125 patients with stage I-III CRC is shown. [Figure 89] Schematic representation of ctDNA profiling results for plasma samples included in the ctDNA analysis at day 30, ordered by relapse status and disease stage. Patients marked with a(s) have synchronous CRC. Plasma marked with ** is positive in the second pool only (n=1). [Figure 90A-B] Schematic representation of ctDNA profiling results for the subset of plasma samples included in the ctDNA analysis and undergoing ACT at day 30, ordered by relapse status and disease stage. Patients marked with an a(n) have synchronous CRC. [Figure 91] Schematic representation of ctDNA profiling results for plasma samples included in the longitudinal post-ACT ctDNA analysis, ordered by recurrence status, post-surgery ctDNA status, and length of follow-up. Patients marked with an a(n) have synchronous CRC (n=2). Plasma samples marked with a ** are positive only in the second pool (n=1). [Figure 92]Schematic representation of CEA profiling results for plasma samples included in the longitudinal post-ACT ctDNA analysis ordered by recurrence status, post-surgery ctDNA status, and length of follow-up. Patients marked with an a(n) have synchronous CRC (n=2). Plasma marked with a ** is positive only in the second pool (n=1). [Fig. 93A-D] (Figures 93-D) Graphs showing the association between ctDNA status and recurrence after definitive treatment. Figure 93A shows recurrence rates stratified by longitudinal ctDNA status. Figure 93B shows Kaplan-Meier estimates of TTR for 75 patients with longitudinal samples stratified by longitudinal ctDNA status. Figure 93C shows a graph comparing time to radiological recurrence with time to ctDNA recurrence. Figure 93D shows that the variant allele frequency (VAF) of plasma ctDNA increased toward radiological recurrence. Early timepoints before and during ACT are omitted. [Figure 94] Schematic representation of ctDNA profiling results of longitudinal plasma samples from relapsed and non-relapsed patients. Patients with only one positive plasma sample during surveillance are considered positive. [Figure 95] Schematic representation of CEA profiling results of longitudinal serum samples from relapsed and non-relapsed patients. Patients with only one positive plasma sample during surveillance are considered positive. [Figure 96] Graph comparing time to radiological recurrence and time to CEA recurrence. [Fig. 97A-C]Detection of disease-causing mutations in relapsed patients. Figure 97A shows the proportion of ctDNA+ relapsed patients with a disease-causing mutation detected during surveillance. First ctDNA+ sample (left column) and all ctDNA+ plasma samples (right column). Figure 97B shows disease-causing variants called in blood. Correlation between mean blood VAF calculated using the Signatera ctDNA+ assay and the variant allele frequency (VAF) of disease-causing mutations, plotted using logarithmic scales on both the horizontal and vertical axes. Figure 97C shows serial ctDNA profiling of two representative relapsed patients with disease-causing mutations. [Figure 98] Schematic comparison of current standard of care and postoperative management of patients guided by potential ctDNA. [Figure 99] Graph showing ctDNA reduction by adjuvant chemotherapy (ACT). [Figure 100A-B] Graphs showing the association between ctDNA status and recurrence after definitive treatment. Figure 100A shows the recurrence rate stratified by longitudinal ctDNA status and Kaplan-Meier estimates of TTR for 58 patients with longitudinal samples stratified by longitudinal ctDNA status. Figure 100B shows the recurrence rate stratified by CEA analysis and Kaplan-Meier estimates of TTR for 58 patients with longitudinal samples stratified by CEA analysis.

[0095] The above-identified figures are provided for purposes of illustration and not limitation. DETAILED DESCRIPTION OF THE INVENTION

[0096] The methods and compositions provided herein improve the detection, diagnosis, staging, screening, treatment, and management of cancer (e.g., breast, bladder, or colon cancer). In exemplary embodiments, the methods provided herein analyze single nucleotide variant mutations (SNVs) in circulating fluids, particularly circulating tumor DNA. The methods offer the advantage of identifying many of the mutations found in tumors and clones, as well as subclonal mutations, in a single test, compared to multiple tests that require the use of tumor samples for any effectiveness. The methods and compositions may be useful on their own, or they may be useful when used in conjunction with other methods for the detection, diagnosis, staging, screening, treatment, and management of cancer (e.g., breast, bladder, or colon cancer), e.g., to help corroborate the results of these other methods and provide more reliable and / or definitive results.

[0097] Thus, in one embodiment, provided herein is a method for determining single nucleotide variants present in a cancer (e.g., breast cancer, bladder cancer, or colon cancer) by determining single nucleotide variants present in a ctDNA sample from an individual, e.g., an individual having or suspected of having cancer (e.g., breast cancer, bladder cancer, or colon cancer), using the ctDNA SNV amplification / sequencing workflow provided herein.

[0098] The terms "cancer" and "cancerous" refer to or describe a physiological condition in animals that is typically characterized by uncontrolled cell growth. A "tumor" contains one or more cancerous cells. There are several major types of cancer. Carcinoma is cancer that begins in the skin or in tissues that outline or cover internal organs. Sarcoma is cancer that begins in bone, cartilage, fat, muscle, blood vessels, or other connective or supportive tissue. Leukemia is cancer that begins in blood-forming tissues, such as the bone marrow, and results in the production of large numbers of abnormal blood cells that enter the blood. Lymphoma and multiple myeloma are cancers that begin in cells of the immune system. Central nervous system cancer is cancer that begins in the tissues of the brain and spinal cord.

[0099] In some embodiments, the cancer is acute lymphoblastic leukemia, acute myeloid leukemia, adrenocortical carcinoma, AIDS-related cancer, AIDS-related lymphoma, anal cancer, appendiceal cancer, astrocytoma, atypical teratoma / rhabdoid tumor, basal cell carcinoma, bladder cancer, brain stem glioma, brain tumor (including brain stem glioma, central nervous system atypical teratoma / rhabdoid tumor, central nervous system embryonal tumor, astrocytoma, craniopharyngioma, ependymoblastoma, ependymoma, medulloblastoma, medulloepithelioma, moderately differentiated pineal parenchymal tumor, supratentorial primitive neuroectodermal tumor, and pineoblastoma), breast cancer, bronchial cancer, Tumor, Burkitt lymphoma, carcinoma of unknown primary site, carcinoid tumor, carcinoma of unknown primary site, central nervous system atypical teratoma / rhabdoid tumor, central nervous system embryonal tumor, cervical cancer, childhood cancer, chordoma, chronic lymphocytic leukemia, chronic myeloid leukemia, chronic myeloproliferative disorder, colon cancer, colorectal cancer, craniopharyngioma, cutaneous T-cell lymphoma, pancreatic endocrine islet cell tumor, endometrial cancer, ependymoblastoma, ependymoma, esophageal cancer, nasal neuroblastoma, Ewing's sarcoma, extracranial germ cell tumor, extragonadal germ cell tumor, extrahepatic bile duct cancer, gallbladder cancer, gastric cancer cancer), gastrointestinal carcinoid tumor, gastrointestinal stromal cell tumor, gastrointestinal stromal tumor (GIST), gestational trophoblastic tumor, glioma, hairy cell leukemia, head and neck cancer, cardiac cancer, Hodgkin's lymphoma, hypopharyngeal cancer, intraocular melanoma, islet cell tumor, Kaposi's sarcoma, kidney cancer, Langerhans cell histiocytosis, laryngeal cancer, lip cancer, liver cancer, malignant fibrous histiocytoma / bone cancer, medulloblastoma, medulloepithelioma, melanoma, Merkel cell carcinoma, Merkel cell skin carcinoma, mesothelioma, metastatic squamous cell carcinoma of the neck of unknown primary origin, mouth cancer, multiple endocrine neoplasia syndrome, multiple myeloma, multiple myeloma / plasmacytoma, mycosis fungoides, myelodysplastic syndrome, myeloproliferative neoplasm, nasal cancer, nasopharyngeal carcinoma, neuroblastoma, non-Hodgkin's lymphoma, non-melanoma skin cancer, non-small cell lung cancer, oral cancercancer), oral cancer, oropharyngeal cancer, osteosarcoma, other brain and spinal cord tumors, ovarian cancer, epithelial ovarian cancer, ovarian germ cell tumor, ovarian low malignant potential tumor, pancreatic cancer, papillomatosis, sinonasal cancer, parathyroid cancer, pelvic cancer, penile cancer, pharyngeal cancer, moderately differentiated pineal parenchymal tumor, pineoblastoma, pituitary tumor, plasmacytoma / multiple myeloma, pleuropulmonary blastoma, primary central nervous system (CNS) lymphoma, primary hepatocellular carcinoma, prostate cancer, rectal cancer, kidney cancer, renal cell (kidney) cancer, renal cell carcinoma, airway cancer, retinoblastoma, rhabdomyosarcoma, salivary gland cancer, Sézary syndrome, small cell lung cancer, small intestine cancer, soft tissue sarcoma, squamous cell carcinoma, cervical squamous cell carcinoma, gastric cancer cancer), supratentorial primitive neuroectodermal tumor, T-cell lymphoma, testicular cancer, throat cancer, thymic carcinoma, thymoma, thyroid cancer, transitional cell carcinoma, transitional cell carcinoma of the renal pelvis and ureter, trophoblastic tumor, ureteral cancer, urinary tract cancer, uterine cancer, uterine sarcoma, vaginal cancer, vulvar cancer, Waldenstrom's macroglobulinemia, or Wilms' tumor.

[0100] In another embodiment, the present invention provides a method for detecting cancer (e.g., breast cancer, bladder cancer, or colon cancer) in a blood sample or a fraction thereof from an individual, for example, an individual suspected of having cancer, comprising determining the single nucleotide variants present in the sample by using the ctDNA SNV amplification / sequencing workflow provided herein to determine the single nucleotide variants present in the ctDNA sample. At a plurality of single nucleotide loci, the presence of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, or 15 SNVs at the lower end of the range, or 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 30, 40, or 50 SNVs at the upper end of the range in the sample is indicative of the presence of cancer (e.g., breast cancer, bladder cancer, or colon cancer).

[0101] In another embodiment, provided herein is a method for detecting clonal single nucleotide variants in an individual's tumor (e.g., breast cancer, bladder cancer, or colon cancer). The method includes performing a ctDNA SNV amplification / sequencing workflow as provided herein and determining a variant allele frequency for each SNV locus based on the sequences of multiple copies of a set of amplicons. A relatively high allele frequency at a plurality of single nucleotide variant loci compared to other single nucleotide variants is indicative of a clonal single nucleotide variant in the tumor. Variant allele frequencies are well known in the art of sequencing. Support for this embodiment is provided, for example, in Figures 12-14.

[0102] In certain embodiments, the method further comprises determining a treatment plan, a therapy, and / or administering to the individual a compound that targets one or more clonal single nucleotide variants. In certain examples, subclones and / or other clonal SNVs are not targeted by the therapy. Specific therapies and associated mutations are provided in other sections of this specification and are known in the art. Thus, in certain examples, the method further comprises administering to the individual a compound that is known to be specifically effective in treating cancers (e.g., breast cancer, bladder cancer, or colon cancer) that have one or more of the determined single nucleotide variants.

[0103] In certain aspects of this embodiment, a variant allele frequency of greater than 0.25%, 0.5%, 0.75%, 1.0%, 5%, or 10% is indicative of a clonal single nucleotide variant. These cutoffs are supported by the data in tabular form in Figures 20A-20B.

[0104] In certain examples of this embodiment, the cancer is stage 1a, 1b, or 2a breast cancer, bladder cancer, or colon cancer. In certain examples of this embodiment, the cancer is stage 1a or 1b breast cancer, bladder cancer, or colon cancer. In certain examples of this embodiment, the individual does not undergo surgery. In a particular example of this embodiment, the individual does not undergo a biopsy.

[0105] In some examples of this embodiment, a clonal SNV is identified or further identified when other tests, such as direct tumor tests, suggest that for any SNV, the variable allele frequency is at least one-quarter, one-third, half, or three-quarters greater than other single nucleotide variants determined in the test, suggesting that the SNV under test is a clonal SNV.

[0106] In some embodiments, the methods herein for detecting SNVs in ctDNA may be used in place of direct analysis of DNA from tumors. The results provided herein show that SNVs that are significantly more likely to be clonal SNVs have a higher VAF (see Figures 12-14).

[0107] In some specific examples of the embodiments of the methods provided herein, before targeted amplification is performed on ctDNA from an individual, data is provided about the SNVs found in the tumor from that individual.Therefore, in these embodiments, SNV amplification / sequencing reaction is performed on one or more tumor samples from the individual.In this method, the ctDNA SNV amplification / sequencing reaction provided herein is still advantageous because it provides a liquid biopsy of clonal and subclonal mutations.Furthermore, as provided herein, clonal mutations can be more clearly identified in individuals with cancer (e.g., breast cancer, bladder cancer, or colon cancer) when a high VAF rate (e.g., VAF greater than 1, 2, 3, 4, 5, 6, 7, 8, 9, 10%) is determined for a certain SNV in ctDNA samples from the individual.

[0108] In certain embodiments, the method provided herein can be used to determine how to isolate and analyze ctDNA from circulating free nucleic acid from individuals with cancer (for example, breast cancer, bladder cancer or colon cancer).First, determine whether cancer is breast cancer, bladder cancer or colon cancer.If cancer is breast cancer, bladder cancer or colon cancer, isolate circulating free nucleic acid from individuals.In some instances, this method further comprises determining the stage of cancer.

[0109] In some methods, provided herein are compositions and / or solid supports of the invention. A composition comprising circulating tumor nucleic acid fragments comprising universal adaptors, wherein the circulating tumor nucleic acid was derived from breast cancer, bladder cancer, or colon cancer.

[0110] In some embodiments, provided herein are compositions comprising circulating tumor nucleic acid fragments comprising universal adaptors, wherein the circulating tumor nucleic acid was derived from a blood sample or fraction thereof from an individual with cancer (e.g., breast cancer, bladder cancer, or colon cancer). These methods typically involve the formation of ctDNA fragments comprising universal adaptors. Furthermore, such methods typically involve the formation of a solid support, particularly a solid support for high-throughput screening, comprising a clonal population of a plurality of nucleic acids, the clonal population comprising amplicons generated from a sample of circulating free nucleic acid, which is ctDNA. In exemplary embodiments based on the surprising results provided herein, the ctDNA was derived from cancer (e.g., breast cancer, bladder cancer, or colon cancer).

[0111] Similarly, provided herein as one embodiment of the present invention is a solid support comprising a clonal population of multiple nucleic acids, the clonal population comprising nucleic acid fragments generated from a sample of blood from an individual with cancer (e.g., breast cancer, bladder cancer, or colon cancer) or a sample of circulating free nucleic acids from a fraction thereof.

[0112] In certain embodiments, nucleic acid fragments in different clonal populations contain the same universal adaptor. Such compositions are typically formed during high-throughput sequencing reactions in the methods of the invention.

[0113] A clonal collection of nucleic acids may be derived from nucleic acid fragments from a set of samples from two or more individuals, in these embodiments, the nucleic acid fragments comprise one of a set of molecular barcodes corresponding to the samples in the set of samples.

[0114] Detailed analysis methods are provided herein as SNV Method 1 and SNV Method 2 in the analysis section of this specification. Any of the methods provided herein may further include an analysis step as provided herein. Thus, in a specific example, a method for determining whether a single nucleotide variant is present in a sample includes identifying a confidence value for each allele determination at each of a set of single nucleotide variant loci, which may be based at least in part on the read depth for the loci. Confidence limits can be set at at least 75%, 80%, 85%, 90%, 95%, 96%, 96%, 98%, or 99%. Confidence limits can be set at different levels for different types of mutations.

[0115] The method can be performed at a read depth for a set of at least 5, 10, 15, 20, 25, 50, 100, 150, 200, 250, 500, 1,000, 10,000, 25,000, 50,000, 100,000, 250,000, 500,000 or 1 million single nucleotide variant loci.

[0116] In certain embodiments, the method of any embodiment herein comprises determining the efficiency and / or error rate per cycle, which is determined for each amplification reaction of the multiplex amplification reaction of single nucleotide variant loci.The efficiency and error rate can then be used to determine whether a single nucleotide variant at a set of single variant loci is present in the sample.In certain embodiments, the more detailed analysis steps provided in the SNV method 2 provided in the analysis method can also be included.

[0117] In an exemplary embodiment, in any of the methods herein, the set of single nucleotide variant loci includes all of the single nucleotide variant loci identified in the TCGA and COSMIC datasets for cancer (e.g., breast cancer, bladder cancer, or colorectal cancer).

[0118] In certain embodiments of any of the methods herein, the set of single nucleotide variant loci includes 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 75, 100, 250, 500, 1000, 2500, 5000, or 10,000 single nucleotide variant loci known to be associated with cancer (e.g., breast cancer, bladder cancer, or colorectal cancer) at the lower end of the range, and 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 40, 50, 75, 100, 250, 500, 1000, 2500, 5000, 10,000, 20,000, and 25,000 single nucleotide variant loci known to be associated with cancer (e.g., breast cancer, bladder cancer, or colorectal cancer) at the higher end of the range.

[0119] In any of the methods for detecting SNVs herein, including ctDNA SNV amplification / sequencing workflow, improved amplification parameters for multiplex PCR may be used. For example, when the amplification reaction is a PCR reaction, the annealing temperature is 1, 2, 3, 4, 5, 6, 7, 8, 9 or 10°C higher than the melting temperature of at least 10, 20, 25, 30, 40, 50, 06, 70, 75, 80, 90, 95 or 100% of the primers in the set of primers at the lower end of the range, and 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14 or 15°C higher at the upper end of the range.

[0120] In certain embodiments, when the amplification reaction is a PCR reaction, the length of the annealing step in the PCR reaction is 10, 15, 20, 30, 45, and 60 minutes at the lower end of the range, and 15, 20, 30, 45, 60, 120, 180, or 240 minutes at the upper end of the range. In certain embodiments, the primer concentration in the amplification (e.g., PCR reaction) is 1 to 10 nM. Furthermore, in exemplary embodiments, the primers in the primer set are designed to minimize primer-dimer formation.

[0121] Thus, in one example of any of the methods herein that includes an amplification step, the amplification reaction is a PCR reaction, the annealing temperature is 1-10°C higher than the melting temperature of at least 90% of the primers in the primer set, the length of the annealing step in the PCR reaction is 15-60 minutes, the primer concentration in the amplification reaction is 1-10 nM, and the primers in the primer set are designed to minimize primer-dimer formation. In a further embodiment of this example, the multiplex amplification reaction is performed under limiting primer conditions.

[0122] In another embodiment, provided herein is a method for confirming a diagnosis of cancer (e.g., breast cancer, bladder cancer, or colon cancer) in an individual, e.g., an individual suspected of having cancer (e.g., breast cancer, bladder cancer, or colon cancer), from a sample of blood from the individual, or a fraction thereof, comprising performing a ctDNA SNV amplification / sequencing workflow provided herein to determine whether one or more single nucleotide variants are present at a plurality of single nucleotide variant loci. In this embodiment, the following elements, statements, guidelines, or rules apply: the absence of a single nucleotide variant supports a diagnosis of stage 1a, 1b, or 2a adenocarcinoma; the presence of a single nucleotide variant supports a diagnosis of squamous cell carcinoma or stage 2b or 3a adenocarcinoma; and / or the presence of 10 or more single nucleotide variants supports a diagnosis of squamous cell carcinoma or stage 2b or 3 adenocarcinoma.

[0123] These results identify analysis of lung ADC and SCC samples from individuals using the ctDNA SNV amplification / sequencing workflow as a valuable method for identifying SNVs found in ADC tumors, particularly for stage 2b and 3a ADC tumors, and particularly for SCC tumors at any stage (see Figure 15 and Figure 20A-B).

[0124] In certain embodiments, the methods herein for detecting SNVs may be used to prescribe a treatment regimen. Therapies targeting specific mutations associated with ADC and SCC are available and under development (Nature Review Cancer. 14:535-551 (2014)). For example, detecting EGFR mutations at L858R or T790M may be useful for selecting a therapy. Erlotinib, gefitinib, afatinib, AZK9291, CO-1686, and HM61713 are currently approved therapies in the United States and in clinical trials that target specific EGFR mutations. In another example, a G12D, G12C, or G12V mutation in KRAS may be used to prescribe a combination of selumetinib and docetaxel to an individual. In another example, a V600E mutation in BRAF may be used to prescribe a treatment regimen for a subject with vemurafenib, dabrafenib, or trametinib.

[0125] In certain exemplary embodiments, the sample analyzed by the method of the present invention is blood sample or its fraction.In certain embodiments, the method provided herein is particularly adapted to amplify DNA fragment, particularly the tumor DNA fragment found in circulating tumor DNA (ctDNA).This fragment is typically about 160 nucleotides long.

[0126] It is known in the art that cell-free nucleic acids (cfNAs), such as cfDNA, can be released into the circulation through various forms of cell death, such as apoptosis, necrosis, autophagy, and necroptosis. cfDNA is fragmented, and the size distribution of the fragments varies from 150-350 bp to over 10,000 bp. (See Kalnina et al., World J Gastroenterol. 2015 Nov. 7;21(41):11636-11653.) For example, the size distribution of plasma DNA fragments in hepatocellular carcinoma (HCC) patients ranges from 100-220 bp in length, with a peak frequency of approximately 166 bp, and the highest tumor DNA concentration among the fragments is 150-180 bp in length. (See Jiang et al., Proc Natl Acad Sci USA 112:E1317-E1325.)

[0127] In an exemplary embodiment, circulating tumor DNA (ctDNA) is isolated from blood using EDTA-2Na tubes after centrifugation to remove cellular debris and platelets. Plasma samples may be stored at -80°C until DNA is extracted, for example, using a QIAamp DNA Mini Kit (Qiagen, Hilden, Germany) (see, e.g., Hamakawa et al., Br J Cancer. 2015;112:352-356). reported that the median concentration of extracted cell-free DNA across all samples was 43.1 ng / ml of plasma (range 9.5-1338 ng / ml), with a mutant frequency range of 0.001-77.8%, with a median of 0.90%.

[0128] In certain exemplary embodiments, sample is tumor.In view of the teachings of this specification, the method of isolating nucleic acid from tumor and the method of making nucleic acid library from this DNA sample are known in the art.In addition, in view of the teachings of this specification, those skilled in the art will know how to make nucleic acid library suitable for the method of this specification from other samples, such as other liquid samples that DNA is floating in free state in addition to ctDNA sample.

[0129] In certain embodiments, the methods of the present invention typically include the step of creating and amplifying a nucleic acid library from a sample (i.e., library preparation). The nucleic acids from the sample during the library preparation step may have attached ligation adapters (often referred to as library tags or ligation adapter tags (LTs)), which contain universal priming sequences followed by universal amplification. In one embodiment, this may be performed using standard protocols designed to create sequencing libraries after fragmentation. In one embodiment, the DNA sample may be blunt-ended, and then an A may be added to its 3' end. A Y adapter with a T overhang may be added and ligated. In some embodiments, other sticky ends other than A or T overhangs may be used. In some embodiments, other adapters, such as looped ligation adapters, may be added. In some embodiments, the adapters may have tags designed for PCR amplification.

[0130] Some embodiments provided herein include detecting SNVs in a ctDNA sample. In exemplary embodiments, such a method includes an amplification step and a sequencing step (sometimes referred to herein as a "ctDNA SNV amplification / sequencing workflow"). In an exemplary example, the ctDNA amplification / sequencing workflow may include: performing a multiplex amplification reaction on nucleic acids isolated from a blood sample or a fraction thereof from an individual, for example, an individual suspected of having cancer (e.g., breast cancer, bladder cancer, or colon cancer), to generate a set of amplicons, where each amplicon in the set of amplicons spans at least one single nucleotide variant locus of a set of single nucleotide variant loci, for example, an SNV locus known to be associated with cancer (e.g., breast cancer, bladder cancer, or colon cancer); and determining the sequence of at least one segment of each amplicon in the set of amplicons, where the segment includes the single nucleotide variant locus. In this manner, this exemplary method determines the single nucleotide variants present in the sample.

[0131] An exemplary ctDNA SNV amplification / sequencing workflow may, in more detail, comprise forming an amplification reaction mixture by combining a polymerase, nucleotide triphosphates, and nucleic acid fragments from a nucleic acid library generated from the sample with a set of primers that each bind at an effective distance from the single-nucleotide variant loci or a set of primer pairs that each span an effective region that includes the single-nucleotide variant loci. In exemplary embodiments, the single-nucleotide variant loci are known to be associated with cancer (e.g., breast cancer, bladder cancer, or colon cancer). The amplification reaction mixture is then subjected to amplification conditions to produce a set of amplicons, preferably including at least one single-nucleotide variant locus from the set of single-nucleotide variant loci known to be associated with cancer (e.g., breast cancer, bladder cancer, or colon cancer), and determining the sequence of at least one segment of each amplicon in the set of amplicons, wherein the segment includes the single-nucleotide variant locus.

[0132] The effective distance of primer binding may be within 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40, 45, 50, 75, 100, 125, or 150 base pairs of the SNV locus. The effective range spanned by a pair of primers typically includes the SNV and is typically 160 base pairs or less, and may be 150, 140, 130, 125, 100, 75, 50, or 25 base pairs or less. In other embodiments, the primer pair spans a working range of 20, 25, 30, 40, 50, 60, 70, 75, 100, 110, 120, 125, 130, 140, or 150 nucleotides at the lower end of the range, or 25, 30, 40, 50, 60, 70, 75, 100, 110, 120, 125, 130, 140, or 150, 160, 170, 175, or 200 nucleotides at the upper end of the range from the SNV locus.

[0133] Further details regarding amplification methods that can be used in a ctDNA SNV amplification / sequencing workflow to detect SNVs for use in the methods herein are provided in other sections of this specification.

[0134] SNV call analysis During the method provided herein, nucleic acid sequencing data is generated for the amplicons generated by tiled multiplex PCR.This data can be analyzed and / or adapted to determine whether a mutation (e.g., SNV) exists in the target gene within a certain confidence limit.

[0135] Sequencing reads may be demultiplexed using in-house tools and mapped in single-end mode using pair-merged reads against the hg19 genome using the Bwa mem function of the Burrows-Wheeler alignment software (BWA, Burrows-Wheeler Alignment Software (see Li H. and Durbin R. (2010) Fast and accurate long-read alignment with Burrows-Wheeler Transform. Bioinformatics, Epub. [PMID:20080505]). Amplification statistics QC can be performed by analyzing total reads, number of mapped reads, number of mapped reads on target, and number of measured reads.

[0136] In certain embodiments, any analytical method for detecting SNVs from nucleic acid sequencing data detection may be used with the methods of the present invention that include detecting SNVs or determining whether an SNV is present. In certain exemplary embodiments, methods of the present invention that utilize SNV Method 1 below are used. In yet other exemplary embodiments, methods of the present invention that include detecting SNVs or determining whether an SNV is present at an SNV locus utilize SNV Method 2 below.

[0137] SNV Method 1: In this embodiment, a background error model is constructed using normal plasma samples, which are sequenced in the same sequencing run to account for run-specific artifacts. In certain embodiments, 5, 10, 15, 20, 25, 30, 40, 50, 100, 150, 200, 250, or more than 250 normal plasma samples are analyzed in the same sequencing run. In certain exemplary embodiments, 20, 25, 40, or 50 normal plasma samples are analyzed in the same sequencing run. Noise positions with a median common variant allele frequency greater than a cutoff are removed. For example, in certain embodiments, this cutoff is greater than 0.1%, 0.2%, 0.25%, 0.5%, 1%, 2%, 5%, or 10%. In certain exemplary embodiments, noise positions with a median common variant allele frequency greater than 0.5% are removed. To account for noise and contamination, outlier samples are repeatedly removed from this model. In certain embodiments, samples with Z-scores greater than 5, 6, 7, 8, 9 or 10 are removed from data analysis. For each base substitution of all genomic loci, the read depth-weighted mean and standard deviation of error are calculated. The position of tumor or cell-free plasma sample that has at least 5 variant reads and has a Z-score of 10 against background error model can be called, for example, as a candidate mutation.

[0138] SNV Method 2: In this embodiment, single nucleotide variants (SNVs) are determined using plasma ctDNA data. The PCR process is modeled as a stochastic process, and parameters are estimated using a training set to generate final SNV calls for a separate test set. Error propagation over multiple PCR cycles is determined, and the mean and variance of background error are calculated; in an exemplary embodiment, background error is distinguished from actual mutations.

[0139] For each base, the following parameters are estimated: p = efficiency (probability that each read is replicated during each cycle) p e = error rate per cycle for variant e (probability of an error of type e occurring) X0 = initial number of molecules

[0140] As a read is replicated over a series of PCR processes, more errors are introduced. Thus, the error profile of a read is determined by its degree of separation from the original read. A read is said to be of generation k if it has undergone k replications before it is created.

[0141] For each base, let's define the following variables: X ij = number of i-th generation reads produced in PCR cycle j Y ij = total number of reads in generation i at the end of cycle j X ij e = number of i-th generation reads with mutation e generated in PCR cycle j

[0142] Furthermore, in addition to the normal molecule X0, an additional f that has the mutation e at the start of the PCR process e If X0 molecules are present (thus fe / (1+fe) will be the fraction of mutated molecules in the initial mixture).

[0143] Given the total number of i-1 generation reads in cycle j-1, the number of i-th generation reads produced in cycle j is determined by the sample size Y i-1,j-1 and has a binomial distribution with probability parameter p. Therefore, E(X ij ,|Y i-1,j-1 ,p)=pY i-1,j-1 and Var(X ij ,|Y i-1,j-1 ,p)=p(1-p)Y i-1,j-1 is.

[0144] The present inventors

number

[0145] Finally, E(Xi j e |Y i-1,j-1 ,p e )=p e Y i-1,j-1 and Var(X ij e |Y i-1,j-1 ,p)=p e (1-p e )Y i-1,j-1 Using these, E(X ij e ) and Var(X ij e ) can be calculated.

[0146] In certain embodiments, SNV method 2 is performed as follows.

[0147] a) The training dataset is used to estimate PCR efficiency and error rate per cycle.

[0148] b) Using the distribution of efficiencies estimated in step (a), estimate the number of starting molecules for the test data set at each base.

[0149] c) If necessary, update the efficiency estimate for the test dataset using the number of starting molecules estimated in step (b).

[0150] d) Using the test set data and parameters estimated in steps (a), (b), and (c), estimate the mean and variance for the total number of molecules, background error molecules, and actual mutant molecules (for a search space consisting of the initial proportion of actual mutant molecules).

[0151] e) Fit the distribution to the number of total error molecules (background errors and actual mutations) in the total molecules to calculate the likelihood of the proportion of each actual mutation in the search space.

[0152] f) determining the proportion of most likely actual mutations and calculating the confidence using the data from step (e);

[0153] A confidence cutoff can be used to identify SNVs at SNV loci, for example, a 90%, 95%, 96%, 97%, 98%, or 99% confidence cutoff can be used to call SNVs.

[0154] Exemplary SNV Method 2 Algorithm The algorithm begins by estimating the efficiency and error rate per cycle using a training set, where n denotes the total number of PCR cycles.

[0155] Read R at each base b b The number of b ) n It can be approximated by X0, where p b is the efficiency at base b. (R b / X0) 1 / n Using 1+p b Then, over all training samples, p b The mean and standard deviation of can be determined to estimate the parameters of the probability distribution for each base (e.g., normal, beta, or similar distribution).

[0156] Similarly, the read R of error e at each base b be Using the number of e After determining the mean and standard deviation of the error rate across all training samples, its probability distribution (e.g., normal, beta, or similar) is estimated, and its parameters are estimated using this mean and standard deviation value.

[0157] Next, for the test data, the initial starting copies at each base are

number

number

[0158] Therefore, this parameter is estimated and used in the stochastic process, and these estimates can then be used to estimate the mean and variance of the molecules created in each cycle (separately for normal, error, and mutant molecules).

[0159] Finally, by using a probabilistic method (e.g., maximum likelihood or similar), the best f that best fits the distribution of errors, mutations, and normal molecules is determined. e More specifically, various f values ​​in the final read can be determined. e For each value, we estimate the expected ratio of error molecules to total molecules, determine the likelihood of our data for each of these values, and then select the value with the greatest likelihood.

[0160] Primer tails can improve detection of fragmented DNA from universally tagged libraries. When the library tag and primer tail contain homologous sequences, hybridization can be improved (e.g., lowering the melting temperature (Tm)) and primer extension can occur when only a portion of the primer target sequence is present in the sample DNA primer fragment. In some embodiments, 13 or more target-specific base pairs may be used. In some embodiments, 10-12 target-specific base pairs may be used. In some embodiments, 8-9 target-specific base pairs may be used. In some embodiments, 6-7 target-specific base pairs may be used.

[0161] In one embodiment, a library is created from the sample by ligating adapters to the ends of DNA fragments in the sample or to the ends of DNA fragments created from DNA isolated from the sample. The fragments can then be amplified using PCR, for example, according to the following exemplary protocol.

[0162] 95°C for 2 min; 15× [95°C for 20 s, 55°C for 20 s, 68°C for 20 s], 68°C for 2 min, hold at 4°C.

[0163] Many kits and methods are known in the art for creating nucleic acid libraries containing universal primer binding sites for subsequent amplification (e.g., clonal amplification) and subsequent sequencing. To facilitate adapter ligation, library preparation and amplification may include end repair and adenylation (i.e., A-tailing). Kits specifically adapted for preparing libraries from small nucleic acid fragments (especially circulating free DNA) may be useful for performing the methods provided herein. For example, the NEXTflex Cell Free Kit available from Bioo Scientific or the Natera Library Prep Kit (available from Natera, Inc., San Carlos, CA). However, such kits are typically modified to include customized adapters for the amplification and sequencing steps of the methods provided herein. Adapter ligation can be performed using commercially available kits, such as the ligation kit found in the AGILENT SURESELECT kit (Agilent, CA).

[0164] The target region of a nucleic acid library made from DNA isolated from a sample, particularly a circulating free DNA sample for the method of the present invention, is then amplified. For this amplification, the set of primers or primer pairs may include 5, 10, 15, 20, 25, 50, 100, 125, 150, 250, 500, 1000, 2500, 5000, 10,000, 20,000, 25,000, or 50,000 primers at the lower end of the range, and 15, 20, 25, 50, 100, 125, 150, 250, 500, 1000, 2500, 5000, 10,000, 20,000, 25,000, 50,000, 60,000, 75,000, or 100,000 primers at the upper end of the range, each binding to one of the set of primer binding sites.

[0165] Primer designs may be generated with Primer3 (Untergrasser A, Cutcutache I, Koressaar T, Ye J, Faircloth BC, Remm M, Rozen SG (2012) "Primer3 - new capabilities and interfaces." Nucleic Acids Research 40(15):e115 and Koressaar T, Remm M (2007) "Enhancements and modifications of primer design program Primer3." Bioinformatics 23(10):1289-91. Source code is available at primer3.sourceforge.net). Primer specificity may be assessed by BLAST, which may be added to existing primer design pipeline criteria.

[0166] Primer specificity can be determined using the BLASTn program from the ncbi-blast-2.2.29+ package. The task option "blastn-short" can be used to map primers to the hg19 human genome. A primer design can be determined to be "specific" if the primer has fewer than 100 hits to the genome, and the top hit is the target-complementary primer-binding region of that genome and scores at least 2 points higher than the other hits (scores are defined by the BLASTn program). This can be done to have a unique hit to the genome and not many other hits throughout the genome.

[0167] The final selected primers can be visualized in IGV (James T. Robinson, Helga Thorvaldsdottir, Wendy Winckler, Mitchell Guttman, Eric S. Lander, Gad Getz, Jill P. Mesirov. Integrative Genomics Viewer. Nature Biotechnology 29, 24-26 (2011)) and the UCSC browser (Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, Haussler D. The human genome browser at UCSC. Genome Res. 2002 June;12(6):996-1006) with a BED file for validation and a coverage map.

[0168] In certain embodiments, the methods of the present invention include forming an amplification reaction mixture. This reaction mixture is typically created by combining a polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library created from a sample, and a set of forward and reverse primers specific to a target region containing an SNV. In exemplary embodiments, the reaction mixtures provided herein form a separate aspect of the present invention.

[0169] Amplification reaction mixtures useful in the present invention contain components known in the art for nucleic acid amplification, particularly PCR amplification. For example, reaction mixtures typically contain nucleotide triphosphates, polymerase, and magnesium. Polymerases useful in the present invention may include any polymerase that can be used in amplification reactions, particularly those useful in PCR reactions. In certain embodiments, hot-start Taq polymerase is particularly useful. Amplification reaction mixtures useful for carrying out the methods provided herein, such as AmpliTaq Gold Master Mix (Life Technologies, Carlsbad, CA), are commercially available.

[0170] PCR amplification (e.g., temperature cycling) conditions are well known in the art. The methods provided herein may include any PCR cycling conditions that allow amplification of target nucleic acids (e.g., target nucleic acids from a library). Non-limiting exemplary cycling conditions are provided in the Examples section of this specification.

[0171] There are many possible workflows when carrying out PCR, and some workflows are provided herein that are typical for the method disclosed herein.The steps outlined herein are not meant to exclude other possible steps, nor do they imply that any of the steps described herein are necessary for the method to function properly.Many parameter variations or other modifications are known in the literature and can be made without affecting the essence of the present invention.

[0172] In certain embodiments of the methods provided herein, at least a portion, and in illustrative examples, the entire sequence of the amplicon (e.g., outer primer-targeted amplicon) is determined. Methods for determining the sequence of an amplicon are known in the art. Any sequencing method known in the art, such as Sanger sequencing, can be used to determine such a sequence. In exemplary embodiments, high-throughput next-generation sequencing technologies (also referred to herein as massively parallel sequencing technologies), such as, but not limited to, those used in MYSEQ (ILLUMINA), HISEQ (ILLUMINA), ION TORRENT (LIFE TECHNOLOGIES), GENOME ANALYZER ILX (ILLUMINA), and GS FLEX+ (ROCHE 454), can be used to sequence the amplicons produced by the methods provided herein.

[0173] High-throughput gene sequencers can be modified to accommodate the use of barcoding (i.e., sample tagging with a distinctive nucleic acid sequence) to identify unique samples from individuals, thereby enabling simultaneous analysis of multiple samples in a single DNA sequencer run. The number of times a given region of the genome is sequenced (number of reads) in a library preparation (or other nucleic acid preparation of interest) will be proportional to the copy number (or expression level in the case of a preparation containing cDNA) of that sequence in the genome of interest. Bias in amplification efficiency may be taken into account in such quantitative determinations.

[0174] Target gene In exemplary embodiments, the target gene of the present invention is a cancer-related gene, and in many exemplary embodiments, it is a cancer-related gene.Cancer-related genes (for example, cancer-related genes or bladder cancer-related genes or colon cancer-related genes) refer to genes that are associated with changes in the risk of cancer (for example, breast cancer, bladder cancer or colon cancer) or changes in the prognosis of cancer.Exemplary cancer-related genes that promote cancer include oncogenes, genes that promote cell proliferation, invasion or metastasis, genes that inhibit apoptosis, and pro-angiogenic genes.Cancer-related genes that inhibit cancer include, but are not limited to, tumor suppressor genes, genes that inhibit cell proliferation, invasion or metastasis, genes that promote apoptosis, and anti-angiogenic genes.

[0175] One embodiment of the mutation detection method begins with the selection of a region of a gene to target. The region with a known mutation is used to develop primers for mPCR-NGS to amplify and detect the mutation.

[0176] The methods provided herein can be used to detect virtually any type of mutation, particularly mutations known to be associated with cancer, and most particularly, the methods provided herein are directed to mutations (particularly SNVs) associated with cancer, particularly breast cancer, bladder cancer or colon cancer.Exemplary SNVs may be in one or more of the following genes: EGFR, FGFR1, FGFR2, ALK, MET, ROS1, NTRK1, RET, HER2, DDR2, PDGFRA, KRAS, NF1, BRAF, PIK3CA, MEK1, NOTCH1, MLL2, EZH2, TET2, DNMT3A, SOX2, MYC, KEAP1, CDKN2A, NRG1, TP53, LKB1, and PTEN have been identified as mutated, or in increased copy number, or fused to other genes, and combinations thereof, in various lung cancer samples (Non-small-cell lung cancers: a heterogeneous set of diseases. Chen et al., Nat. Rev. Cancer. 2014 Aug. 14(8):535-551). In another example, the list of genes is as listed above, and the SNVs are as reported, for example, in the Chen et al. reference.

[0177] Amplification (e.g., PCR) reaction mixture: In certain embodiments, the methods of the present invention include forming an amplification reaction mixture. This reaction mixture is typically formed by combining a polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library created from a sample with a set of forward target-specific outer primers and a first-strand reverse outer universal primer. Another exemplary embodiment is a reaction mixture that includes a forward target-specific inner primer instead of the forward target-specific outer primer, and amplicons from a first PCR reaction using the outer primer instead of the nucleic acid fragments from the nucleic acid library. In exemplary embodiments, the reaction mixture provided herein forms a separate aspect of the present invention. In exemplary embodiments, the reaction mixture is a PCR reaction mixture. The PCR reaction mixture typically includes magnesium.

[0178] In some embodiments, the reaction mixture contains ethylenediaminetetraacetic acid (EDTA), magnesium, tetramethylammonium chloride (TMAC), or any combination thereof. In some embodiments, the concentration of TMAC is 20 to 70 mM, inclusive. Without being bound by any particular theory, it is believed that TMAC binds to DNA, stabilizes the duplex, increases primer specificity, and / or equalizes the melting temperatures of different primers. In some embodiments, TMAC increases the uniformity of the amount of amplification product for different targets. In some embodiments, the concentration of magnesium (e.g., magnesium from magnesium chloride) is 1 to 8 mM.

[0179] The large number of primers used in multiplex PCR of multiple targets can chelate a lot of magnesium (two phosphate groups in a primer chelate one magnesium). For example, if enough primers are used so that the concentration of primer-derived phosphate groups is about 9 mM, the primers can reduce the effective magnesium concentration to about 4.5 mM. In some embodiments, EDTA is used to reduce the amount of magnesium available as a polymerase cofactor, because high concentrations of magnesium can cause PCR errors (e.g., amplification of non-target loci). In some embodiments, the concentration of EDTA reduces the amount of available magnesium to 1-5 mM (e.g., 3-5 mM).

[0180] In some embodiments, the pH is 7.5 to 8.5, e.g., 7.5 to 8, 8 to 8.3, or 8.3 to 8.5 (inclusive). In some embodiments, Tris is used at a concentration of, e.g., 10 to 100 mM, e.g., 10 to 25 mM, 25 to 50 mM, 50 to 75 mM, or 25 to 75 mM (inclusive). In some embodiments, Tris at any of these concentrations is used at a pH of 7.5 to 8.5. In some embodiments, a combination of KCl and (NH4)2SO4 is used, e.g., 50 to 150 mM KCl and 10 to 90 mM (NH4)2SO4 (inclusive). In some embodiments, the concentration of KCl is 0 to 30 mM, 50 to 100 mM, or 100 to 150 mM (inclusive). In some embodiments, the concentration of (NH4)2SO4 is 10-50 mM, 50-90 mM, 10-20 mM, 20-40 mM, 40-60 mM, or 60-80 mM (NH4)2SO4, inclusive. + ] concentration is 0 to 160 mM, for example, 0 to 50, 50 to 100, or 100 to 160 mM (boundaries included). In some embodiments, the sum of the potassium concentration and the ammonium concentration ([K + ]+[NH4 +]) is 0 to 160 mM, for example, 0 to 25, 25 to 50, 50 to 150, 50 to 75, 75 to 100, 100 to 125, or 125 to 160 mM (including the boundary values). + ]+[NH4 + An exemplary buffer with a pH of 0.05 is 20 mM KCl and 50 mM (NH4)2SO4. In some embodiments, the buffer contains 25-75 mM Tris (pH 7.2-8), 0-50 mM KCl, 10-80 mM ammonium sulfate, and 3-6 mM magnesium (inclusive). In some embodiments, the buffer contains 25-75 mM Tris (pH 7-8.5), 3-6 mM MgCl2, 10-50 mM KCl, and 20-80 mM (NH4)2SO4 (inclusive). In some embodiments, 100-200 units / mL of polymerase are used. In some embodiments, 7 ul of DNA template in a final volume of 20 ul in 100 mM KCl, 50 mM (NH4)2SO4, 3 mM MgCl2, 7.5 nM of each primer in the library and pH 8.1 is used.

[0181] In some embodiments, a crowding agent, such as polyethylene glycol (PEG, e.g., PEG 8,000) or glycerol, is used. In some embodiments, the amount of PEG (e.g., PEG 8,000) is 0.1-20%, e.g., 0.5-15%, 1-10%, 2-8%, or 4-8%, inclusive. In some embodiments, the amount of glycerol is 0.1-20%, e.g., 0.5-15%, 1-10%, 2-8%, or 4-8%, inclusive. In some embodiments, the crowding agent allows for the use of either a lower polymerase concentration and / or a shorter annealing time. In some embodiments, the crowding agent improves the uniformity of DOR and / or reduces dropouts (undetected alleles). In some embodiments, a polymerase with proofreading activity, a polymerase without (or negligible) proofreading activity, or a mixture of a polymerase with and without (or negligible) proofreading activity is used. In some embodiments, a hot-start polymerase, a non-hot-start polymerase, or a mixture of a hot-start polymerase and a non-hot-start polymerase is used. In some embodiments, HotStarTaq DNA polymerase is used (see, e.g., QIAGEN catalog number 203203). In some embodiments, AmpliTaq Gold® DNA polymerase is used. In some embodiments, PrimeSTAR GXL DNA polymerase is used (Takara Clontech, Mountain View, CA), a high-fidelity polymerase that provides efficient PCR amplification when excess template is present in the reaction mixture and when amplifying long products. In some embodiments, KAPA Taq DNA polymerase or KAPA Taq HotStart DNA polymerase is used. These are derived from the single-subunit wild-type Taq DNA polymerase of the thermophilic bacterium Thermus aquaticus.KAPA Taq and KAPA Taq HotStart DNA Polymerase have 5'-3' polymerase activity and 5'-3' exonuclease activity, but not 3' to 5' exonuclease (proofreading) activity (see, e.g., KAPA BIOSYSTEMS catalog number BK1000). In some embodiments, Pfu DNA polymerase is used. This polymerase is a thermostable DNA polymerase derived from the hyperthermophilic archaeon Pyrococcus furiosus. This enzyme catalyzes the template-dependent polymerization of nucleotides into double-stranded DNA in the 5' to 3' direction. Pfu DNA Polymerase also exhibits 3' to 5' exonuclease (proofreading) activity, allowing the polymerase to correct nucleotide incorporation errors. This polymerase does not have 5' to 3' exonuclease activity (see, e.g., Thermo Scientific catalog number EP0501). In some embodiments, Klentaq1 is used. It is a Klenow fragment analog of Taq DNA polymerase and does not have exonuclease or endonuclease activity (see, e.g., DNA POLYMERASE TECHNOLOGY, Inc., St. Louis, Missouri, catalog number 100). In some embodiments, the polymerase is a PHUSION DNA polymerase, such as PHUSION High Fidelity DNA polymerase (M0530S, New England BioLabs, Inc.) or PHUSION Hot Start Flex DNA polymerase (M0535S, New England BioLabs, Inc.). In some embodiments, the polymerase is Q5® DNA polymerase, e.g., Q5® High-Fidelity DNA Polymerase (M0491S, New England BioLabs, Inc.) or Q5® Hot Start High-Fidelity DNA Polymerase (M0493S, New England BioLabs, Inc.). In some embodiments, the polymerase is T4 DNA polymerase (M0203S, New England BioLabs, Inc.).

[0182] In some embodiments, 5 to 600 units / mL (units per mL of reaction volume), for example, 5 to 100, 100 to 200, 200 to 300, 300 to 400, 400 to 500, or 500 to 600 units / mL (inclusive) of polymerase is used.

[0183] PCR method In some embodiments, hot-start PCR is used to reduce or prevent polymerization before PCR thermal cycling. Exemplary hot-start PCR methods include initial inhibition of DNA polymerase or physical separation of reaction components until the reaction mixture reaches a higher temperature. In some embodiments, delayed release of magnesium is used. Because DNA polymerase requires magnesium ions for activity, magnesium is chemically separated from the reaction by binding to a chemical compound and released into solution only at elevated temperatures. In some embodiments, non-covalent binding of an inhibitor is used. In this method, a peptide, antibody, or aptamer non-covalently binds to the enzyme at low temperatures and inhibits its activity. After incubation at elevated temperatures, the inhibitor is released and the reaction begins. In some embodiments, a cold-sensitive Taq polymerase is used, e.g., a modified DNA polymerase that has little activity at low temperatures. In some embodiments, chemical modification is used. In this method, a molecule is covalently attached to the side chain of an amino acid in the active site of the DNA polymerase. The molecule is released from the enzyme by incubating the reaction mixture at elevated temperatures. Once released, the enzyme is activated.

[0184] In some embodiments, the amount of nucleic acid (e.g., RNA or DNA sample) for template assembly is 20 to 5,000 ng, e.g., 20 to 200, 200 to 400, 400 to 600, 600 to 1,000, 1,000 to 1,500, or 2,000 to 3,000 ng, inclusive.

[0185] In some embodiments, a QIAGEN Multiplex PCR Kit is used (QIAGEN catalog number 206143). For 100 x 50 μl multiplex PCR reactions, the kit contains 2 x QIAGEN Multiplex PCR Master Mix (3 x 0.85 ml, providing a final concentration of 3 mM MgCl), 5 x Q-Solution (1 x 2.0 ml), and RNase-Free Water (2 x 1.7 ml). QIAGEN Multiplex PCR Master Mix (MM) contains a combination of KCl and (NH4)2SO4, as well as the PCR additive Factor MP, which increases the local concentration of primers at the template. Factor MP stabilizes specifically bound primers, allowing efficient primer extension by HotStarTaq DNA Polymerase. HotStarTaq DNA Polymerase is a modified form of Taq DNA polymerase that has no polymerase activity at ambient temperature. In some embodiments, HotStarTaq DNA Polymerase is activated by incubation at 95°C for 15 minutes, which can be incorporated into any existing thermal cycler program.

[0186] In some embodiments, a final concentration of 1×QIAGEN MM (recommended concentration), 7.5 nM of each primer in the library, 50 mM TMAC, and 7 ul of DNA template in a final volume of 20 ul are used. In some embodiments, PCR thermal cycling conditions include 95° C. for 10 minutes (hot start), 20 cycles of 96° C. for 30 seconds, 65° C. for 15 minutes, 72° C. for 30 seconds, followed by 72° C. for 2 minutes (final extension), then a 4° C. hold.

[0187] In some embodiments, a final concentration of 2x QIAGEN MM (twice the recommended concentration), 2 nM of each primer in the library, 70 mM TMAC, and 7 ul of DNA template in a total volume of 20 ul are used. In some embodiments, up to 4 mM EDTA is also included. In some embodiments, PCR thermal cycling conditions include 95°C for 10 minutes (hot start), 96°C for 30 seconds, 65°C for 20, 25, 30, 45, 60, 120, or 180 minutes, optionally 72°C for 30 seconds (25 cycles), followed by 72°C for 2 minutes (final extension), then a hold at 4°C.

[0188] Another exemplary set of conditions involves a semi-nested PCR approach: The first PCR reaction uses a 20 ul reaction volume containing 2x QIAGEN MM final concentration, 1.875 nM of each primer in the library (forward and reverse outer primers) and DNA template. The thermal cycling parameters included 95°C for 10 minutes, 96°C for 30 seconds, 65°C for 1 minute, 58°C for 6 minutes, 60°C for 8 minutes, 25 cycles of 65°C for 4 minutes and 72°C for 30 seconds, followed by 72°C for 2 minutes, followed by a 4°C hold. Two microliters of the resulting product, diluted 1:200, was then used as input for a second PCR reaction. This reaction used a 10-microliter reaction volume containing a final concentration of 1x QIAGEN MM, 20 nM of each inner forward primer, and 1 μM of the reverse primer tag. The thermal cycling parameters included 95°C for 10 minutes, 95°C for 30 seconds, 65°C for 1 minute, 60°C for 5 minutes, 15 cycles of 65°C for 5 minutes and 72°C for 30 seconds, followed by 72°C for 2 minutes, followed by a 4°C hold. The annealing temperature may optionally be higher than the melting temperature of some or all of the primers, as discussed herein (see U.S. Patent Application No. 14 / 918,544, filed October 20, 2015, which is incorporated by reference in its entirety).

[0189] Melting point (T mThe annealing temperature (T) is the temperature at which half (50%) of the DNA duplex of an oligonucleotide (e.g., a primer) and its perfect complement dissociates into single-stranded DNA. A ) is the temperature at which the PCR protocol is run. For conventional methods, this temperature is usually the lowest T of the primers used. m Because the temperature is 5°C lower than the T temperature, nearly all possible duplexes are formed (so that virtually all primer molecules bind to the template nucleic acid). Although this is highly efficient, lower temperatures are sure to result in more non-specific reactions. A One consequence of having T that is too low is that a primer may anneal to a sequence other than the true target because an internal single-base mismatch or partial annealing may be tolerated. A is T m At higher temperatures, at a given moment, only a small fraction of the target has primers annealed (e.g., only about 1-5%). As these are extended, they are removed from the equilibrium of primer and target annealing and dissociation (extension is called T m By rapidly increasing the temperature above 70 °C, approximately 1-5% of the new targets will have primers. Therefore, by allowing the reaction to run for a long time for annealing, it is possible to get approximately 100% of the targets copied per cycle.

[0190] In various embodiments, the annealing temperature is at least 25, 50, 60, 70, 75, 80, 90, 95, or 100% of the melting points (e.g., empirically determined or calculated T m1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13°C to 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13 or 15°C higher at the upper end of the range than the melting temperature (e.g., empirically measured or calculated T ) of the non-identical primers. In various embodiments, the annealing temperature is at least 25, 50, 75, 100, 300, 500, 750, 1,000, 2,000, 5,000, 7,500, 10,000, 15,000, 19,000, 20,000, 25,000, 27,000, 28,000, 30,000, 40,000, 50,000, 75,000, 100,000, or all of the melting temperatures (e.g., empirically measured or calculated T ). m In various embodiments, the annealing temperature is 1 to 15°C (e.g., 1 to 10, 1 to 5, 1 to 3, 3 to 5, 5 to 10, 5 to 8, 8 to 10, 10 to 12, or 12 to 15°C, inclusive) higher than the melting points (e.g., empirically measured or calculated T ) of at least 25%, 50%, 60%, 70%, 75%, 80%, 90%, 95% or all of the non-identical primers. m ), and the length of the annealing step (per PCR cycle) is 5 to 180 minutes, e.g., 15 to 120 minutes, 15 to 60 minutes, 15 to 45 minutes, or 20 to 60 minutes (inclusive).

[0191] Exemplary Multiplex PCR Methods In various embodiments, long annealing times (as discussed herein and exemplified in Example 10) and / or low primer concentrations are used. Indeed, in certain embodiments, limited primer concentrations and / or conditions are used. In various embodiments, the length of the annealing step ranges from 15, 20, 25, 30, 35, 40, 45, or 60 minutes at the lower end of the range to 20, 25, 30, 35, 40, 45, 60, 120, or 180 minutes at the higher end of the range. In various embodiments, the length of the annealing step (per PCR cycle) is 30 to 180 minutes. For example, the annealing step may be 30 to 60 minutes, and the concentration of each primer may be less than 20, 15, 10, or 5 nM. In other embodiments, primer concentrations are from 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20 or 25 nM at the lower end of the range to 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25 and 50 nM at the upper end of the range.

[0192] At high levels of multiplexing, the solution may become viscous due to the large amount of primers in the solution. If the solution is too viscous, the primer concentration may be reduced to a level that is still sufficient for the primers to bind to the template DNA. In various embodiments, 1,000 to 100,000 different primers are used, and the concentration of each primer is less than 20 nM, for example, less than 10 nM or 1 to 10 nM (boundary values ​​included).

[0193] Copy number variation (CNV) detection In addition to SNVs and indels, the methods for monitoring and detecting early recurrence and metastasis described herein can also benefit from the detection of CNVs.

[0194] In one aspect, the present invention generally relates, at least in part, to improved methods for determining the presence or absence of copy number variations (e.g., deletions or duplications of chromosomal segments or entire chromosomes). The methods are particularly useful for detecting small deletions or duplications that can be difficult to detect with high specificity and sensitivity using conventional methods due to the limited data available from the relevant chromosomal segments. The methods include improved analytical methods, improved bioassay methods, and combinations of improved analytical and bioassay methods. The methods of the present invention can also be used to detect deletions or duplications that are present in only a small percentage of the cells or nucleic acid molecules tested. This allows for detection of deletions or duplications before the onset of disease (e.g., in a precancerous state) or early in the disease, e.g., before the accumulation of a large number of diseased cells (e.g., cancer cells) with deletions or duplications. More accurate detection of deletions or duplications associated with a disease or disorder allows for improved methods for diagnosing, predicting, preventing, delaying, stabilizing, or treating the disease or disorder. Some deletions or duplications are known to be associated with cancer or severe intellectual or physical disabilities.

[0195] In another aspect, the present invention generally relates, at least in part, to improved methods for detecting single nucleotide variants (SNVs). These improved methods include improved analytical methods, improved bioassay methods, and improved methods using a combination of improved analytical and bioassay methods. In certain exemplary embodiments, the methods are used to detect, diagnose, monitor, or stage cancer in samples (e.g., circulating free DNA samples) in which SNVs are present at very low concentrations (e.g., less than 10%, 5%, 4%, 3%, 2.5%, 2%, 1%, 0.5%, 0.25%, or 0.1% of the total number of normal copies of the SNV locus). That is, in certain exemplary embodiments, these methods are particularly well suited to samples in which a relatively low proportion of mutations or variants exist relative to the normal polymorphic alleles present for the locus. Finally, methods are provided herein that combine improved methods for detecting copy number variation with improved methods for detecting single nucleotide variants.

[0196] Successful treatment of diseases such as cancer depends largely on early diagnosis, correct staging of the disease, selection of an effective treatment regimen, and close monitoring to prevent or detect recurrence. For cancer diagnosis, histological evaluation of tumor material obtained from tissue biopsy is often considered to be the most reliable method.However, due to the invasive nature of biopsy-based sampling, it is not practical for mass screening and regular follow-up.Therefore, this method has the advantage that it can be carried out non-invasively when it is desired to have a relatively low cost and fast turnaround time.The targeted sequencing that can be used by the method of the present invention requires fewer reads than shotgun sequencing (for example, several hundred reads instead of 40 million reads), thereby reducing costs.Multiplex PCR and next-generation sequencing that can be used increase throughput and reduce costs.

[0197] In some exemplary embodiments, analysis of AAI patterns in ctDNA provides more detailed insight into the clonal architecture of tumors, helps predict their therapeutic response, and optimizes treatment strategies.Therefore, in certain embodiments, mmPCR-NGS panels are selected that target clinically causative CNVs and SNVs.In certain exemplary embodiments, such panels are particularly useful for patients with cancers in which CNVs represent a substantial proportion of mutation burden, as is common in breast cancer, ovarian cancer, and lung cancer.

[0198] In some embodiments, this method is used to detect deletions, duplications, or single nucleotide variants in individuals. Samples from individuals containing cells or nucleic acids suspected of having deletions, duplications, or single nucleotide variants may be analyzed. In some embodiments, the sample is derived from tissues or organs suspected of having deletions, duplications, or single nucleotide variants, such as cells or masses suspected of being cancerous. The method of the present invention can be used to detect deletions, duplications, or single nucleotide variants that are present in only one cell or a small number of cells in a mixture containing cells that have deletions, duplications, or single nucleotide variants and cells that do not have deletions, duplications, or single nucleotide variants. In some embodiments, cfDNA or cfRNA from blood samples from individuals is analyzed. In some embodiments, cfDNA or cfRNA is secreted by cells (e.g., cancer cells). In some embodiments, cfDNA or cfRNA is released by the cell undergoing necrosis or apoptosis (for example, cancer cell).The method of the present invention can be used to detect the deletion, duplication or single nucleotide variant that only exists in a small proportion of cfDNA or cfRNA.In some embodiments, one or more cells from embryo are tested.

[0199] In addition to determining the presence or absence of copy number polymorphisms, if desired, one or more other factors can be analyzed.These factors can be used to improve the accuracy of diagnosis (for example, determining the presence or absence of cancer or the increased risk of cancer, classifying cancer, or determining the stage of cancer) or the accuracy of prognosis.These factors can also be used to select a specific therapy or treatment regimen that is likely to be effective in a subject.Exemplary factors include the presence or absence of polymorphisms or mutations, changes (increases or decreases) in the level of total or specific cfDNA, cfRNA, and microRNA (miRNA), changes (increases or decreases) in tumor fraction, changes (increases or decreases) in methylation levels, changes (increases or decreases) in DNA integrity, changes (increases or decreases) or alternative mRNA splicing.

[0200] The following sections describe methods for detecting deletions or duplications using phasing data (e.g., inferred or measured phasing data) or non-phasing data, samples that can be tested, methods for sample preparation, amplification and quantification, methods for phasing genetic data, detectable polymorphisms, mutations, nucleic acid changes, changes in mRNA splicing and changes in nucleic acid levels, databases resulting from the methods, other risk factors and screening methods, cancers that can be diagnosed or treated, cancer treatments, cancer models for testing treatments, and methods for prescribing and administering treatments.

[0201] Exemplary Methods for Determining Ploidy Using Phasing Data Some of the methods of the present invention are based, in part, on the discovery that using phased data to detect CNVs reduces the false negative and false positive rates compared to using non-phased data. This improvement is greatest for samples with low levels of CNVs. Thus, phased data improves the accuracy of CNV detection compared to using non-phased data (e.g., methods that calculate allele ratios at one or more loci or aggregate allele ratios to provide an aggregate value (e.g., average value) across chromosomes or chromosome segments without considering whether the allele ratios at different loci indicate that the same or different haplotypes appear to be present in unusual amounts). The use of phased data allows for a more accurate determination of whether the difference between the measured allele ratio and the predicted allele ratio is due to noise or the presence of a CNV. For example, if the difference between the measured allele ratio and the predicted allele ratio indicates that a sample haplotype is overrepresented at most or all loci in a region, a CNV is likely to be present. By using the association between alleles in haplotypes, it is possible to determine whether the measured genetic data corresponds to the same haplotype that is overrepresented (rather than random noise). In contrast, if the difference between the measured allele ratio and the predicted allele ratio is due only to noise (e.g., experimental error), in some embodiments, about half of the time a first haplotype will appear overrepresented and about half of the time a second haplotype will appear overrepresented.

[0202] In some embodiments, the phasing genetic data is used to determine whether there is an overrepresentation of the copy number of a first homologous chromosomal segment compared to a second homologous chromosomal segment in the genome of an individual (e.g., in the genome of one or more cells, or in cfDNA or cfRNA). Exemplary overrepresentation includes duplication of the first homologous chromosomal segment or deletion of the second homologous chromosomal segment. In some embodiments, there is no overrepresentation because the first chromosomal segment and the homologous chromosomal segment are present in equal proportions (e.g., one copy of each segment in a diploid sample). In some embodiments, the calculated allele ratio in the nucleic acid sample is compared to the predicted allele ratio to determine whether there is an overrepresentation, as further described below. As used herein, the phrase "first homologous chromosomal segment compared to a second homologous chromosomal segment" refers to the first homolog of the chromosomal segment and the second homolog of the chromosomal segment.

[0203] In some embodiments, the method includes: obtaining phased genetic data for a first homologous chromosome segment, the phased genetic data including, for each locus in a set of polymorphic loci on a first homologous chromosome segment, the identity of the allele present at the locus on the first homologous chromosome segment; obtaining phased genetic data for a second homologous chromosome segment, the phased genetic data including, for each locus in a set of polymorphic loci on a second homologous chromosome segment, the identity of the allele present at the locus on the second homologous chromosome segment; and obtaining measured genetic allele data for each allele at each locus in the set of polymorphic loci, the amount of each allele present in DNA or RNA samples from one or more target cells and one or more non-target cells from the individual. In some embodiments, the method includes: enumerating a set of one or more hypotheses indicating the degree of overrepresentation of the first homologous chromosomal segment; for each of the hypotheses, calculating predicted genetic data for a plurality of loci in the sample from phased genetic data obtained for one or more possible ratios of DNA or RNA from one or more target cells to total DNA or RNA in the sample; for each possible DNA or RNA ratio and for each hypothesis, calculating (e.g., by computer) a data fitting between the obtained genetic data of the sample and the predicted genetic data of the sample for that possible DNA or RNA ratio and for that hypothesis; ranking the one or more hypotheses according to the data fitting; and selecting the highest ranked hypothesis to determine the degree of overrepresentation of the copy number of the first homologous chromosomal segment in the genome of one or more cells from the individual.

[0204] In some embodiments, the method involves obtaining phasing genetic data using any of the methods described herein or any known method.In some embodiments, the method involves simultaneously or sequentially in any order: (i) obtaining phasing genetic data for a first homologous chromosome segment, the phasing genetic data for each locus in a set of polymorphic loci on a first homologous chromosome segment, the phasing genetic data for the first homologous chromosome segment includes the identity of the allele present at the locus on the first homologous chromosome segment; (ii) obtaining phasing genetic data for a second homologous chromosome segment, the phasing genetic data for each locus in a set of polymorphic loci on a second homologous chromosome segment, the phasing genetic data for the second homologous chromosome segment, the phasing genetic data for each locus in a set of polymorphic loci on a second homologous chromosome segment includes the identity of the allele present at the locus on the second homologous chromosome segment; and (iii) obtaining measured gene allele data, the phasing genetic data for each locus in a set of polymorphic loci in a sample of DNA from one or more cells from an individual, the measured gene allele data including the amount of each allele for each locus in the set of polymorphic loci.

[0205] In some embodiments, the method involves calculating the allele ratio for one or more loci in a set of polymorphic loci that are heterozygous in at least one cell from which the sample is derived. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one allele divided by the total measured amount of all alleles for that locus. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one allele (e.g., an allele on a first homologous chromosome segment) divided by the measured amounts of one or more other alleles for that locus (e.g., an allele on a second homologous chromosome segment). The calculated allele ratio may be calculated using any of the methods described herein or any standard method (e.g., any mathematical transformation of the calculated allele ratios described herein).

[0206] In some embodiments, the method involves determining whether there is an overrepresentation of a copy number of a first homologous chromosomal segment by comparing the calculated allele ratio of one or more alleles for a locus with the expected allele ratio for that locus when the first and second homologous chromosomal segments are present in equal proportions. In some embodiments, the expected allele ratio assumes that the possible alleles for a locus are equally likely to be present. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one allele divided by the total measured amount of all alleles for that locus, and the corresponding expected allele ratio is 0.5 for a biallelic locus or 1 / 3 for a triallelic locus. In some embodiments, the expected allele ratio is the same for all loci, e.g., 0.5 for all loci. In some embodiments, the expected allele ratio assumes that the likelihood of the possible alleles for a locus being present may differ, for example, based on the frequency of each allele in a particular population to which the subject belongs (e.g., a population based on the subject's ancestry). Such allele frequencies are publicly available (see, e.g., the HapMap Project; the Perlegen Human Haplotype Project; online at ncbi.nlm.nih.gov / projects / SNP / ; Sherry ST, Ward MH, Kholodov M, et al., dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001 Jan. 1;29(1):308-11, each of which is incorporated herein by reference in its entirety). In some embodiments, the predicted allele ratio is the allele ratio expected for a particular individual tested for a particular hypothesis indicating the degree of overrepresentation of a first homologous chromosomal segment.For example, the predicted allele ratio for a particular individual may be determined based on phased or non-phased genetic data from that individual (e.g., samples from that individual that are unlikely to have deletions or duplications, such as non-cancerous samples), or data from one or more blood relatives of that individual.

[0207] In some embodiments, a calculated allele ratio is indicative of overrepresentation of a copy number of a first homologous chromosomal segment if either (i) the allele ratio for the measured amount of an allele present at a locus on a first homologous chromosomal segment divided by the total measured amount of all alleles at that locus is greater than the predicted allele ratio for that locus, or (ii) the allele ratio for the measured amount of an allele present at a locus on a second homologous chromosomal segment divided by the total measured amount of all alleles at that locus is less than the predicted allele ratio for that locus. In some embodiments, a calculated allele ratio is considered to be indicative of overrepresentation only if it is significantly greater or less than the predicted ratio for that locus. In some embodiments, a calculated allele ratio indicates that there is no overrepresentation of copies of a first homologous chromosomal segment if either (i) the allele ratio for the measured amount of an allele present at a locus on a first homologous chromosomal segment divided by the total measured amount of all alleles at that locus is less than or equal to the predicted allele ratio for that locus, or (ii) the allele ratio for the measured amount of an allele present at a locus on a second homologous chromosomal segment divided by the total measured amount of all alleles at that locus is greater than or equal to the predicted allele ratio for that locus. In some embodiments, calculated ratios that are equal to the predicted corresponding ratios are ignored (because they are an indication of no overrepresentation).

[0208] In various embodiments, one or more of the calculated allele ratios are compared to the corresponding predicted allele ratios using one or more of the following methods: In some embodiments, it is determined whether the calculated allele ratio is above or below the predicted allele ratio for a particular locus, regardless of the magnitude of the difference; In some embodiments, it is determined whether the calculated allele ratio is above or below the predicted allele ratio, regardless of the magnitude of the difference; In some embodiments, it is determined whether the calculated allele ratio is above or below the predicted allele ratio, and the magnitude of the difference for a particular locus; In some embodiments, it is determined whether the average or weighted average of the calculated allele ratios is above or below the average or weighted average of the predicted allele ratios, regardless of the magnitude of the difference. In some embodiments, the magnitude of the difference between the average or weighted average calculated allele ratios and the average or weighted average predicted allele ratios is determined, regardless of whether the average or weighted average calculated allele ratios is above or below the average or weighted average predicted allele ratios. In some embodiments, the magnitude of the difference between the average or weighted average calculated allele ratios and the average or weighted average predicted allele ratios is determined. In some embodiments, the magnitude of the difference between the calculated allele ratios and the predicted allele ratios is determined.

[0209] In some embodiments, the magnitude of the difference between the calculated allele ratio and the predicted allele ratio for one or more loci is used to determine whether an overrepresentation of the copy number of a first homologous chromosomal segment is due to a duplication of the first homologous chromosomal segment or a deletion of a second homologous chromosomal segment in the genome of one or more cells.

[0210] In some embodiments, copy number overrepresentation of the first homologous chromosomal segment is determined to be present if one or more of the following conditions are met: In some embodiments, the numerical value of the calculated allele ratio indicative of copy number overrepresentation of the first homologous chromosomal segment is above a threshold; In some embodiments, the numerical value of the calculated allele ratio indicative of no copy number overrepresentation of the first homologous chromosomal segment is below a threshold; In some embodiments, the magnitude of the difference between the calculated allele ratio indicative of copy number overrepresentation of the first homologous chromosomal segment and the corresponding predicted allele ratio is above a threshold; In some embodiments, the sum of the magnitudes of the differences between the calculated allele ratio and the corresponding predicted allele ratio for all calculated allele ratios indicative of overrepresentation is above a threshold; In some embodiments, the magnitude of the difference between the calculated allele ratio indicative of no copy number overrepresentation of the first homologous chromosomal segment and the corresponding predicted allele ratio is below a threshold. In some embodiments, the average or weighted average of the calculated allele ratios for the measured amounts of alleles present on the first homologous chromosomal segment, divided by the total measured amounts of all alleles for that locus, is greater than the average or weighted average of the predicted allele ratios by at least one threshold. In some embodiments, the average or weighted average of the calculated allele ratios for the measured amounts of alleles present on the second homologous chromosomal segment, divided by the total measured amounts of all alleles for that locus, is less than the average or weighted average of the predicted allele ratios by at least one threshold. In some embodiments, the data fit between the calculated allele ratios and the predicted allele ratios for the overrepresentation of copy numbers of the first homologous chromosomal segment is below a threshold (indicating good data fit).In some embodiments, the data fit between the calculated allele ratio and the allele ratio expected for the absence of copy number overrepresentation of the first homologous chromosomal segment is above a threshold (indicative of poor data fit).

[0211] In some embodiments, copy number overrepresentation of the first homologous chromosomal segment is determined to be absent if one or more of the following conditions are met: In some embodiments, the numerical value of the calculated allele ratio indicative of copy number overrepresentation of the first homologous chromosomal segment is below a threshold; In some embodiments, the numerical value of the calculated allele ratio indicative of copy number overrepresentation of the first homologous chromosomal segment is above a threshold; In some embodiments, the magnitude of the difference between the calculated allele ratio indicative of copy number overrepresentation of the first homologous chromosomal segment and the corresponding predicted allele ratio is below a threshold; In some embodiments, the magnitude of the difference between the calculated allele ratio indicative of copy number overrepresentation of the first homologous chromosomal segment and the corresponding predicted allele ratio is above a threshold. In some embodiments, the average or weighted average of the calculated allele ratios for the measured amounts of alleles present on the first homologous chromosomal segment, divided by the total measured amounts of all alleles for that locus, minus the average or weighted average of the predicted allele ratios, is below a threshold. In some embodiments, the average or weighted average of the predicted allele ratios, subtracted by the average or weighted average of the calculated allele ratios for the measured amounts of alleles present on the second homologous chromosomal segment, divided by the total measured amounts of all alleles for that locus, is below a threshold. In some embodiments, the data fit between the calculated allele ratios and the predicted allele ratios for the overrepresentation of copies of the first homologous chromosomal segment exceeds the threshold. In some embodiments, the data fit between the calculated allele ratios and the predicted allele ratios for the absence of overrepresentation of copies of the first homologous chromosomal segment is below the threshold. In some embodiments, the threshold is determined from empirical testing of samples known to have the CNV of interest and / or samples known to lack the CNV.

[0212] In some embodiments, determining whether there is an overrepresentation of the copy number of the first homologous chromosomal segment includes listing a set of one or more hypotheses indicating the degree of overrepresentation of the first homologous chromosomal segment. In an exemplary hypothesis, there is no overrepresentation because the chromosomal segments homologous to the first chromosomal segment are present in equal proportions (e.g., one copy of each segment in a diploid sample). Another exemplary hypothesis includes the first homologous chromosomal segment being replicated one or more times (e.g., one, two, three, four, five, or more excess copies of the first homologous chromosomal segment compared to the copy number of the second homologous chromosomal segment). Another exemplary hypothesis includes a deletion of the second homologous chromosomal segment. Yet another exemplary hypothesis is a deletion of both the first and second homologous chromosomal segments. In some embodiments, the predicted allele ratio for a locus that is heterozygous in at least one cell is estimated for each hypothesis, taking into account the degree of overrepresentation indicated by that hypothesis. In some embodiments, the likelihood that the hypothesis is correct is calculated by comparing the calculated allele ratios with the predicted allele ratios, and the hypothesis with the greatest likelihood is selected.

[0213] In some embodiments, an expected distribution of a test statistic is calculated using the predicted values ​​of allele ratios for each hypothesis, hi some embodiments, the likelihood that the hypothesis is correct is calculated by comparing the test statistic calculated using the calculated values ​​of allele ratios with the expected distribution of the test statistic calculated using the predicted values ​​of allele ratios, and the hypothesis with the greatest likelihood is selected.

[0214] In some embodiments, the predicted allele ratio for a locus that is heterozygous in at least one cell is estimated taking into account the phasing genetic data for the first homologous chromosome segment, the phasing genetic data for the second homologous chromosome segment, and the degree of overrepresentation indicated by the hypothesis. In some embodiments, the likelihood that the hypothesis is correct is calculated by comparing the calculated allele ratio with the predicted allele ratio, and the hypothesis with the greatest likelihood is selected.

[0215] Use of mixed samples It will be understood that for many embodiments, the sample is a mixed sample containing DNA or RNA from one or more target cells and one or more non-target cells. In some embodiments, the target cells are cells with a CNV (e.g., a deletion or duplication of interest), and the non-target cells are cells without the copy number variation of interest (e.g., a mixture of cells with the deletion or duplication of interest and cells without any of the deletions or duplications being tested). In some embodiments, the target cells are cells associated with a disease or disorder or an elevated risk of a disease or disorder (e.g., cancer cells), and the non-target cells are cells not associated with a disease or disorder or an elevated risk of a disease or disorder (e.g., non-cancerous cells). In some embodiments, the target cells all have the same CNV. In some embodiments, two or more target cells have different CNVs.In some embodiments, one or more of target cells have the CNV, polymorphism or mutation associated with the disease or disorder or the increased risk of disease or disorder, which is not found in at least one other target cell.In some such embodiments, it is assumed that the proportion of cells associated with the disease or disorder or the increased risk of disease or disorder in the total cells from sample is greater than or equal to the most frequent proportion of these CNVs, polymorphisms or mutations in this sample.For example, if 6% of cells have K-ras mutation and 8% of cells have BRAF mutation, it is assumed that at least 8% of cells are cancerous.

[0216] In some embodiments, the ratio of DNA (or RNA) from one or more target cells to the total DNA (or RNA) in the sample is calculated. In some embodiments, a set of one or more hypotheses is enumerated, which indicates the degree of overrepresentation of the first homologous chromosome segment. In some embodiments, the predicted allele ratio for the heterozygous locus in at least one cell is estimated by taking into account the calculated DNA or RNA ratio, and the degree of overrepresentation indicated by the hypothesis is estimated for each hypothesis. In some embodiments, the likelihood that the hypothesis is correct is calculated by comparing the calculated allele ratio with the predicted allele ratio, and the hypothesis with the highest likelihood is selected.

[0217] In some embodiments, the predictive distribution of the test statistic calculated using the predicted allele ratios and the calculated DNA or RNA ratios is estimated for each hypothesis. In some embodiments, the likelihood that the hypothesis is correct is determined by comparing the test statistic calculated using the calculated allele ratios and the calculated DNA or RNA ratios with the predictive distribution of the test statistic calculated using the predicted allele ratios and the calculated DNA or RNA ratios, and the hypothesis with the greatest likelihood is selected.

[0218] In some embodiments, the method includes enumerating a set of one or more hypotheses that indicate the degree of overrepresentation of the first homologous chromosome segment. In some embodiments, the method includes, for each hypothesis, estimating either (i) a predicted allele ratio for the locus that is heterozygous in at least one cell, taking into account the degree of overrepresentation indicated by that hypothesis, or (ii) a predicted distribution of a test statistic calculated using the predicted allele ratio and the possible ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample for one or more possible ratios of DNA or RNA. In some embodiments, data fitting is calculated by comparing (i) the calculated allele ratio with the predicted allele ratio, or (ii) the calculated allele ratio with the possible ratio of DNA or RNA with the predicted distribution of a test statistic calculated using the predicted allele ratio and the possible ratio of DNA or RNA. In some embodiments, one or more of the hypotheses are ranked according to data fitting, and the highest-ranked hypothesis is selected. In some embodiments, a technique or algorithm, such as a search algorithm, is used to calculate a data fitting, rank hypotheses, or select the highest-ranked hypothesis. In some embodiments, the data fitting is a fitting to a beta-binomial distribution or a fitting to a binomial distribution. In some embodiments, the technique or algorithm is selected from the group consisting of maximum likelihood estimation, empirical maximum estimation, Bayesian estimation, dynamic estimation (e.g., dynamic Bayesian estimation), and expectation maximization estimation. In some embodiments, the method includes applying the above-mentioned technique or algorithm to the obtained genetic data and the predicted value of the genetic data.

[0219] In some embodiments, the method includes creating a distribution of possible ratios of DNA or RNA from one or more target cells to the total DNA or RNA in the sample, ranging from a lower limit to an upper limit. In some embodiments, a set of one or more hypotheses is enumerated, each of which indicates the degree of overrepresentation of a first homologous chromosome segment. In some embodiments, the method includes, for each possible DNA or RNA ratio in the distribution and for each hypothesis, estimating either (i) a predicted allele ratio for a locus that is heterozygous in at least one cell, taking into account the possible DNA or RNA ratios and the degree of overrepresentation indicated by the hypothesis, or (ii) a predictive distribution of test probabilities calculated using the predicted allele ratios and the possible DNA or RNA ratios. In some embodiments, the method calculates the likelihood that a hypothesis is correct by comparing (i) the calculated allele ratio with the predicted allele ratio, or (ii) the calculated allele ratio and a test statistic calculated using the possible DNA or RNA ratios with the predicted allele ratio and a test statistic calculated using the possible DNA or RNA ratios, for each possible ratio of DNA or RNA, with the predicted distribution of the test statistic. In some embodiments, the joint probability for each hypothesis is determined by adding up the probabilities of the hypotheses for each possible ratio in the distribution, and the hypothesis with the greatest joint probability is selected. In some embodiments, the joint probability for each hypothesis is determined by weighting the probability of a hypothesis based on the likelihood that the possible ratio is the correct ratio for a particular possible ratio.

[0220] In some embodiments, the ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample is estimated using a technique selected from the group consisting of maximum likelihood estimation, empirical maximum estimation, Bayesian estimation, dynamic estimation (for example, dynamic Bayesian estimation) and expectation maximization estimation.In some embodiments, the ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample is assumed to be the same for two or more (or all) of the CNVs of interest.In some embodiments, for each CNV of interest, the ratio of DNA or RNA from one or more target cells to the total DNA or RNA in the sample is calculated.

[0221] Exemplary Method for Using Incomplete Fading Data It should be understood that in many embodiments, incomplete phasing data is used. For example, it may not be known with 100% certainty which alleles are present for one or more of the loci on the first and / or second homologous chromosome segments. In some embodiments, prior probabilities for the individual's possible haplotypes (e.g., haplotypes based on population-based haplotype frequencies) are used in calculating the probability of each hypothesis. In some embodiments, the prior probabilities for the possible haplotypes are adjusted by using another method for phasing genetic data or by using phasing data from other subjects (e.g., previous subjects) to refine the population data used for the individual's informatics-based phasing.

[0222] In some embodiments, the phasing genetic data includes probability data for two or more possible sets of phasing genetic data, each possible set of phasing data including the possible identity of an allele present at each locus in a set of polymorphic loci on a first homologous chromosome segment and the possible identity of an allele present at each locus in a set of polymorphic loci on a second homologous chromosome segment. In some embodiments, the probability of at least one of the hypotheses is determined for each possible set of phasing genetic data. In some embodiments, the joint probability for a hypothesis is determined by adding up the probabilities of that hypothesis for each possible set of phasing genetic data, and the hypothesis with the greatest joint probability is selected.

[0223] Any of the methods disclosed herein or any known method may be used to generate incomplete phasing data for use in the claimed methods (e.g., using ensemble-based haplotype frequencies to infer the most likely phase). In some embodiments, the phasing data is obtained by probabilistically combining haplotypes of smaller segments. For example, possible haplotypes may be determined based on the possible combinations of one haplotype from a first region with another haplotype from another region on the same chromosome. The probability that particular haplotypes from different regions are part of the same, larger haplotype block on the same chromosome may be determined, for example, using ensemble-based haplotype frequencies and / or known recombination rates between different regions.

[0224] In some embodiments, a single-hypothesis rejection test is used for the null hypothesis of disomy. In some embodiments, the probability of the disomy hypothesis is calculated, and the disomy hypothesis is rejected if the probability is below a given threshold (e.g., less than 1 in 1,000). If the null hypothesis is rejected, this may be due to an error in incomplete phasing data or the presence of CNV. In some embodiments, more accurate phasing data is obtained (e.g., phasing data from any of the molecular phasing methods disclosed herein for obtaining actual phasing data, rather than phasing data estimated based on bioinformatics). In some embodiments, the probability of the disomy hypothesis is recalculated using this more accurate phasing data to determine whether the disomy hypothesis should still be rejected. Rejection of this hypothesis indicates the presence of duplication or deletion of a chromosomal segment. If desired, the false positive rate can be changed by adjusting the threshold.

[0225] Further Exemplary Embodiments for Determining Ploidy Using Phasing Data In an exemplary embodiment, a method for determining the ploidy of a chromosome segment in a sample from an individual is provided herein. The method includes receiving allele frequency data at each locus in a set of polymorphic loci on the chromosome segment, the allele frequency data including the amount of each allele present in the sample, generating phased allele information for the set of polymorphic loci by estimating the phase of the allele frequency data, using the allele frequency data to generate individual probabilities of allele frequencies for the polymorphic loci for different ploidy states, using the individual probabilities and the phased allele information to generate joint probabilities for the set of polymorphic loci, and determining the ploidy of the chromosome segment by selecting a best-fitting model that is indicative of chromosome ploidy based on the joint probabilities.

[0226] As disclosed herein, allele frequency data (also referred to herein as measured gene allele data) may be generated by methods known in the art. For example, the data may be generated using qPCR or microarrays. In an exemplary embodiment, the data is generated using nucleic acid sequence data, particularly high-throughput nucleic acid sequence data.

[0227] In certain illustrative examples, the allele frequency data is corrected for errors before being used to generate individual probabilities. In specific illustrative embodiments, the errors corrected include allele amplification efficiency bias. In other embodiments, the errors corrected include ambient contamination and genotypic contamination. In some embodiments, the errors corrected include allele amplification bias, sequencing errors, ambient contamination, and genotypic contamination.

[0228] In certain embodiments, the individual probabilities are generated using a set of models of different ploidy states and allelic imbalance fractions for the set of polymorphic loci. In these and other embodiments, the linkage probabilities are generated by considering linkages between polymorphic loci on chromosomal segments.

[0229] Thus, in one exemplary embodiment combining several of these embodiments, provided herein is a method for detecting chromosomal ploidy in a sample from an individual, the method including: receiving nucleic acid sequence data for alleles at a set of polymorphic loci on a chromosomal segment in the individual; using the nucleic acid sequence data to detect allele frequencies at the set of loci; correcting for allele amplification efficiency bias in the detected allele frequencies to create corrected allele frequencies for the set of polymorphic loci; creating phasing allele information for the set of polymorphic loci by estimating the phase of the nucleic acid sequence data; creating individual probabilities of allele frequencies for the polymorphic loci for different ploidy states by comparing the corrected allele frequencies to a set of models of different ploidy states and allele imbalance fractions for the set of polymorphic loci; creating a joint probability for the set of polymorphic loci by combining the individual probabilities that take into account the joint probabilities between the polymorphic loci on the chromosomal segment; and selecting a best-fitting model that is indicative of chromosomal aneuploidy based on the joint probabilities.

[0230] As disclosed herein, individual probabilities may be generated using a set of models or hypotheses of different ploidy states and average allelic imbalance fractions for a set of polymorphic loci. For example, in a particularly illustrative example, individual probabilities are generated by modeling the ploidy states of a first homolog of a chromosomal segment and a second homolog of the chromosomal segment. The modeled ploidy states include: (1) none of the cells have a deletion or amplification of the first homolog or the second homolog of the chromosomal segment; (2) at least some of the cells have a deletion of the first homolog or an amplification of the second homolog of the chromosomal segment; and (3) at least some of the cells have a deletion of the second homolog or an amplification of the first homolog of the chromosomal segment.

[0231] It will be understood that the above model may also be referred to as the hypotheses used to constrain the model. Thus, shown above are three hypotheses that can be used.

[0232] The modeled average allelic imbalance fraction may include any range of average allelic imbalance, including the actual average allelic imbalance of the chromosomal segment.For example, in certain exemplary embodiments, the modeled average allelic imbalance range may be 0, 0.1, 0.2, 0.25, 0.3, 0.4, 0.5, 0.6, 0.75, 1, 2, 2.5, 3, 4, and 5% at the lower end, and 1, 2, 2.5, 3, 4, 5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 95, and 99% at the upper end.The interval for modeling with this range may be any interval depending on the computing power used and the time allowed for analysis.For example, intervals of 0.01, 0.05, 0.02, or 0.1 may be modeled.

[0233] In certain exemplary embodiments, the sample has an average allelic imbalance for the chromosomal segment of 0.4% to 5%. In certain embodiments, the average allelic imbalance is low. In these embodiments, the average allelic imbalance is typically less than 10%. In certain exemplary embodiments, the allelic imbalance has a lower limit of 0.25, 0.3, 0.4, 0.5, 0.6, 0.75, 1, 2, 2.5, 3, 4, and 5% and an upper limit of 1, 2, 2.5, 3, 4, and 5%. In other exemplary embodiments, the average allelic imbalance has a lower limit of 0.4, 0.45, 0.5, 0.6, 0.7, 0.8, 0.9, or 1.0% and an upper limit of 0.5, 0.6, 0.7, 0.8, 0.9, 1.0, 1.5, 2.0, 3.0, 4.0, or 5.0%. For example, the average allelic imbalance of the samples is, in an illustrative example, 0.45-2.5%. In another example, the average allelic imbalance is detected with a sensitivity of 0.45, 0.5, 0.6, 0.8, 0.9, or 1.0%. That is, the test method can detect chromosomal aneuploidies with an AAI down to 0.45, 0.5, 0.6, 0.8, 0.8, 0.9, or 1.0%. Exemplary samples with low allelic imbalance for the methods of the invention include plasma samples from individuals with cancer who have circulating tumor DNA or plasma samples from pregnant women who have circulating fetal DNA.

[0234] It will be understood that for SNV, the proportion of abnormal DNA is typically measured using mutant allele frequency (the number of mutant alleles at a locus / the total number of alleles at that locus).Because the difference in the amount of two homologs in tumors is similar, the average allelic imbalance (AAI) measures the proportion of abnormal DNA for CNV (defined as |(H1-H2)| / (H1+H2)), where Hi is the average copy number of homolog i in the sample, and Hi / (H1+H2) is the abundance fraction of homolog i, i.e., the homolog ratio.The maximum homolog ratio is the homolog ratio of the more abundant homolog.

[0235] The assay dropout rate is the proportion of SNPs with no reads, estimated using all SNPs. The single allele dropout (ADO) rate is the proportion of SNPs with only one allele, estimated using only heterozygous SNPs. Genotype confidence is determined by fitting a binomial distribution to the number of reads at each SNP that were B allele reads using the ploidy state of the SNP's focal region, allowing the probability of each genotype to be estimated.

[0236] For tumor tissue samples, chromosomal aneuploidy (exemplified in this paragraph by CNV) can be represented by transitions between allele frequency distributions. In plasma samples from cancer patients, individuals suspected of having cancer, individuals previously diagnosed with cancer, or as cancer screening for at-risk individuals or the general population, CNVs can be identified by a maximum likelihood algorithm that searches for plasma CNVs in regions known to exhibit aneuploidy in cancer and / or when tumor samples from the same individuals also have CNVs. In an exemplary embodiment, this algorithm uses the haplotype phase information of the individual whose sample is analyzed for the presence of circulating tumor DNA to fit the measured and corrected allele counts of the test sample to predicted allele counts, for example, using a joint distribution model. Such haplotype phase information can be deduced from any sample from an individual, such as, but not limited to, a buffy coat sample, saliva sample, or skin sample, that contains most, or at least 60, 70, 80, 90, 95, 96, 97, 98, 99%, or all, normal cellular DNA, from parental genetic information, or by de novo haplotype phasing, using a variety of methods (e.g., Snyder, M. et al., Haplotype-resolved genome sequencing: experimental methods and applications. Nat Rev Genet 16, 344-358 (2015)), such as haplotyping by dilution (Kaper, F. et al., Whole-genome haplotyping by dilution, amplification, and sequencing. Proc Natl Acad Sci USA 110, 5552-5557 (2013)) or long-read sequencing (Kuleshov, V. et al., Whole-genome haplotyping using long This can be achieved by using reads and statistical methods. Nat Biotech 32, 261-266 (2014)).The algorithm can model predicted allele frequencies across all allelic imbalance rates at 0.025% intervals for three sets of hypotheses: (1) all cells are normal (no allelic imbalance); (2) some / all cells have a deletion of homolog 1 or an amplification of homolog 2; or (3) some / all cells have a deletion of homolog 2 or an amplification of homolog 1. The likelihood of each hypothesis can be determined at each SNP using a Bayesian classifier based on a beta-binomial model of predicted and observed allele frequencies at all heterozygous SNPs. The joint likelihood across multiple SNPs can then be calculated, in certain exemplary embodiments, taking into account the joint SNP locus, as exemplified herein. Indeed, in exemplary embodiments, haplotype phase information from normal cells, obtained as disclosed above, is used by the algorithm to fit measured, typically corrected, test sample allele counts to predicted allele counts using a joint distribution model. The maximum likelihood hypothesis can then be selected.

[0237] Considering a chromosomal region with an average of N copies in the tumor, where c denotes the fraction of DNA in plasma that comes from a mixture of normal and tumor cells in the disomic region, the AAI is calculated as follows:

number

[0238] In certain illustrative examples, the allele frequency data is corrected for errors before being used to generate individual probabilities. Correction of different types of errors and / or biases is disclosed herein. In specific illustrative embodiments, the error corrected is allele amplification efficiency bias. In other embodiments, the error corrected includes sequencing errors, ambient contamination, and genotype contamination. In some embodiments, the errors corrected include allele amplification bias, sequencing errors, ambient contamination, and genotype contamination.

[0239] It will be appreciated that allele amplification efficiency bias can be determined for a given allele as part of an experimental or laboratory determination involving the sample under test, or can be determined at a different time using a set of samples that include the allele for which the efficiency is being calculated. Ambient contamination and genotypic contamination are typically determined in the same run as the sample under test analysis.

[0240] In certain embodiments, ambient contamination and genotype contamination are determined for homozygous alleles in the sample.For any given sample from an individual, even if a locus is selected for analysis because it has a relatively high heterozygosity in the population, it will be understood that some loci in the sample will be heterozygous and others will be homozygous.In some embodiments, it is advantageous to determine the ploidy of a chromosome segment using heterozygous loci for an individual, while ambient contamination and genotype contamination can be calculated using homozygous loci.

[0241] In a particular illustrative example, the above selection is made by analyzing the magnitude of the difference between the phasing allele information and the estimated allele frequencies generated for the model.

[0242] In an illustrative example, individual probabilities of allele frequencies are generated based on a beta-binomial model of the predicted and observed allele frequencies at a set of polymorphic loci. In an illustrative example, the individual probabilities are generated using a Bayesian classifier.

[0243] In certain exemplary embodiments, nucleic acid sequence data is generated by performing high-throughput DNA sequencing of multiple copies of a series of amplicons generated using a multiplex amplification reaction, each amplicon of the series spanning at least one polymorphic locus of a set of polymorphic loci, and each of the polymorphic loci of the set is amplified. In certain embodiments, the multiplex amplification reaction is performed under limited primer conditions for at least half of the reactions. In some embodiments, limited primer concentrations are used in 1 / 10, 1 / 5, 1 / 4, 1 / 3, 1 / 2, or all of the reactions in the multiplex reaction. Provided herein are factors to consider in achieving limited primer conditions in amplification reactions such as PCR.

[0244] In certain embodiments, the methods provided herein detect the ploidy of multiple chromosome segments across multiple chromosomes. Thus, in these embodiments, the chromosome ploidy is determined for a set of chromosome segments in a sample. For these embodiments, a larger number of multiplex amplification reactions are required. Thus, for these embodiments, the multiplex amplification reaction may include, for example, 2,500 to 50,000 multiplex reactions. In certain embodiments, multiplex reactions are performed in the following ranges: from 100, 200, 250, 500, 1000, 2500, 5000, 10,000, 20,000, 25,000, and 50,000 at the lower end to 200, 250, 500, 1000, 2500, 5000, 10,000, 20,000, 25,000, 50,000, and 100,000 at the upper end.

[0245] In exemplary embodiments, the set of polymorphic loci is a set of loci known to show high heterozygosity.However, for any given individual, it is expected that some of these loci will be homozygous.In certain exemplary embodiments, the method of the present invention utilizes nucleic acid sequence information for both the homozygous loci and heterozygous loci of an individual.The homozygous loci of an individual are used, for example, for error correction, while the heterozygous loci are used to determine the allele imbalance of the sample.In certain embodiments, at least 10% of the polymorphic loci are heterozygous loci of an individual.

[0246] As disclosed herein, it is preferred to analyze target SNP loci known to be heterozygous in the population. Thus, in certain embodiments, polymorphic loci are selected in which 10, 20, 25, 50, 75, 80, 90, 95, 99, or 100% of the polymorphic loci are known to be heterozygous in the population.

[0247] As disclosed herein, in certain embodiments, the sample is a plasma sample from a pregnant woman.

[0248] In some examples, the method further includes performing the method on a control sample with a known average allelic imbalance ratio. The control may have an average allelic imbalance ratio for a particular allelic state that is indicative of aneuploidy of a chromosomal segment of 0.4-10% to mimic the average allelic imbalance of alleles in a sample present at low concentrations, such as would be expected for circulating free DNA from a tumor.

[0249] In some embodiments, as disclosed herein, PlasmArt control is used as control.Therefore, in certain embodiments, this is the sample that comprises fragmenting the nucleic acid sample that is known to show chromosomal aneuploidy into fragments that mimic the size of the fragments of DNA that circulate in individual's plasma, and is prepared by a method.In certain embodiments, the control that does not have aneuploidy for chromosomal segment is used.

[0250] In exemplary embodiments, data from one or more controls may be analyzed in this method along with the test sample. The controls may include, for example, different samples from individuals not suspected of having chromosomal aneuploidy, or samples suspected of having CNV or chromosomal aneuploidy. For example, if the test sample is a tumor sample suspected of containing circulating free tumor DNA, the method may be performed on the plasma sample as well as a tumor-derived control sample from the subject. As disclosed herein, the control sample may be prepared by fragmenting a DNA sample known to exhibit chromosomal aneuploidy. Such fragmentation can obtain a DNA sample that mimics the DNA composition of apoptotic cells, especially when the sample is from an individual suffering from cancer. Data from the control sample will increase the reliability of the detection of chromosomal aneuploidy.

[0251] In certain embodiments of the methods of determining ploidy, the sample is a plasma sample from an individual suspected of having cancer. In these embodiments, the method further comprises determining whether copy number variations are present in tumor cells of the individual based on the selecting step described above. For these embodiments, the sample may be a plasma sample from the individual. In these embodiments, the method may further comprise determining whether cancer is present in the individual based on the selecting step described above.

[0252] These embodiments for determining the ploidy of a chromosomal segment may further include detecting a single nucleotide variant at a single nucleotide variant position in the set of single nucleotide variant positions, wherein detecting either the chromosomal aneuploidy or the single nucleotide variant, or both, indicates the presence of circulating tumor nucleic acid in the sample.

[0253] These embodiments may further include receiving haplotype information of chromosomal segments for a tumor of an individual and using the haplotype information to generate a set of models of different ploidy states and allelic imbalance fractions for the set of polymorphic loci.

[0254] As disclosed herein, certain embodiments of the methods for determining ploidy may further include removing outliers from the initial or revised allele frequency data before comparing the initial or revised allele frequencies to the set of models. For example, in certain embodiments, locus allele frequencies that are at least two or three standard deviations above or below the mean value for other loci on the chromosome segment are removed from the data before being used for modeling.

[0255] As noted herein, it will be understood that for many of the embodiments provided herein, including those for determining the ploidy of a chromosome segment, incomplete or complete phasing data is preferably used. It will also be understood that several features are provided herein that provide improvements over conventional methods for detecting ploidy, and that many different combinations of these features may be used.

[0256] In certain embodiments, a computer system and a computer-readable medium are provided herein for carrying out any of the methods of the present invention. These include a system and a computer-readable medium for carrying out a method for determining ploidy. Thus, as a non-limiting example of a system embodiment, to demonstrate that any of the methods provided herein can be implemented using the disclosures of the present specification and a system and a computer-readable medium, in another aspect, a system for detecting chromosomal ploidy in a sample from an individual is provided herein, the system comprising: an input processor configured to receive allele frequency data including the amount of each allele present in the sample at each locus in a set of polymorphic loci on a chromosomal segment; a modeler configured to generate phased allele information for the set of polymorphic loci by estimating the phase of the allele frequency data, use the allele frequency data to generate individual probabilities of allele frequencies for the polymorphic loci for different ploidy states, and use the individual probabilities and the phased allele information to generate a joint probability for the set of polymorphic loci; and a hypothesis manager that determines the ploidy of the chromosomal segment by selecting a best-fit model that is an indicator of chromosomal ploidy based on the joint probabilities.

[0257] In certain embodiments of this system, the allele frequency data is data generated by a nucleic acid sequencing system. In certain embodiments, the system further comprises an error correction unit configured to correct errors in the allele frequency data, and the corrected allele frequency data is used by the modeler to generate the individual probabilities. In certain embodiments, the error correction unit corrects for allele amplification efficiency bias. In certain embodiments, the modeler generates the individual probabilities using a set of models of different ploidy states and allele imbalance fractions for the set of polymorphic loci. In certain exemplary embodiments, the modeler generates the association probabilities by considering associations between polymorphic loci on chromosomal segments.

[0258] In one exemplary embodiment, provided herein is a system for detecting chromosomal ploidy in a sample from an individual, the system comprising: an input processor configured to receive nucleic acid sequence data for alleles at a set of polymorphic loci on a chromosomal segment in the individual and use the nucleic acid sequence data to detect allele frequencies at the set of loci; an error correction unit configured to correct errors in the detected allele frequencies and create corrected allele frequencies for the set of polymorphic loci; a modeler configured to create phased allele information for the set of polymorphic loci by estimating the phase of the nucleic acid sequence data, create individual probabilities of allele frequencies for the polymorphic loci for different ploidy states by comparing the phased allele information to a set of models of different ploidy states and allelic imbalance fractions for the set of polymorphic loci, and create a joint probability for the set of polymorphic loci by combining the individual probabilities taking into account the relative distance between the polymorphic loci on the chromosomal segment; and a hypothesis manager configured to select a best-fitting model that is indicative of chromosomal aneuploidy based on the joint probabilities.

[0259] In certain exemplary system embodiments provided herein, the set of polymorphic loci includes 1,000 to 50,000 polymorphic loci. In certain exemplary system embodiments provided herein, the set of polymorphic loci includes 100 known heterozygosity hotspot loci. In certain exemplary system embodiments provided herein, the set of polymorphic loci includes 100 loci that are within 0.5 kb of or within a recombination hotspot.

[0260] In certain exemplary system embodiments provided herein, the best fitting model analyzes the following ploidy states of a first homolog of a chromosomal segment and a second homolog of a chromosomal segment: (1) none of the cells have a deletion or amplification of the first homolog or the second homolog of the chromosomal segment, (2) some or all of the cells have a deletion of the first homolog of the chromosomal segment or an amplification of the second homolog, or (3) some or all of the cells have a deletion of the second homolog of the chromosomal segment or an amplification of the first homolog.

[0261] In certain exemplary system embodiments provided herein, the errors corrected include allele amplification efficiency bias, contamination, and / or sequencing errors. In certain exemplary system embodiments provided herein, the contamination includes ambient contamination and genotypic contamination. In certain exemplary system embodiments provided herein, ambient contamination and genotypic contamination are determined for homozygous alleles.

[0262] In certain exemplary system embodiments provided herein, the hypothesis manager is configured to analyze the magnitude of the difference between the phasing allele information generated for the model and the estimated allele frequencies. In certain exemplary system embodiments provided herein, the modeler generates individual probabilities of allele frequencies based on a beta-binomial model of predicted and observed allele frequencies at the set of polymorphic loci. In certain exemplary system embodiments provided herein, the modeler generates the individual probabilities using a Bayesian classifier.

[0263] In certain exemplary system embodiments provided herein, nucleic acid sequence data are generated by performing high-throughput DNA sequencing of multiple copies of a series of amplicons generated using a multiplex amplification reaction, where each amplicon in the series spans at least one polymorphic locus in a set of polymorphic loci, and each polymorphic locus in the set is amplified. In certain exemplary system embodiments provided herein, the multiplex amplification reaction is performed under restrictive primer conditions for at least half of the reactions. In certain exemplary system embodiments provided herein, the samples have an average allelic imbalance of 0.4% to 5%.

[0264] In certain exemplary system embodiments provided herein, the sample is a plasma sample from an individual suspected of having cancer, and the hypothesis manager is further configured to determine whether copy number variation is present in tumor cells of the individual based on the best fitting model.

[0265] In certain exemplary system embodiments provided herein, the sample is a plasma sample from the individual, and the hypothesis manager is further configured to determine whether cancer is present in the individual based on the best fitting model. In these embodiments, the hypothesis manager may be further configured to detect a single nucleotide variant at a single nucleotide variant position in the set of single nucleotide variant positions, wherein detecting either a chromosomal aneuploidy or a single nucleotide variant, or both, indicates the presence of circulating tumor nucleic acid in the sample.

[0266] In certain exemplary system embodiments provided herein, the input processor is further configured to receive haplotype information of chromosomal segments for a tumor of an individual, and the modeler is configured to use this haplotype information to generate a set of models of different ploidy states and allelic imbalance fractions for the set of polymorphic loci.

[0267] In certain exemplary system embodiments provided herein, the modeler creates models over a range of allelic imbalance fractions from 0% to 25%.

[0268] It will be understood that any of the methods provided herein may be implemented by computer-readable code stored on a non-transitory computer-readable medium. Thus, in one embodiment, provided herein is a non-transitory computer-readable medium for detecting chromosomal ploidy in a sample from an individual, the non-transitory computer-readable medium comprising computer-readable code that, when executed by a processing device, causes the processing device to receive allele frequency data including the amount of each allele present in the sample at each locus in a set of polymorphic loci on a chromosomal segment, generate phasing allele information for the set of polymorphic loci by estimating the phase of the allele frequency data, use the allele frequency data to generate individual probabilities of allele frequencies for the polymorphic loci for different ploidy states, use the individual probabilities and the phasing allele information to generate joint probabilities for the set of polymorphic loci, and determine the ploidy of the chromosomal segment by selecting a best-fitting model that is indicative of chromosomal ploidy based on the joint probabilities.

[0269] In certain computer-readable medium embodiments, the allele frequency data is generated from nucleic acid sequence data. Certain computer-readable medium embodiments further include correcting errors in the allele frequency data and using the corrected allele frequency data in the step of generating the individual probabilities. In certain computer-readable medium embodiments, the error corrected is allele amplification efficiency bias. In certain computer-readable medium embodiments, the individual probabilities are generated using a set of models of different ploidy states and allele imbalance fractions for the set of polymorphic loci. In certain computer-readable medium embodiments, the linkage probabilities are generated by considering linkages between polymorphic loci on chromosomal segments.

[0270] In one particular embodiment, provided herein is a non-transitory computer-readable medium for detecting chromosomal ploidy in a sample from an individual, the non-transitory computer-readable medium comprising computer-readable code that, when executed by a processing device, causes the processing device to receive nucleic acid sequence data for alleles at a set of polymorphic loci on a chromosomal segment in the individual, use the nucleic acid sequence data to detect allele frequencies at the set of loci, correct for allele amplification efficiency bias in the detected allele frequencies to create corrected allele frequencies for the set of polymorphic loci, estimate the phase of the nucleic acid sequence data to create phasing allele information for the set of polymorphic loci, compare the corrected allele frequencies with a set of models of different ploidy states and allele imbalance fractions for the set of polymorphic loci to create individual probabilities of allele frequencies for the polymorphic loci for different ploidy states, combine the individual probabilities that take into account the linkages between the polymorphic loci on the chromosomal segment to create a joint probability for the set of polymorphic loci, and select a best-fitting model that is indicative of chromosomal aneuploidy based on the joint probabilities.

[0271] In certain exemplary computer-readable medium embodiments, the selecting is performed by analyzing the magnitude of the difference between the phasing allele information and the estimated allele frequencies generated for the model.

[0272] In certain exemplary computer-readable medium embodiments, individual probabilities of allele frequencies are generated based on a beta-binomial model of predicted and observed allele frequencies at a set of polymorphic loci.

[0273] It will be appreciated that any method embodiments provided herein may be performed by executing code stored on a non-transitory computer-readable medium.

[0274] Exemplary Embodiments for Detecting Cancer In certain embodiments, the present invention provides a method for detecting cancer.It will be understood that the sample can be a tumor sample or a liquid sample, such as plasma, from an individual suspected of having cancer.This method is particularly effective for detecting genetic mutations, such as single nucleotide variations, such as SNVs, or copy number variations, such as CNVs, in samples that contain low levels of these genetic variations as part of the total DNA in the sample.Therefore, the sensitivity of detecting DNA or RNA from cancer in samples is exceptional.To achieve this exceptional sensitivity, this method can combine any or all of the improvements provided herein for detecting CNVs and SNVs.

[0275] Thus, in certain embodiments, provided herein are a method for determining whether circulating tumor nucleic acid is present in an individual's sample, and a non-transitory computer-readable medium, the non-transitory computer-readable medium comprising computer-readable code that, when executed on a processing device, causes the processing device to perform the method. The method includes analyzing the sample to determine the ploidy at a set of polymorphic loci on a chromosomal segment in the individual, and determining the level of average allelic imbalance present at the polymorphic loci based on the determined ploidy, wherein an average allelic imbalance equal to or greater than 0.4%, 0.45%, 0.5%, 0.6%, 0.7%, 0.75%, 0.8%, 0.9%, or 1% is indicative of the presence of circulating tumor nucleic acid (e.g., ctDNA) in the sample.

[0276] In certain exemplary embodiments, an average allelic imbalance of more than 0.4, 0.45, or 0.5% is indicative of the presence of ctDNA. In certain embodiments, the method for determining whether circulating tumor nucleic acid is present further comprises detecting a single nucleotide variant at the single nucleotide variance site in the set of single nucleotide variance positions, and detecting an allelic imbalance of equal to or greater than 0.5%, or detecting a single nucleotide variant, or both, is indicative of the presence of circulating tumor nucleic acid in the sample. It will be understood that any of the methods provided for detecting chromosome ploidy or CNV can be used to determine the level of allelic imbalance (typically expressed as average allelic imbalance). It will be understood that any of the methods provided herein for detecting SNV can be used to detect a single nucleotide for this aspect of the present invention.

[0277] In certain embodiments, the method for determining whether circulating tumor nucleic acid is present further comprises performing the method on a control sample having a known average allelic imbalance ratio. The control may be, for example, a sample from an individual's tumor. In some embodiments, the control has an expected average allelic imbalance relative to the sample being analyzed. For example, an AAI of 0.5% to 5%, or an average allelic imbalance ratio of 0.5%.

[0278] In certain embodiments, the analyzing step of the method for determining whether circulating tumor nucleic acid is present comprises analyzing a set of chromosomal segments known to exhibit aneuploidy in cancer. In certain embodiments, the analyzing step of the method for determining whether circulating tumor nucleic acid is present comprises analyzing 1,000 to 50,000 or 100 to 1,000 polymorphic loci for ploidy. In certain embodiments, the analyzing step of the method for determining whether circulating tumor nucleic acid is present comprises analyzing 100 to 1,000 single-nucleotide variant sites. For example, in these embodiments, the analyzing step may comprise performing multiplex PCR to amplify amplicons across 1,000 to 50,000 polymorphic loci and 100 to 1,000 single-nucleotide variant sites. This multiplex reaction can be configured as a single reaction or as a pool of multiplex reactions of different subsets. The multiplex reaction methods provided herein (e.g., the massively multiplex PCR disclosed herein) provide exemplary processes for performing amplification reactions to help achieve improved multiplexing and, therefore, sensitivity levels.

[0279] In certain embodiments, multiplex PCR reactions are performed under restrictive primer conditions for at least 10%, 20%, 25%, 50%, 75%, 90%, 95%, 98%, 99%, or 100% of the reactions. Improved conditions for performing large-scale multiplex reactions provided herein can be used.

[0280] In certain aspects, the above-described method for determining whether circulating tumor nucleic acid is present in a sample from an individual, and all embodiments thereof, can be performed using a system. This disclosure provides teachings regarding specific functional and structural features for carrying out the above-described methods. By way of non-limiting example, the system may include:

[0281] an input processor configured to analyze data from the sample to determine ploidy at a set of polymorphic loci on a chromosomal segment in the individual;

[0282] A modeler configured to determine the level of allelic imbalance present at the polymorphic locus based on the ploidy determination, wherein an allelic imbalance equal to or greater than 0.5% is indicative of the presence of cycling.

[0283] Exemplary embodiments for detecting single nucleotide variants In certain aspects, the present specification provides a method for detecting single nucleotide variants in a sample.The improved method provided herein can achieve a detection limit of 0.015, 0.017, 0.02, 0.05, 0.1, 0.2, 0.3, 0.4 or 0.5% SNV in a sample.All embodiments for detecting SNV can be performed using a system.The present disclosure provides teachings on specific functional and structural features for performing the above-mentioned method.Furthermore, the present specification provides an embodiment comprising a non-transitory computer-readable medium, which comprises computer-readable code, and when executed by a processing device, causes the processing device to perform the method for detecting SNV provided herein.

[0284] Thus, in one embodiment, provided herein is a method for determining whether a single nucleotide variant is present at a set of genomic locations in a sample from an individual, the method comprising: generating, for each genomic location, estimates of efficiency and error rate per cycle for the amplicon spanning that genomic location using a training dataset; receiving observed nucleotide identity information for each genomic location in the sample; determining a set of probabilities of single nucleotide variant fractions resulting from one or more actual mutations at each genomic location by comparing the observed nucleotide identity information at each genomic location to different variant fraction models using the estimates of amplification efficiency and error rate per cycle for each genomic location independently; and determining the fraction and confidence of the most likely actual variant from the set of probabilities for each genomic location.

[0285] In an exemplary embodiment of a method for determining whether a single nucleotide variant is present, efficiency and error rate per cycle estimates are generated for a set of amplicons spanning a genomic location. For example, the set may include 2, 3, 4, 5, 10, 15, 20, 25, 50, 100, or more amplicons spanning the genomic location.

[0286] In an exemplary embodiment of a method for determining whether a single nucleotide variant is present, the observed nucleotide identity information includes the observed number of total reads for each genomic location and the observed number of variant allele reads for each genomic location.

[0287] In an exemplary embodiment of the method for determining whether a single nucleotide variant is present, the sample is a plasma sample and the single nucleotide variant is present in circulating tumor DNA of the sample.

[0288] In another embodiment, the present invention provides a method for estimating the proportion of single nucleotide variants present in a sample from an individual.The method includes: At a set of genome locations, using a training data set, generate estimates of the efficiency and error rate per cycle for one or more amplicons that span these genome locations; For each genome location in the sample, receive observed nucleotide identity information; Using the amplification efficiency and error rate per cycle of the amplicon, generate mean and variance estimates for the mean and variance of the total number of molecules, background error molecules and actual mutant molecules for a search space that includes the initial proportion of actual mutant molecules; Using the mean and variance estimates, determine the proportion of the most likely actual single nucleotide variants by fitting the distribution to the observed nucleotide identity information in the sample, thereby determining the proportion of the single nucleotide variants present in the sample that result from actual mutations.

[0289] In an illustrative example of this method for estimating the proportion of a single nucleotide variant present in a sample, the sample is a plasma sample and the single nucleotide variant is present in circulating tumor DNA of the sample.

[0290] The training dataset in this embodiment of the present invention typically includes samples from one healthy individual or, preferably, a group of healthy individuals. In certain exemplary embodiments, the training dataset is analyzed on the same day or in the same run as one or more samples under test. For example, samples from a group of 2, 3, 4, 5, 10, 15, 20, 25, 30, 36, 48, 96, 100, 192, 200, 250, 500, 1000, or more healthy individuals may be used to create a training dataset. If data on a larger number of healthy individuals (e.g., 96 or more) is available, the confidence in the estimation of amplification efficiency increases, even if a run is performed before the method is performed on the samples under test. Since the PCR error rate is per amplicon, nucleic acid sequence information generated not only for the SNV base position but also for the entire amplified region surrounding the SNV may be used. For example, using samples from 50 individuals and sequencing 20-base-pair amplicons surrounding the SNV, error frequency data from 1000-base reads can be used to determine the error frequency rate.

[0291] Typically, amplification efficiency is estimated by estimating the mean and standard deviation of the amplification efficiency for the amplified segment and then fitting this to a distribution model (e.g., binomial distribution or beta-binomial distribution). The error rate is determined for PCR with a known number of cycles, and then the error rate per cycle is estimated.

[0292] In certain exemplary embodiments, estimating the starting molecules for the test dataset further comprises updating the efficiency estimate for the test dataset using the starting molecule number estimated in step (b) if the observed number of reads significantly differs from the estimate of the number of reads. This estimate can then be updated for the new efficiency and / or starting molecules.

[0293] The search space used to estimate the total number of molecules, background error molecules, and actual mutant molecules may include a search space with a lower limit of 0.1%, 0.2%, 0.25%, 0.5%, 1%, 2.5%, 5%, 10%, 15%, 20%, or 25% of the copies of the base at the SNV position, which is the SNV base, and an upper limit of 1%, 2%, 2.5%, 5%, 10%, 12.5%, 15%, 20%, 25%, 50%, 75%, 90%, or 95%. Lower ranges, such as a lower limit of 0.1%, 0.2%, 0.25%, 0.5%, or 1%, and an upper limit of 1%, 2%, 2.5%, 5%, 10%, 12.5%, or 15%, may be used in an exemplary example for plasma samples, where the method detects circulating tumor DNA. For tumor samples, even higher ranges are used.

[0294] A distribution is fitted to the number of total error molecules (background errors and actual mutations) in the total molecules to calculate the likelihood or probability for each possible actual mutation in the search space. This distribution may be a binomial or beta-binomial distribution.

[0295] The most likely actual mutation is determined by determining the proportion of the most likely actual mutation and calculating the reliability using the data from the distribution fitting.As an illustrative example, without intending to limit the clinical interpretation provided herein, when the average mutation rate is high, the reliability rate required to make a positive SNV determination will be low.For example, if the average mutation rate for SNVs in samples using the most likely hypothesis is 5% and the reliability rate is 99%, a positive SNV call will be made.On the other hand, for this illustrative example, if the average mutation rate for SNVs in samples using the most likely hypothesis is 1% and the reliability rate is 50%, in certain circumstances, a positive SNV call will not be made.It will be understood that the clinical interpretation of data can be a function of sensitivity, specificity, prevalence and the availability of alternative products.

[0296] In certain exemplary embodiments, the sample is a circulating DNA sample, eg, a circulating tumor DNA sample.

[0297] In another embodiment, provided herein is a method for detecting one or more single nucleotide variants in a test sample from an individual, the method comprising the steps of:

[0298] determining a median variant allele frequency for a plurality of control samples from each of a plurality of normal individuals for each single nucleotide variant position in the set of single nucleotide variant positions based on results produced in the sequencing run to identify selected single nucleotide variant positions having a median variant allele frequency in the normal samples that is below a threshold, and determining a background error for each single nucleotide variant position after removing outlier samples for each single nucleotide variant position; determining a weighted mean and variance of observed read depth for the selected single nucleotide variant positions for the test samples based on data produced in the sequencing run for the test samples; and detecting one or more single nucleotide variants by using a computer to identify one or more single nucleotide variant positions having a statistically significant weighted mean read depth compared to the background error for that position.

[0299] In certain embodiments of this method for detecting one or more SNVs, the sample is a plasma sample, the control sample is a plasma sample, and the one or more detected single nucleotide variants are present in the circulating tumor DNA of the sample.In certain embodiments of this method for detecting one or more SNVs, the plurality of control samples comprises at least 25 samples.In certain exemplary embodiments, the plurality of control samples is at least 5, 10, 15, 20, 25, 50, 75, 100, 200, or 250 samples at the lower limit, and at least 10, 15, 20, 25, 50, 75, 100, 200, 250, 500, and 1000 samples at the upper limit.

[0300] In certain embodiments of this method for detecting one or more SNVs, outliers are removed from the data generated in the high-throughput sequencing run, and a weighted average of the observed read depths is calculated, and the observed variance is determined. In certain embodiments of this method for detecting one or more SNVs, the read depth for each single nucleotide variant position for the test sample is at least 100 reads.

[0301] In certain embodiments of this method for detecting one or more SNVs, the sequencing run includes a multiplex amplification reaction performed under restrictive primer reaction conditions. These embodiments are performed in illustrative examples using the improved methods for performing multiplex amplification reactions provided herein.

[0302] Without being limited by theory, the method of the present embodiment utilizes a background error model using normal plasma samples, sequenced in the same sequencing run as the sample under test, to account for run-specific artifacts. Noisy positions with median common variant allele frequencies above thresholds, e.g., 0.1%, 0.2%, 0.25%, 0.5%, 0.75%, and 1.0%, are removed.

[0303] In order to consider noise and contamination, outlier samples are repeatedly removed from this model.For each base substitution of all genome loci, calculate the average and standard deviation of error weighted by read depth.In certain exemplary embodiments, the sample (for example, tumor or cell-free plasma sample) that has at least a threshold number of reads (for example, at least 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 50, 100, 250, 500 or 1000 variant reads) and in certain embodiments, has a single nucleotide variant position that a1 Z score against background error model is greater than 2.5, 5, 7.5 or 10 is counted as candidate mutation.

[0304] In certain embodiments, a read depth of greater than 100, 250, 500, 1,000, 2000, 2500, 5000, 10,000, 20,000, 25,0000, 50,000 or 100,000 at the lower end of the range and 2000, 2500, 5,000, 7,500, 10,000, 25,000, 50,000, 100,000, 250,000 or 500,000 reads at the upper end is achieved in a sequencing run for each single nucleotide variant position in the set of single nucleotide variant positions. Typically, sequencing run is a high-throughput sequencing run.In exemplary embodiments, the average or median value generated for the sample under test is weighted by read depth.Therefore, the likelihood that variant allele determination is actual in the sample in which one variant allele is detected in 1000 reads is weighted more than the likelihood that one variant allele is detected in 10,000 reads.Because the determination of variant allele (i.e., mutation) is not performed with 100% confidence, the identified single nucleotide variant may be considered as a candidate variant or candidate mutation.

[0305] Exemplary Test Statistics for Analysis of Fading Data Exemplary test statistics are described below for analyzing phasing data from a sample known or suspected to be a mixed sample containing DNA or RNA from two or more genetically non-identical cells. f represents the fraction of DNA or RNA of interest, e.g., the fraction of DNA or RNA containing a CNV of interest, or the fraction of DNA or RNA from a cell of interest, e.g., a cancer cell. In some embodiments of cancer testing, f represents the fraction of DNA or RNA from a cancer cell in a mixture of cancer cells and normal cells, or f represents the fraction of cancer cells in a mixture of cancer cells and normal cells. Note that this refers to the fraction of DNA from a cell of interest, assuming that each cell of interest provides two copies of DNA. This is different from the fraction of DNA from a cell of interest that contains a deleted or duplicated segment.

[0306] The possible allele values ​​for each SNP are designated A and B. AA, AB, BA, and BB are used to represent all possible ordered allele pairs. In some embodiments, SNPs containing the ordered alleles AB or BA are analyzed. i indicates the number of sequence reads for the i-th SNP, and A i and B i denotes the number of reads of the i-th SNP representing alleles A and B, respectively. Assume the following: N i =A i +B i

[0307] Allele ratio R i is defined as follows:

number

[0308] T indicates the number of targeted SNPs.

[0309] Without loss of generality, some embodiments focus on a single chromosome segment.For the sake of clarity, in this specification, the phrase "first homologous chromosome segment compared with second homologous chromosome segment" refers to the first homolog of chromosome segment and the second homolog of chromosome segment.In some such embodiments, all of the target SNPs are included in the target segment chromosome.In other embodiments, multiple chromosome segments are analyzed for possible copy number variation.

[0310] MAP estimation This method exploits knowledge of phasing through ordered allele pairs to detect deletions or duplications of target segments. For each SNPi, we define:

number

[0311] Then, the following definitions are made:

number

[0312] X under various copy number hypotheses (e.g., disomy hypothesis, deletion of the first or second homolog, or duplication of the first or second homolog) i and the distribution of S is given below.

[0313] Disomy hypothesis Under the hypothesis that the target segment is not deleted or duplicated,

number

number

[0314] Assuming a constant read depth N, we have a binomial distribution S with the following parameters:

number

[0315] Negative hypothesis Under the hypothesis that the first homologue is deleted (i.e., the AB SNP becomes B and the BA SNP becomes A), R i has a binomial distribution and parameters for AB SNPs

number

number

number

[0316] Assuming a constant read depth N, we have a binomial distribution S with the following parameters:

number

[0317] Under the hypothesis that the second homolog is deleted (i.e., the AB SNP becomes A and the BA SNP becomes B), R i has a binomial distribution and parameters for AB SNPs

number

number

number

[0318] Assuming a constant read depth N, we have a binomial distribution S with the following parameters:

number

[0319] overlap hypothesis Under the hypothesis that the first homologues are duplicated (i.e., the AB SNP becomes AAB and the BA SNP becomes BBA), R i has a binomial distribution and parameters for AB SNPs

number

number

number

[0320] Assuming a constant read depth N, we have a binomial distribution S with the following parameters:

number

[0321] Under the hypothesis that the second homologue is duplicated (i.e., the AB SNP becomes ABB and the BA SNP becomes BAA), R i has a binomial distribution and parameters for AB SNPs

number

number

number

[0322] Assuming a constant read depth N, we have a binomial distribution S with the following parameters:

number

[0323] classification As shown in the above chapter, X i is a binary random variable with:

number

[0324] This allows the probability of the test statistic S under each hypothesis to be calculated. The probability of each hypothesis given the measured data can be calculated. In some embodiments, the hypothesis with the greatest probability is selected. If desired, a distribution for S can be calculated for each N i can be simplified by approximating it with a constant penetration depth N or by truncating the read depth to a constant value N. This simplification gives:

number

[0325] The value of f can be estimated by selecting the most likely value of f given the measured data, e.g., the value of f that produces the best data fitting using an algorithm (e.g., a search algorithm), e.g., maximum likelihood estimation, empirical maximum estimation, or Bayesian estimation. In some embodiments, multiple chromosomal segments are analyzed, and the value of f is estimated based on the data for each segment. If all target cells have these duplications or deletions, the estimated values ​​of f based on the data for these different segments will be similar. In some embodiments, f is experimentally measured, e.g., by determining the fraction of DNA or RNA from cancer cells based on the difference in methylation (hypomethylation or hypermethylation) of cancer and non-cancerous DNA or RNA.

[0326] Single hypothesis rejection The distribution of S for the disomy hypothesis does not depend on f. Therefore, the probability of measurement data can be calculated for the disomy hypothesis without calculating f. A single hypothesis rejection test can be used for the null hypothesis of disomy. In some embodiments, the probability of S for the disomy hypothesis is calculated, and the disomy hypothesis is rejected if the probability is below a given threshold (e.g., less than 1 in 1,000). This indicates that there is duplication or deletion of a chromosome segment. If desired, the false positive rate can be changed by adjusting the threshold.

[0327] Exemplary Method for Analysis of Fading Data Exemplary methods are described below for analyzing data from samples known or suspected to be mixed samples containing DNA or RNA from two or more genetically non-identical cells. In some embodiments, phasing data is used. In some embodiments, the method involves determining, for each calculated allele ratio, whether the calculated allele ratio for a particular locus is above or below the predicted allele ratio and the magnitude of the difference. In some embodiments, a likelihood distribution is determined for the allele ratios at loci for a particular hypothesis, and the closer the calculated allele ratio is to the center of the likelihood distribution, the more likely the hypothesis is correct. In some embodiments, the method involves determining the likelihood that a hypothesis is correct for each locus. In some embodiments, the method involves determining the likelihood that a hypothesis is correct for each locus and combining the probabilities of the hypotheses for each locus, and the hypothesis with the greatest joint probability is selected. In some embodiments, the method involves determining the likelihood that a hypothesis is correct for each locus and for each possible ratio of DNA or RNA from one or more target cells to total DNA or RNA in the sample. In some embodiments, the joint probability for each hypothesis is determined by adding together the probabilities of the hypotheses for each locus and each possible ratio, and the hypothesis with the greatest joint probability is selected.

[0328] In one embodiment, the following hypotheses are considered: 11 (All cells are normal), H 10 (presence of cells with only homolog 1, and therefore lack of homolog 2), H 01 (presence of cells with only homolog 2, and therefore lack of homolog 1), H 21 (presence of cells with homolog 1 duplication), H 12 (Presence of cells with homolog 2 duplication). For a fraction f of target cells, such as cancer cells or mosaic cells (or fraction of DNA or RNA from target cells), the predicted allele ratio for a heterozygous (AB or BA) SNP can be found as follows: Formula (1):

number

[0329] Correction of bias, contamination and sequencing errors: Observation D in SNP s is the number of originally mapped reads in which each allele is present, n A 0 and n B 0 Then, using the expected bias in the amplification of the A and B alleles, the corrected read n A and n B can be found.

[0330] c a indicates ambient contamination (e.g., contamination from DNA in the air or environment), and r(c a ) indicates the allele ratio for the ambient pollutant (initially taken as 0.5). Furthermore, c g indicates the genotype contamination rate (e.g., contamination from another sample), and r(c g ) is the allele ratio for that contaminant. e (A,B) and s e (B,A) indicates a sequencing error that calls one allele as a different allele (e.g., by mistakenly detecting the A allele when the B allele is present).

[0331] By correcting for ambient contamination, genotypic contamination, and sequencing errors, for a given predicted allele ratio r, we can estimate the observed allele ratio q(r,c a ,r(c a ),cg ,r(c g ),s e (A,B),s e (B,A)) can be found.

[0332] Since the genotype of the contaminant is unknown, the ensemble frequency is used to calculate P(r(c g ) can be found. More specifically, p is the population frequency for one of the alleles (sometimes called the reference allele). Then, P(r(c g )=0)=(1-p) 2 , P(r(c g )=0)=2p(1-p) and P(r(cg)=0)=p 2 r(c g ) to find the conditional expectation over E[q(r,c a ,r(c a ),c g ,r(c g ),s e (A,B),s e (B, A)) can be determined. Note that ambient and genotypic contamination are determined using homozygous SNPs and are therefore not affected by the presence or absence of deletions or duplications. Furthermore, if desired, ambient and genotypic contamination can be measured using a reference chromosome.

[0333] Likelihood at each SNP: The following formula takes into account the allele ratio r and A and n B gives the probability of observing Formula (2):

number

[0334] D s indicates the data for the SNP. 11 ,H 01 ,H 10 ,H 21 ,H 12}, in formula (1), r = r(AB, h) or r = r(BA, h), and g ) and find the conditional expectation over the observed allele ratios E[q(r,c a ,r(c a ),c g ,r(c g Then, in equation (2), r = E[q(r,c a ,r(c a ),c g ,r(c g ),s e (A,B),s e (B,A))], and P(D s |h,f) can be determined.

[0335] Search algorithm: In some embodiments, SNPs with allele ratios that appear to be outliers are ignored (e.g., by ignoring or excluding SNPs with allele ratios that are at least two or three standard deviations above or below the mean). Note that an identified advantage of this approach is that it ensures that SNPs are not trimmed due to mosaicism, as the allele ratios may be more variable in the presence of higher rates of mosaicism.

[0336] F={f1, ,f N} denotes the search space for the proportion of mosaicism (e.g., tumor fraction). P(Ds|h,f) at each SNP and fε F can be determined and the likelihoods for all SNPs can be combined.

[0337] The algorithm runs for each hypothesis over each f. A search method is used to find the range F of f where the confidence of the deletion or duplication hypothesis is higher than the confidence of the no deletion and no duplication hypotheses. * When F is present, it is concluded that mosaicism exists. * In P(D s The maximum likelihood estimate of |h,f) is determined. If desired, fε F *If desired, the confidence for each hypothesis can be determined.

[0338] In some embodiments, a beta-binomial distribution is used instead of a binomial distribution. In some embodiments, a reference chromosome or chromosome segment is used to determine the sample-specific parameters of the beta-binomial.

[0339] Theoretical performance using simulation: If desired, the theoretical performance of the algorithm can be evaluated by randomly assigning the number of reference reads to SNPs at a given depth of read (DOR). Typically, p = 0.5 is used for the binomial probability parameter, and p is modified accordingly for deletions or duplications. Exemplary input parameters for each simulation are: (1) the number of SNPs, S; (2) a constant DOR per SNP, D; (3) p; and (4) the number of experiments.

[0340] First simulation experiment: The experiment focused on S ∈ {500, 1000}, D ∈ {500, 1000}, and p ∈ {0%, 1%, 2%, 3%, 4%, 5%}. For each setting, 1,000 simulated experiments were performed (thus, 24,000 experiments with phases and 24,000 experiments without phases). The number of reads was simulated from a binomial distribution (other distributions may be used if desired). The false positive rate (when p = 0%) and false negative rate (when p > 0%) were determined with and without phase information. Note that phase information is particularly useful for S = 1000, D = 1000. However, for S = 500, D = 500, the algorithm has the highest false positive rate, regardless of whether or not the condition being tested is phased out.

[0341] Phase information is especially useful at low mosaic rates (≤3%). Without phase information, the confidence in the deletion is limited to H 10 and H 01Because the probability of a hypothesis being determined by assigning equal chances to all hypothesis, a high level of false negatives is observed for p=1%, and small deviations in favor of one hypothesis are not enough to compensate for the low likelihood of the other hypothesis. This is similarly true for overlaps. The algorithm also appears to be more sensitive to read depth compared to the number of SNPs. For results using phase information, we assume that complete phase information is available for a large number of consecutive heterozygous SNPs. If desired, haplotype information can be obtained by probabilistically matching haplotypes for smaller segments.

[0342] Second simulation experiment: This experiment focused on random experiments with S ∈ {100, 200, 300, 400, 500}, D ∈ {1000, 2000, 3000, 4000, 5000}, and p ∈ {0%, 1%, 1.5%, 2%, 2.5%, 3%} and 10,000 for each setting. The false positive rate (for p = 0%) and false negative rate (for p > 0%) were determined with and without phase information. The false negative rate was less than 10% for D ≥ 3000 and N ≥ 200 using haplotype information, while the same performance was achieved for D = 5000 and N ≥ 400. For small mosaic fractions, the difference in false negative rate was particularly striking. For example, for p = 1%, without haplotype data, a false negative rate of less than 20% was never achieved, whereas it was close to 0% for N ≥ 300 and D ≥ 3000. For p=3%, a 0% false negative rate is observed using haplotype data, whereas without haplotype data, N≧300 and D≧3000 are required to reach the same performance.

[0343] Exemplary Methods for Detecting Deletions and Duplications Without Phasing Data In some embodiments, non-phasing genetic data is used to determine whether there is an over-representation of the copy number of a first homologous chromosomal segment compared to a second homologous chromosomal segment in an individual's genome (for example, in the genome of one or more cells, or in cfDNA or cfRNA).In some embodiments, phasing genetic data is used, but phasing is ignored.In some embodiments, the DNA or RNA sample is a mixed sample of cfDNA or cfRNA from an individual, which contains cfDNA or cfRNA from two or more genetically different cells.In some embodiments, this method utilizes the magnitude of the difference between the calculated allele ratio and the predicted allele ratio for each locus.

[0344] In some embodiments, the methods involve obtaining genetic data at a set of polymorphic loci on chromosomes or chromosome segments in a DNA or RNA sample from one or more cells from an individual by measuring the amount of each allele at each locus. In some embodiments, allele ratios are calculated for loci that are heterozygous in at least one cell from which the sample is derived. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one allele divided by the total measured amount of all alleles for that locus. In some embodiments, the calculated allele ratio for a particular locus is the measured amount of one allele (e.g., an allele on a first homologous chromosome segment) divided by the measured amounts of one or more other alleles for that locus (e.g., an allele on a second homologous chromosome segment). Calculated allele ratios and predicted allele ratios may be calculated using any of the methods described herein or any standard method (e.g., any mathematical transformation of the calculated allele ratios or predicted allele ratios described herein).

[0345] In some embodiments, a test statistic is calculated for each locus based on the magnitude of the difference between the calculated allele ratio and the expected allele ratio. In some embodiments, the test statistic Δ is calculated using the following formula:

number

[0346] In the formula, δ i is the magnitude of the difference between the calculated allele ratio and the predicted allele ratio for the i locus, In the formula, μ i is δ i is the average value of During the ceremony,

number

[0347] For example, if the predicted allele ratio is 0.5, then δ i can be defined as follows:

number

[0348] μ i and δ i The value of R i is a binomial random variable. In some embodiments, the standard deviation is assumed to be the same for all loci. In some embodiments, the mean or weighted mean of the standard deviations, or an estimate of the standard deviation, is

number

[0349] In some embodiments, a set of one or more hypotheses is enumerated, which indicates the copy number of a chromosome or chromosome segment in one or more genomes of a cell. In some embodiments, the most likely hypothesis is selected based on test statistics, thereby determining the copy number of a chromosome or chromosome segment in one or more genomes of a cell. In some embodiments, if the probability that a test statistic belongs to the distribution of test statistics for a hypothesis exceeds an upper threshold, the hypothesis is selected. If the probability that a test statistic belongs to the distribution of test statistics for a hypothesis is below a lower threshold, the one or more hypotheses are rejected, or if the probability that a test statistic belongs to the distribution of test statistics for a hypothesis is between the lower and upper thresholds, or if the probability is not determined with sufficient confidence, the hypothesis is not selected or rejected. In some embodiments, the upper and / or lower thresholds are determined, for example, from empirical distributions from training data (e.g., samples with known copy numbers, such as diploid samples or samples known to have a specific deletion or duplication). Such empirical distributions can be used to select thresholds for single-hypothesis rejection testing. Note that the test statistic Δ is independent of S, so either can be used independently if desired.

[0350] Exemplary Methods for Detecting Deletions or Duplications Using Allele Distributions or Patterns This section includes methods for determining whether there is an overrepresentation of a copy number of a first homologous chromosomal segment compared to a second homologous chromosomal segment. In some embodiments, the method involves enumerating (i) a plurality of hypotheses indicating the copy number of a chromosome or chromosomal segment present in the genome of one or more cells (e.g., cancer cells) of an individual, or (ii) a plurality of hypotheses indicating the degree of overrepresentation of a copy number of a first homologous chromosomal segment compared to a second homologous chromosomal segment in the genome of one or more cells of the individual. In some embodiments, the method involves obtaining genetic data from an individual at a plurality of polymorphic loci (e.g., SNP loci) on a chromosome or chromosomal segment. In some embodiments, a probability distribution of the individual's predicted genotype for each hypothesis is generated. In some embodiments, a data fit between the obtained individual's genetic data and the probability distribution of the individual's predicted genotype is calculated. In some embodiments, one or more hypotheses are ranked according to the data fit, and the highest-ranked hypothesis is selected. In some embodiments, a technique or algorithm, such as a search algorithm, is used to calculate a data fitting, rank hypotheses, or select the highest ranked hypothesis. In some embodiments, the data fitting is a fitting to a beta-binomial distribution or a fitting to a binomial distribution. In some embodiments, the technique or algorithm is selected from the group consisting of maximum likelihood estimation, empirical maximum estimation, Bayesian estimation, dynamic estimation (e.g., dynamic Bayesian estimation), and expectation maximization estimation. In some embodiments, the method includes applying the above-mentioned technique or algorithm to the obtained genetic data and the predicted value of the genetic data.

[0351] In some embodiments, the method involves enumerating (i) a plurality of hypotheses indicating the copy number of a chromosome or chromosome segment present in the genome of one or more cells (e.g., cancer cells) of an individual, or (ii) a plurality of hypotheses indicating the degree of copy number overrepresentation of a first homologous chromosome segment compared to a second homologous chromosome segment in the genome of one or more cells of the individual. In some embodiments, the method involves obtaining genetic data from the individual at a plurality of polymorphic loci (e.g., SNP loci) on a chromosome or chromosome segment. In some embodiments, the genetic data includes allele counts for the plurality of polymorphic loci. In some embodiments, a joint distribution model is created for predicted values ​​of allele counts at the plurality of polymorphic loci on a chromosome or chromosome segment for each hypothesis. In some embodiments, the relative probabilities of one or more of the hypotheses are determined using the joint distribution model and the allele counts measured for the sample, and the hypothesis with the greatest probability is selected.

[0352] In some embodiments, the distribution or pattern of alleles (e.g., the pattern of calculated allele ratios) is used to determine the presence or absence of a CNV (e.g., a deletion or duplication). If desired, the parental origin of the CNV can be determined based on this pattern.

[0353] Exemplary Counting / Quantification Methods In some embodiments, one or more counting methods (also called quantification methods) are used to detect one or more CNS (e.g., deletion or duplication of a chromosome segment or entire chromosome). In some embodiments, one or more counting methods are used to determine whether the overrepresentation of the copy number of a first homologous chromosome segment is due to duplication of the first homologous chromosome segment or deletion of a second homologous chromosome segment. In some embodiments, one or more counting methods are used to determine the excess copy number of a duplicated chromosome segment or chromosome (e.g., whether there are 1, 2, 3, 4, or more excess copies). In some embodiments, one or more counting methods are used to distinguish samples with many duplications and a small tumor fraction from samples with few duplications and a large tumor fraction. For example, one or more counting methods may be used to distinguish a sample with four excess chromosome copies and a tumor fraction of 10% from a sample with two excess chromosome copies and a tumor fraction of 20%. Exemplary methods are disclosed, for example, in U.S. Publication Nos. 2007 / 0184467, 2013 / 0172211, and 2012 / 0003637, U.S. Patent Nos. 8,467,976, 7,888,017, 8,008,018, 8,296,076, and 8,195,415, U.S. Application No. 62 / 008,235 filed June 5, 2014, and U.S. Application No. 62 / 032,785 filed August 4, 2014, each of which is incorporated herein by reference in its entirety.

[0354] In some embodiments, the counting method comprises counting the number of reads based on DNA sequences that map to one or more given chromosomes or chromosome segments.Some such methods involve creating a reference value (cutoff value) for the number of DNA sequence reads that map to a specific chromosome or chromosome segment, and excessive read numbers are indicative of specific genetic abnormalities.

[0355] In some embodiments, the total measured amount of all alleles for one or more loci (e.g., the total number of polymorphic or non-polymorphic loci) is compared to a reference value. In some embodiments, the reference amount is (i) a threshold or (ii) a predicted amount for a particular copy number hypothesis. In some embodiments, the reference amount (in the absence of CNV) is the total measured amount of all alleles for one or more loci for one or more chromosomes or chromosome segments known or predicted to have no deletions or duplications. In some embodiments, the reference amount (in the presence of CNV) is the total measured amount of all alleles for one or more loci for one or more chromosomes or chromosome segments known or predicted to have deletions or duplications. In some embodiments, the reference amount is the total measured amount of all alleles for one or more loci for one or more reference chromosomes or chromosome segments. In some embodiments, the reference amount is the average or median of values ​​determined for two or more different chromosomes, chromosome segments, or different samples. In some embodiments, random (e.g., massively parallel shotgun sequencing) or targeted sequencing is used to determine the amount of one or more polymorphic or non-polymorphic loci.

[0356] In some embodiments utilizing a reference amount, the method includes: (a) measuring the amount of genetic material for a chromosome or chromosome segment of interest; (b) comparing the amount from step (a) to a reference amount; and (c) identifying the presence or absence of a deletion or duplication based on the comparison.

[0357] In some embodiments utilizing a reference chromosome or chromosome segment, the method includes sequencing DNA or RNA from the sample to obtain multiple sequence tags that align to target loci. In some embodiments, the sequence tags are of sufficient length (e.g., 15-100 nucleotides) to be assigned to specific target loci, and the target loci are derived from multiple different chromosomes or chromosome segments, including at least one first chromosome or chromosome segment suspected of having an abnormal distribution in the sample and at least one second chromosome or chromosome segment presumed to have a normal distribution in the sample. In some embodiments, the multiple sequence tags are assigned to their corresponding target loci. In some embodiments, the number of sequence tags assigned to the target loci of the first chromosome or chromosome segment and the number of sequence tags assigned to the target loci of the second chromosome or chromosome segment are determined. In some embodiments, these numbers are compared to determine the presence or absence of an abnormal distribution (e.g., deletion or duplication) of the first chromosome or chromosome segment.

[0358] In some embodiments, the value of f (e.g., tumor fraction) is used in CNV determination, e.g., to compare the observed difference in the dosage of two chromosomes or chromosome segments with the difference expected for a particular type of CNV given the value of f (see, e.g., U.S. Publication Nos. 2012 / 0190020, 2012 / 0190021, 2012 / 0190557, and 2012 / 0191358, each of which is incorporated by reference in its entirety). For example, the difference in dosage of a duplicated chromosome segment in a tumor compared to a disomic reference chromosome segment increases as the tumor fraction increases. In some embodiments, the method includes comparing the relative frequency of a chromosome or chromosome segment of interest with a reference chromosome or chromosome segment (e.g., a chromosome or chromosome segment predicted or known to be disomic) to the value of f to determine the likelihood of CNV. For example, the difference in dosage between a first chromosome or chromosome segment and a reference chromosome or chromosome segment can be compared to what would be expected given the values ​​of f for various possible CNVs (e.g., one or two extra copies of the chromosome segment of interest).

[0359] The following hypothetical example illustrates the use of counting / quantification methods to distinguish between duplication of a first homologous chromosomal segment and deletion of a second homologous chromosomal segment. Considering the host's normal disomic genome as the baseline, analysis of a mixture of normal and cancer cells provides the average difference between the baseline and cancer DNA in the mixture. For example, imagine that 10% of the DNA in a sample is derived from cells with a deletion across the chromosomal region targeted by the assay. In some embodiments, the quantification method indicates that the amount of reads corresponding to this region is predicted to be 95% of the amount expected for a normal sample. This means that because each tumor cell with a deletion of the target region lacks one of the two target chromosomal regions, the total amount of DNA mapping to this region is 90% (for normal cells) + ½ × 10% (for tumor cells) = 95%. Alternatively, in some embodiments, the allele method indicates that the allele ratio at heterozygous loci is, on average, 19:20. Next, imagine that 10% of the DNA in a sample is derived from cells with a 5-fold focal amplification of the chromosomal region targeted by the assay. In some embodiments, quantitative methods indicate that the amount of reads corresponding to this region is predicted to be 125% of the amount expected for a normal sample. This means that one of the two target chromosomal regions in each tumor cell with a 5-fold focal amplification is overcopied 5 times across the target region, so the total amount of DNA mapping to this region is 90% (for normal cells) + (2 + 5) × 10% (for tumor cells) / 2 = 125%. Alternatively, in some embodiments, allele methods indicate that the allele ratio at heterozygous loci is, on average, 25:20. Note that using only allele methods, a 5-fold focal amplification across a chromosomal region in a sample containing 10% cfDNA may appear to be the same as a deletion across the same region in a sample containing 40% cfDNA.In these two cases, the under-represented haplotype in the deletion case appears to be the haplotype without a CNV in the focal duplication case, and the haplotype without a CNV in the deletion case appears to be the over-represented haplotype in the focal duplication case. The likelihoods generated by the allele approach are combined with those generated by the quantitative approach to distinguish between these two probabilities.

[0360] Exemplary Counting / Quantification Methods Using Reference Samples Exemplary quantification methods using one or more reference samples are described in U.S. Application Nos. 62 / 008,235, filed June 5, 2014, and 62 / 032,785, filed August 4, 2014, which are incorporated by reference herein in their entireties. In some embodiments, one or more reference samples (e.g., normal samples) most likely to be free of CNVs on one or more chromosomes or chromosomes of interest are identified by selecting the sample with the highest tumor DNA fraction, selecting the sample with a z-score closest to 0, selecting a sample whose data fit the hypothesis corresponding to the absence of CNVs with the highest confidence or likelihood, selecting a sample known to be normal, selecting a sample from an individual with the lowest likelihood of having cancer (e.g., young age, male if screening for breast cancer, no family history, etc.), selecting the sample with the highest DNA input, selecting the sample with the highest signal-to-noise ratio, selecting samples based on other criteria believed to correlate with the likelihood of having cancer, or selecting samples using some combination of criteria. Once the reference set is selected, these cases can be assumed to be disomy, and the bias per SNP, i.e., experiment-specific amplification and other processing biases for each locus, can be estimated. This experiment-specific bias estimate can then be used to correct the bias in the measurement of the locus of interest, for example, chromosome 21, and, if appropriate, for other chromosome loci, for samples that are not part of the subset in which disomy is not assumed for chromosome 21. Once the bias has been corrected for these samples with unknown ploidy, the data for these samples can be analyzed twice using the same or different methods to determine whether the individual has trisomy 21. For example, a quantitative method can be used for the remaining samples with unknown ploidy, and a z-score can be calculated using the measurements of the corrected genetic data for chromosome 21. Alternatively, the tumor fraction of samples from individuals suspected of having cancer can be calculated as part of a preliminary estimation of the ploidy status of chromosome 21.The expected corrected read percentages for disomy (disomy hypothesis) and trisomy (trisomy hypothesis) can be calculated for cases with that tumor fraction. Alternatively, if the tumor fraction has not been measured in advance, a set of disomy and trisomy hypotheses can be created for different tumor fractions. For each case, a predicted distribution of corrected read percentages can be calculated, taking into account given expected statistical variation in the selection and measurement of various DNA loci. The observed corrected read percentages can be compared to the predicted distribution of corrected read percentages, and likelihood ratios can be calculated for the disomy and trisomy hypotheses for each sample with unknown ploidy. The ploidy state associated with the hypothesis with the highest calculated likelihood can be selected as the correct ploidy state.

[0361] In some embodiments, a subset of samples with a sufficiently low likelihood of having cancer may be selected to serve as a control set of samples. This subset may be a fixed number, or may be a variable number based on selecting only samples below a threshold. Quantitative data from a subset of samples may be combined, averaged, or combined using a weighted average, with the weighting based on the likelihood of the sample being normal. The quantitative data may be used to determine per locus bias for amplification sequencing of samples in an immediate batch of control samples. Per locus bias may also include data from other batches of samples. Per locus bias may indicate the relative over- or under-amplification observed for that locus compared to other loci, and may indicate that any observations of over- or under-amplification are due to amplification and / or sequencing or other biases, assuming the subset of samples does not contain CNVs. Per locus bias may also take into account the GC content of the amplicon. Loci may be grouped into locus groups for purposes of calculating per locus bias. Once the bias per locus has been calculated for each locus in the plurality of loci, the sequencing data for one or more samples not in the sample subset, and optionally one or more samples in the sample subset, may be corrected by adjusting the quantitative measurements for each locus to remove the effect of bias at that locus. For example, if SNP1 is observed to have a read depth twice as large as the average in a subset of patients, the adjustment may involve replacing the corresponding number of reads from SNP1 with a number half that size. If the locus in question is a SNP, the adjustment may involve halving the number of reads corresponding to each allele at that locus. Once the sequencing data for each locus in one or more samples has been adjusted, it may be analyzed using a method to detect the presence of CNVs in one or more chromosomal regions.

[0362] In one example, sample A is a mixture of amplified DNA derived from a mixture of normal and cancerous cells analyzed using a quantitative method. The following shows exemplary possible data: The region of the q arm on chromosome 22 is found to have only 90% of the expected value of DNA mapping to that region, the focal region corresponding to the HER2 gene is found to have 150% of the expected value of DNA mapping to that region, and the p arm of chromosome 5 is found to have 105% of the expected value of DNA mapping to that region. A physician may infer that the sample has a deletion of a region on the q arm of chromosome 22 and a duplication of the HER2 gene. Because 22q deletions are common in breast cancer, and because cells with deletions of the 22q regions on both chromosomes do not usually survive, a physician may infer that approximately 20% of the DNA in the sample is derived from cells with a 22q deletion on one of the two chromosomes. A physician can also infer that a cell contains a five-fold duplication of the HER2 region if DNA from a mixed sample derived from a tumor cell comes from a set of genetic tumor cells that are homogeneous for the HER2 and 22q regions.

[0363] In one example, sample A is also analyzed using the allele method. The following shows exemplary possible data: Two haplotypes for the same region on the q arm of chromosome 22 exist in a ratio of 4:5, two haplotypes in the focal region corresponding to the HER2 gene exist in a ratio of 1:2, and two haplotypes in the p arm of chromosome 5 exist in a ratio of 20:21. All other assayed regions of the genome do not contain any haplotypes in statistically significant excess. A physician can infer that the sample contains DNA from a tumor with CNVs in the 22q region, the HER2 region, and the 5p arm. Based on the knowledge that 22q deletions are very common in breast cancer and / or a quantitative analysis showing an underrepresentation of DNA mapping to the 22q region of the genome, a physician can infer the presence of a tumor with 22q deletion. Based on the knowledge that HER2 amplification is very common in breast cancer and / or quantitative analysis showing an overrepresentation of the amount of DNA mapping to the HER2 region of the genome, a physician can infer the presence of a tumor with HER2 amplification.

[0364] Exemplary Reference Chromosomes or Chromosome Segments In some embodiments, any of the methods described herein are also performed on one or more reference chromosomes or chromosome segments, and the results are compared to the results for one or more chromosomes or chromosome segments of interest.

[0365] In some embodiments, a reference chromosome or chromosome segment is used as a control that is predicted to be free of CNVs. In some embodiments, the reference is the same chromosome or chromosome segment from one or more different samples that are known or predicted to have no deletions or duplications in the chromosome or chromosome segment. In some embodiments, the reference is a different chromosome or chromosome segment from the test sample that is predicted to be disomic. In some embodiments, the reference is a different segment from one of the chromosomes of interest in the same sample as the test. For example, the reference may be one or more segments outside the region of the potential deletion or duplication. Having a reference for the same chromosome as the test avoids variations between different chromosomes, such as differences in metabolism, apoptosis, histone, inactivation, and / or amplification between chromosomes. Analyzing segments that do not contain CNVs on the same chromosome as the test can also be used to determine differences in metabolism, apoptosis, histone, inactivation, and / or amplification between chromosomes, allowing the level of variation between homologs that do not contain CNVs to be determined for comparison with the results from potential CNVs. In some embodiments, the magnitude of the difference between the calculated and expected allele ratios for the potential CNV is greater than the corresponding magnitude for the reference, thereby confirming the presence of the CNV.

[0366] In some embodiments, a reference chromosome or chromosome segment is used as a control in which a CNV (e.g., a specific deletion or duplication of interest) is expected to exist. In some embodiments, the reference is the same chromosome or chromosome segment from one or more different samples known or predicted to have a deletion or duplication in the chromosome or chromosome segment. In some embodiments, the reference is a different chromosome or chromosome segment from the test sample known or predicted to have a CNV. In some embodiments, the magnitude of the difference between the calculated and predicted allele ratios for a potential CNV is similar (e.g., not significantly different) to the corresponding magnitude for the reference CNV, thereby confirming the presence of the CNV. In some embodiments, the magnitude of the difference between the calculated and predicted allele ratios for a potential CNV is smaller (e.g., significantly smaller) than the corresponding magnitude for the reference CNV, thereby confirming the absence of the CNV. In some embodiments, the tumor fraction is determined using one or more loci (or DNA or RNA from cancer cells, such as cfDNA or cfRNA) for the genotype of cancer cells that differs from the genotype of non-cancerous cells (or DNA or RNA, e.g., cfDNA or cfRNA, from non-cancerous cells). Tumor fraction can be used to determine whether the overrepresentation of the copy number of a first homologous chromosomal segment is due to duplication of the first homologous chromosomal segment or deletion of the second homologous chromosomal segment.Tumor fraction can also be used to determine the excess copy number of the duplicated chromosomal segment or chromosome (for example, whether there are 1, 2, 3, 4, or more excess copies), and can distinguish, for example, a sample with 4 excess chromosomal copies and a tumor fraction of 10% from a sample with 2 excess chromosomal copies and a tumor fraction of 20%.Tumor fraction can also be used to determine how well observed data matches with predicted data for possible CNVs.In some embodiments, the degree of overrepresentation of CNVs is used to select a specific therapy or treatment regimen for an individual.For example, some therapeutic agents are only effective against at least four, six, or more copies of a chromosomal segment.

[0367] In some embodiments, the one or more loci used to determine tumor fraction are relative to a reference chromosome or chromosome segment (e.g., a chromosome or chromosome segment known or predicted to be disomic, a chromosome or chromosome segment that is rarely duplicated or deleted in cancer cells in general or in particular types of cancer in individuals known to have or at elevated risk of having, or a chromosome or chromosome segment that is unlikely to be aneuploid (e.g., such a segment that, when deleted or duplicated, is predicted to cause cell death). In some embodiments, any of the methods of the invention are used to confirm that the reference chromosome or chromosome segment is disomic in both cancer cells and non-cancerous cells. In some embodiments, one or more chromosomes or chromosome segments with high confidence in calling disomy are used.

[0368] Exemplary loci that can be used to determine tumor fraction include polymorphisms or mutations (e.g., SNPs) in cancer cells (or DNA or RNA, such as cfDNA or cfRNA, from cancer cells) that are not present in non-cancerous cells (or DNA or RNA from non-cancerous cells) in an individual. In some embodiments, tumor fraction is determined by identifying polymorphic loci at which cancer cells (or DNA or RNA from cancer cells) in a sample (e.g., a plasma sample or tumor specimen) from an individual have alleles that are not present in non-cancerous cells (or DNA or RNA from non-cancerous cells), and using the amount of alleles unique to cancer cells at one or more of the identified polymorphic loci to determine the tumor fraction in the sample. In some embodiments, non-cancerous cells are homozygous for a first allele at the polymorphic locus, and cancer cells are (i) heterozygous for the first allele and a second allele, or (ii) homozygous for the second allele at the polymorphic locus. In some embodiments, non-cancerous cells are heterozygous for the first and second alleles at the polymorphic locus, and cancer cells have (i) one or two copies of the third allele at the polymorphic locus. In some embodiments, cancer cells are assumed or known to have only one copy of an allele that is not present in non-cancerous cells. For example, if the genotype of non-cancerous cells is AA, the genotype of cancer cells is AB, and 5% of the signals at that locus in a sample are from the B allele and 95% are from the A allele, the tumor fraction of the sample is 10%. In some embodiments, cancer cells are assumed or known to have two copies of an allele that is not present in non-cancerous cells. For example, if the genotype of non-cancerous cells is AA, the genotype of cancer cells is BB, and 5% of the signals at that locus in a sample are from the B allele and 95% are from the A allele, the tumor fraction of the sample is 5%. In some embodiments, cancer cells analyze multiple loci that have alleles that are not present in non-cancerous cells to determine which loci are heterozygous and which are homozygous in cancer cells. For example, for loci where non-cancerous cells are AA, if the signal from the B allele is about 5% at some loci and about 10% at some loci, then cancer cells are assumed to be heterozygous at loci with about 5% B alleles and homozygous at loci with about 10% B alleles (indicating that the tumor fraction is about 10%).

[0369] Exemplary loci that can be used to determine tumor fraction include loci where cancer cells and non-cancerous cells share one allele in common (e.g., a locus where cancer cells are AB and non-cancerous cells are BB, or where cancer cells are BB and non-cancerous cells are AB). The amount of A signal, the amount of B signal, or the ratio of A signal to B signal in a mixed sample (containing DNA or RNA from cancer cells and non-cancerous cells) is compared with the corresponding value for (i) a sample containing DNA or RNA from only cancer cells or (ii) a sample containing DNA or RNA from only non-cancerous cells. The difference in values ​​is used to determine the tumor fraction of the mixed sample.

[0370] In some embodiments, loci that can be used to determine tumor fraction are selected based on the genotype of (i) a sample containing DNA or RNA from only cancer cells and / or (ii) a sample containing DNA or RNA from only non-cancerous cells. In some embodiments, loci are selected based on the analysis of mixed samples, e.g., loci in which the absolute or relative amount of each allele differs from the amount expected if both cancer cells and cancerous cells have the same genotype at a particular locus are selected. For example, if cancer cells and non-cancerous cells have the same genotype, a locus is predicted to produce a 0% B signal if all cells are AA, a 50% B signal if all cells are AB, or a 100% B signal if all cells are BB. Other values ​​of the B signal indicate that the genotypes of cancer cells and non-cancerous cells differ at that locus, and therefore the locus can be used to determine tumor fraction.

[0371] In some embodiments, the tumor fraction calculated based on alleles at one or more loci is compared to the tumor fraction calculated using one or more of the counting methods disclosed herein.

[0372] Exemplary Methods for Detecting Phenotypes or Analyzing Multiple Mutations In some embodiments, the method involves analyzing a sample for a set of mutations associated with a disease or disorder (e.g., cancer) or an elevated risk of a disease or disorder. There is a strong correlation between events within a class (e.g., M or C cancer classes) that can be used to improve the signal-to-noise ratio of a method and classify tumors into distinct clinical subsets. For example, borderline results for several mutations (e.g., several CNVs) for one or more chromosomes or chromosomal segments considered together can be a very strong signal. In some embodiments, determining the presence or absence of multiple polymorphisms or mutations of interest (e.g., 2, 3, 4, 5, 8, 10, 12, 15, or more) increases the sensitivity and / or specificity of determining the presence or absence of a disease or disorder (e.g., cancer) or elevated risk of a disease or disorder (e.g., cancer). In some embodiments, correlations between events across multiple chromosomes are used to generate a stronger signal compared to looking at each of them individually. The design of the method itself can be optimized to optimally classify tumors. This would be very useful for early detection and screening for recurrence, where sensitivity to one specific mutation / CNV may be most important. In some embodiments, events are not always correlated, but have a probability of being correlated. In some embodiments, a matrix estimation composition is used that has a noise covariance matrix with off-diagonal terms used.

[0373] In some embodiments, the invention features a method for detecting a phenotype (e.g., a cancer phenotype) in an individual, where the phenotype is defined by the presence of at least one of a set of mutations. In some embodiments, the method includes obtaining DNA or RNA measurements from one or more cells from the individual, where one or more cells are suspected of having the phenotype, and analyzing the DNA or RNA measurements to determine, for each mutation in the set of mutations, the likelihood that at least one of the cells has that mutation. In some embodiments, the method includes determining that the individual has the phenotype if (i) for at least one of the mutations, the likelihood that at least one of the cells has that mutation is greater than a threshold, or (ii) for at least one of the mutations, the likelihood that at least one of the cells has that mutation is less than a threshold, and for multiple mutations, the combined likelihood that at least one of the cells has at least one of the mutations is greater than a threshold. In some embodiments, one or more cells have a subset or all of the mutations in the set of mutations. In some embodiments, the subset of mutations is associated with cancer or an elevated risk of cancer. In some embodiments, the set of mutations includes a subset or all of the mutations in the M class of cancer mutations (Ciriello, Nat Genet. 45(10):1127-1133, 2013, doi:10.1038 / ng.2762, incorporated herein by reference in its entirety). In some embodiments, the set of mutations includes a subset or all of the mutations in the C class of cancer mutations (Ciriello, supra). In some embodiments, the sample comprises cell-free DNA or RNA. In some embodiments, the DNA or RNA measurements include measurements at a set of polymorphic loci (e.g., the dosage of each allele at each locus) on one or more chromosomes or chromosomal segments of interest.

[0374] Exemplary Combinations of Methods To improve the accuracy of the results, two or more methods (e.g., any of the methods of the present invention or any known method) for detecting the presence or absence of CNV are performed. In some embodiments, one or more methods (e.g., any of the methods of the present invention or any known method) for analyzing factors that are indicative of the presence or absence of a disease or disorder or an increased risk of a disease or disorder are performed.

[0375] In some embodiments, standard mathematical techniques are used to calculate the covariance and / or correlation between two or more methods. Standard mathematical techniques may also be used to determine the joint probability of a particular hypothesis based on two or more tests. Exemplary techniques include meta-analysis, Fisher's joint probability test for independent tests, Brown's method for combining dependent p-values ​​with known covariances, and cost methods for combining dependent p-values ​​with unknown covariances. In cases where the likelihood is determined by a first method in an orthogonal or independent manner to the way the likelihood is determined for a second method, combining the likelihoods is straightforward and can be done by multiplication and normalization, or by using a formula such as:

[0376] R 結合 =R1R2 / [R1R2+(1-R1)(1-R2)]

[0377] R 結合 is the joint likelihood, and R1 and R2 are the individual likelihoods. For example, if the likelihood of trisomy from Method 1 is 90% and the likelihood of trisomy from Method 2 is 95%, by combining the outputs from the two methods, a clinician can conclude that the fetus is trisomic with a likelihood of (0.90)(0.95) / [(0.90)(0.95)+(1-0.90)(1-0.95)]=99.42%. The likelihoods can also be combined when the first and second methods are not orthogonal, i.e., when there is a correlation between the two methods.

[0378] Exemplary methods for analyzing multiple factors or variables are disclosed in U.S. Patent No. 8,024,128, issued September 20, 2011, U.S. Publication No. 2007 / 0027636, filed July 31, 2006, and U.S. Publication No. 2007 / 0178501, filed December 6, 2006, each of which is incorporated herein by reference.

[0379] In various embodiments, the joint probability of a particular hypothesis or diagnosis is greater than 80, 85, 90, 92, 94, 96, 98, 99, or 99.9%, or greater than some other threshold.

[0380] Detection limit As shown by the experiments provided in the Examples section, the methods provided herein can detect average allelic imbalance in a sample with a detection or sensitivity limit of 0.45% AAI (which is the detection limit of aneuploidy for the exemplary method of the present invention).Similarly, in certain embodiments, the methods provided herein can detect average allelic imbalance in a sample of 0.45, 0.5, 0.6, 0.8, 0.8, 0.9 or 1.0%.That is, the test method can detect chromosomal aneuploidy in a sample with an AAI of as low as 0.45, 0.5, 0.6, 0.8, 0.8, 0.9 or 1.0%.As shown by the experiments provided in the Examples section, the methods provided herein can detect the presence of SNVs in a sample for at least some SNVs, with a detection or sensitivity limit of 0.2%, which is the detection limit for at least some SNVs in an exemplary embodiment. Similarly, in certain embodiments, the methods are capable of detecting SNVs at a frequency or SNV AAI of 0.2, 0.3, 0.4, 0.5, 0.6, 0.8, 0.8, 0.9, or 1.0%, i.e., the test methods are capable of detecting SNVs in samples down to a detection limit of 0.2, 0.3, 0.4, 0.5, 0.6, 0.8, 0.8, 0.9, or 1.0% of the total allele count at the chromosomal locus of the SNV.

[0381] In some embodiments, the detection limit of mutations (e.g., SNVs or CNVs) of the methods of the present invention is less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01, or 0.005%. In some embodiments, the detection limit of mutations (e.g., SNVs or CNVs) of the methods of the present invention is 15 to 0.005%, for example, 10 to 0.005%, 10 to 0.01%, 10 to 0.1%, 5 to 0.005%, 5 to 0.01%, 5 to 0.1%, 1 to 0.005%, 1 to 0.01%, 1 to 0.1%, 0.5 to 0.005%, 0.5 to 0.01%, 0.5 to 0.1%, or 0.1 to 0.01 (inclusive).

[0382] In some embodiments, detection limit is the value that can be detected (or can be detected) that the mutation (for example, SNV or CNV) present in the DNA or RNA that comprises locus in sample (for example, cfDNA or cfRNA sample) is less than or equal to 10,5,2,1,0.5,0.1,0.05,0.01 or 0.005%.For example, mutation can be detected when the DNA or RNA molecule that comprises the locus that has mutation in the locus (for example, instead of the wild-type or non-mutated form of the locus or a different mutation at the locus) is less than or equal to 10,5,2,1,0.5,0.1,0.05,0.01 or 0.005%. In some embodiments, the detection limit is the value that can detect (or be detected) mutations (such as SNVs or CNVs) that exist in less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of the DNA or RNA molecules in sample (such as cfDNA or cfRNA sample).In some embodiments where CNV is deletion, deletion can be detected even if it exists in less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of the DNA or RNA molecules with target region in sample, which may or may not contain deletion.In some embodiments where CNV is deletion, deletion can be detected even if it exists in less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of the DNA or RNA molecules in sample. In some embodiments where the CNV is a duplication, the duplication can be detected even if the excess duplicated RNA or DNA present is present in an amount less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of the DNA or RNA molecules having the region of interest, which may or may not be duplicated in the sample.In some embodiments where the CNV is a duplication, the duplication can be detected even if the excess duplicated RNA or DNA present is only present in an amount less than or equal to 10, 5, 2, 1, 0.5, 0.1, 0.05, 0.01 or 0.005% of the DNA or RNA molecules in the sample.

[0383] Illustrative Sample In some embodiments of any of the aspects of the invention, the sample contains intracellular and / or extracellular genetic material from cells suspected of having a deletion or duplication, e.g., cells suspected of being cancerous. In some embodiments, the sample includes any tissue or bodily fluid (e.g., tumor) suspected of containing cells with a deletion or duplication, DNA, or RNA, or other samples containing cancer cells, DNA, or RNA. The genetic measurements used as part of these methods may be performed on any sample containing DNA or RNA, including, but not limited to, tissue, blood, serum, plasma, urine, hair, tears, saliva, skin, fingernails, feces, bile, lymph, cervical mucus, semen, tumor, or other cells or substances containing nucleic acids. The sample may include any cell type, or DNA or RNA from any cell type may be used (e.g., cells from any organ or tissue suspected of being cancerous, or neurons). In some embodiments, the sample contains nuclear and / or mitochondrial DNA. In some embodiments, the sample is derived from any of the target individuals disclosed herein. In some embodiments, the target individual is a cancer patient.

[0384] Exemplary samples include those containing cfDNA or cfRNA.In some embodiments, cfDNA can be used for analysis without the need to dissolve cells.Cell-free DNA can be obtained from various tissues, for example, tissues in liquid form, such as blood, plasma, lymph, ascites or cerebrospinal fluid.In some cases, cfDNA is composed of DNA derived from fetal cells.In some cases, cfDNA is isolated from plasma separated from whole blood, which is centrifuged to remove cellular material.cfDNA can also be a mixture of DNA derived from target cells (for example, cancer cells) and non-target cells (for example, non-cancer cells).

[0385] In some embodiments, the sample contains or is suspected to contain a mixture of DNA (or RNA), for example, a mixture of DNA (or RNA) derived from cancer cells and DNA (or RNA) derived from non-cancerous (i.e., normal) cells. In some embodiments, at least 0.5, 1, 3, 5, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99, or 100% of the cells in the sample are cancer cells. In some embodiments, at least 0.5, 1, 3, 5, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99, or 100% of the DNA (e.g., cfDNA) or RNA (e.g., cfRNA) in the sample is derived from cancer cell(s). In various embodiments, the percentage of cells in a sample that are cancerous is 0.5-99%, e.g., 1-95%, 5-95%, 10-90%, 5-70%, 10-70%, 20-90%, or 20-70%, inclusive. In some embodiments, the sample is enriched for cancer cells or enriched for DNA or RNA from cancer cells. In some embodiments of samples enriched for cancer cells, at least 0.5, 1, 2, 3, 4, 5, 6, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99, or 100% of the cells in the enriched sample are cancer cells. In some embodiments of a sample enriched in DNA or RNA from cancer cells, at least 0.5, 1, 2, 3, 4, 5, 6, 7, 10, 15, 20, 30, 40, 50, 60, 70, 80, 90, 92, 94, 95, 96, 98, 99 or 100% of the DNA or RNA in the enriched sample is derived from cancer cell(s).In some embodiments, cell sorting (e.g., fluorescence-activated cell sorting (FACS)) is used to enrich for cancer cells (Barteneva et al., Biochim Biophys Acta., 1836(1):105-22, August 2013 doi:10.1016 / j.bbcan.2013.02.004. Epub 2013 Feb 24, and Ibrahim et al., Adv Biochem Eng Biotechnol. 106:19-39, 2007, each of which is incorporated herein by reference in its entirety).

[0386] In some embodiments, the sample is enriched for fetal cells. In some embodiments of samples enriched for fetal cells, at least 0.5, 1, 2, 3, 4, 5, 6, 7% or more of the cells in the enriched sample are fetal cells. In some embodiments, the percentage of cells in the sample that are fetal cells is between 0.5 and 100%, e.g., between 1 and 99%, 5 and 95%, 10 and 95%, 10 and 95%, 20 and 90%, or 30 and 70%, inclusive. In some embodiments, the sample is enriched for fetal DNA. In some embodiments of samples enriched for fetal DNA, at least 0.5, 1, 2, 3, 4, 5, 6, 7% or more of the DNA in the enriched sample is fetal DNA. In some embodiments, the percentage of DNA in the sample that is fetal DNA is between 0.5 and 100%, for example, between 1 and 99%, 5 and 95%, 10 and 95%, 10 and 95%, 20 and 90%, or 30 and 70% (boundaries inclusive).

[0387] In some embodiments, the sample comprises a single cell or comprises DNA and / or RNA from a single cell. In some embodiments, multiple individual cells (e.g., at least 5, 10, 20, 30, 40, or 50 cells from the same or different subjects) are analyzed in parallel. In some embodiments, cells from multiple samples from the same individual are combined, reducing the amount of work compared to analyzing these samples separately. Combining multiple samples also allows multiple tissues to be tested for cancer simultaneously (which can be used to provide more thorough screening for cancer or to determine whether cancer may have metastasized to other tissues).

[0388] In some embodiments, the sample contains a single cell or a small number of cells, e.g., 2, 3, 5, 6, 7, 8, 9, or 10 cells. In some embodiments, the sample contains 1 to 100, 100 to 500, or 500 to 1,000 cells (inclusive). In some embodiments, the sample contains 1 to 10 picograms, 10 to 100 picograms, 100 picograms to 1 nanogram, 1 to 10 nanograms, 10 to 100 nanograms, or 100 nanograms to 1 microgram of RNA and / or DNA (inclusive).

[0389] In some embodiments, the sample is embedded in parafilm. In some embodiments, the sample is preserved with a preservative such as formaldehyde and optionally embedded in paraffin, which may cause DNA cross-linking so that a small amount is available for PCR. In some embodiments, the sample is a formaldehyde-fixed, paraffin-embedded (FFPE) sample. In some embodiments, the sample is a fresh sample (e.g., a sample obtained within one or two days of analysis). In some embodiments, the sample is frozen prior to analysis. In some embodiments, the sample is a historical sample.

[0390] These samples can be used in any of the methods of the present invention.

[0391] Exemplary Sample Preparation Methods In some embodiments, the method includes isolating or purifying DNA and / or RNA. There are several standard procedures known in the art to achieve this goal. In some embodiments, the sample may be centrifuged to separate the various layers. In some embodiments, DNA or RNA may be isolated using filtration. In some embodiments, DNA or RNA preparation may involve amplification, separation, chromatographic purification, liquid-liquid separation, isolation, preferential enrichment, preferential amplification, targeted amplification, or any of several other techniques known in the art or described herein. In some embodiments for DNA isolation, RNase is used to degrade RNA. In some embodiments for RNA isolation, DNase (e.g., DNase I from Invitrogen, Carlsbad, CA, USA) is used to degrade DNA. In some embodiments, an RNeasy Mini Kit (Qiagen) is used to isolate RNA according to the manufacturer's protocol. In some embodiments, small RNAs are isolated using the mirVana PARIS kit (Ambion, Austin, TX, USA) according to the manufacturer's protocol (Gu et al., J.Neurochem.122:641-649, 2012, incorporated herein by reference in its entirety). RNA concentration and purity can optionally be determined using Nanovue (GE Healthcare, Piscataway, NJ, USA), and RNA integrity can optionally be measured using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA) (Gu et al., J.Neurochem.122:641-649, 2012, incorporated herein by reference in its entirety). In some embodiments, TRIZOL or RNAlater (Ambion) is used to stabilize RNA during storage.

[0392] In some embodiments, universal tagging adapters are added to create the library. Prior to ligation, the sample DNA may be blunt-ended, and then a single adenosine base is added to the 3' end. Prior to ligation, the DNA may be cleaved using a restriction enzyme or some other cleavage method. During ligation, the 3' adenosine of the sample fragment and the complementary 3' tyrosine overhang of the adapter can increase ligation efficiency. In some embodiments, adapter ligation is performed using a ligation kit such as that found in the Agilent SureSelect kit. In some embodiments, the library is amplified using universal primers. In one embodiment, the amplified library is fractionated by size separation or by using products such as Agencourt Ampure beads or other similar methods. In some embodiments, PCR amplification is used to amplify the target loci. In some embodiments, the amplified DNA is sequenced (e.g., using an Illumina IIGAX or HiSeq sequencer). In some embodiments, the amplified DNA is sequenced from each end of the amplified DNA to reduce sequencing errors. If there is a sequence error at a particular base when sequencing from one end of the amplified DNA, there is less likely to be a sequence error in the complementary base when sequencing from the other end of the amplified DNA (compared to multiple sequencing from the same end of the amplified DNA).

[0393] In some embodiments, whole genome applications (WGA) are used to amplify nucleic acid samples. Several methods are available for WGA, including ligation-mediated PCR (LM-PCR), degenerate oligonucleotide primer PCR (DOP-PCR), and multiple displacement amplification (MDA). In LM-PCR, short DNA sequences called adapters are ligated to blunt ends of DNA. These adapters contain universal amplification sequences that are used to amplify DNA by PCR. In DOP-PCR, random primers that also contain universal amplification sequences are used in the first round of annealing and PCR. A second round of PCR is then used to further amplify the sequences using the universal primer sequences. MDA uses phi-29 polymerase, a highly processive, non-specific enzyme that replicates DNA and has been used in single-cell analysis. In some embodiments, WGA is not performed.

[0394] In some embodiments, selective amplification or enrichment is used to amplify or enrich target loci. In some embodiments, amplification and / or selective enrichment techniques may involve PCR (e.g., ligation-mediated PCR), hybridization fraction capture, molecular inversion probes, or other circularization probes. In some embodiments, real-time quantitative PCR (RT-qPCR), digital PCR, or emulsion PCR, single-allele base extension reactions followed by mass spectrometry are used (Hung et al., J Clin Pathol 62:308-313, 2009, incorporated herein by reference in its entirety). In some embodiments, hybridization capture using hybrid capture probes is used to preferentially enrich DNA. In some embodiments, methods for amplification or selective enrichment may involve using probes such that, upon correct hybridization to the target sequence, the 3' or 5' end of the nucleotide probe is separated from the polymorphic site of the polymorphic allele by a small number of nucleotides. This separation reduces preferential amplification of one allele, referred to as allelic bias. This is an improvement over methods involving the use of probes in which the 3' or 5' end of a correctly hybridized probe is directly adjacent to or very close to the polymorphic site of an allele. In one embodiment, probes whose hybridizing region may or certainly contains a polymorphic site are excluded. Polymorphic sites at the hybridization site may cause unequal hybridization in some alleles or completely inhibit hybridization, resulting in preferential amplification of specific alleles. These embodiments are an improvement over other methods involving targeted amplification and / or selective enrichment, in that they effectively preserve the original allele frequency of the sample at each polymorphic locus, where the sample is a pure genomic sample from a single individual or a mixture of individuals.

[0395] In some embodiments, PCR (called mini-PCR) is used to generate very short amplicons (U.S. Application No. 13 / 683,604, filed November 21, 2012; U.S. Publication No. 2013 / 0123120; U.S. Application No. 13 / 300,235, filed November 18, 2011; U.S. Publication No. 2012 / 0270212, filed November 18, 2011; and U.S. Application No. 61 / 994,791, filed May 16, 2014, each of which is incorporated by reference in its entirety). cfDNA (e.g., cancer cfDNA released by necrosis or apoptosis) is highly fragmented. In fetal cfDNA, fragment sizes are distributed in an approximately Gaussian manner, with a mean of 160 bp, a standard deviation of 15 bp, a minimum size of approximately 100 bp, and a maximum size of approximately 220 bp. A polymorphic site at a particular target locus may occupy any position from the beginning to the end of various fragments derived from that locus. Because cfDNA fragments are short, the likelihood that both primer sites are present—that is, the likelihood that a fragment of length L contains both forward and reverse primer sites—is the ratio of the amplicon length to the length of the fragment. Under ideal conditions, assays with amplicons of 45, 50, 55, 60, 65, or 70 bp successfully amplify 72%, 69%, 66%, 63%, 59%, or 56% of the available template fragment molecules, respectively. In certain embodiments most relevant to cfDNA from samples of individuals suspected of having cancer, cfDNA is amplified using primers with a melting temperature of 50-65°C, and in certain preferred embodiments, 54-60.5°C, yielding a maximum amplicon length of 85, 80, 75, or 70 bp, and in certain preferred embodiments, 75 bp. The amplicon length is the distance between the 5' ends of the forward and reverse priming sites. Amplicon lengths shorter than those typically used by those known in the art may result in more efficient measurement of desired polymorphic loci by requiring only short sequence reads. In one embodiment, a substantial fraction of the amplicons are less than 100 bp, less than 90 bp, less than 80 bp, less than 70 bp, less than 65 bp, less than 60 bp, less than 55 bp, less than 50 bp, or less than 45 bp.

[0396] In some embodiments, the amplification is performed using direct multiplex PCR, sequential PCR, nested PCR, double nested PCR, one-and-a-half PCR, or any combination thereof. Mini-PCR may be performed using one-sided nested PCR, fully nested PCR, one-sided fully nested PCR, one-sided nested PCR, heminested PCR, heminested PCR, triplex heminested PCR, semi-nested PCR, one-sided semi-nested PCR, reverse semi-nested PCR, or one-sided PCR, as described in U.S. Application No. 13 / 683,604, filed November 21, 2012, U.S. Publication No. 2013 / 0123120, U.S. Application No. 13 / 300,235, filed November 18, 2011, U.S. Publication No. 2012 / 0270212, and U.S. Application No. 61 / 994,791, filed May 16, 2014, which are incorporated by reference in their entireties. Any of these methods may be used for mini-PCR, if desired.

[0397] If desired, the extension step of the PCR amplification may be limited in time to reduce amplification from fragments longer than 200, 300, 400, 500, or 1,000 nucleotides, which may result in enrichment of fragmented or shorter DNA (e.g., fetal DNA, or DNA from cancer cells undergoing apoptosis or necrosis), improving test performance.

[0398] In some embodiments, multiplex PCR is used. In some embodiments, a method for amplifying target loci in a nucleic acid sample involves (i) contacting a nucleic acid sample with a library of primers that simultaneously hybridize at least 100, 200, 500, 750, 1,000, 2,000, 5,000, 7,500, 10,000, 20,000, 25,000, 30,000, 40,000, 50,000, 75,000, or 100,000 different target loci to generate a reaction mixture; and (ii) subjecting the reaction mixture to primer extension reaction conditions (e.g., PCR conditions) to generate amplification products containing target amplicons. In some embodiments, at least 50, 60, 70, 80, 90, 95, 96, 97, 98, 99, or 99.5% of the target loci are amplified. In various embodiments, less than 60, 50, 40, 30, 20, 10, 5, 4, 3, 2, 1, 0.5, 0.25, 0.1, or 0.05% of the amplification products are primer-dimers. In some embodiments, the primers are in solution (e.g., dissolved in a liquid phase rather than a solid phase). In some embodiments, the primers are in solution and not immobilized on a solid support. In some embodiments, the primers are not part of a microarray. In some embodiments, the primers do not comprise a molecular inversion probe (MIP).

[0399] In some embodiments, two or more (e.g., three or four) target amplicons (e.g., amplicons from the mini-PCR methods disclosed herein) are ligated together, and the ligated product is then sequenced. Combining multiple amplicons into a single ligation product increases the efficiency of the subsequent sequencing step. In some embodiments, the target amplicons are less than 150, 100, 90, 75, or 50 base pairs in length before they are ligated. Selective enrichment and / or amplification may involve tagging each individual molecule with a different tag, molecular barcode, tag for amplification, and / or tag for sequencing. In some embodiments, the amplification products are analyzed by sequencing (e.g., high-throughput sequencing) or by hybridization to an array, such as a SNP array, ILLUMINA INFINIUM array, or AFFYMETRIX gene chip. In some embodiments, nanopore sequencing is used, for example, the nanopore sequencing technology developed by Genia (see, for example, the World Wide Web at geniachip.com / technology, which is incorporated herein by reference in its entirety). In some embodiments, double sequencing is used (Schmitt et al., "Detection of ultra-rare mutations by next-generation sequencing," Proc Natl Acad Sci US A. 109(36):14508-14513, 2012, which is incorporated herein by reference in its entirety). This approach significantly reduces errors by independently tagging and sequencing each of the two strands of a DNA duplex. Because the two strands are complementary, true mutations are found at the same position in both strands. In contrast, PCR or sequencing errors can be discounted as technical errors because they only result in mutations in one strand. In some embodiments, the method involves tagging both strands of double-stranded DNA with random but complementary double-stranded nucleotide sequences (called double-stranded tags).First, a single-stranded randomized nucleotide sequence is introduced into one adapter strand, and then the opposite strand is extended using DNA polymerase to obtain a complementary double-stranded tag, thereby incorporating the double-stranded tag sequence into a standard sequencing adapter. After ligating the tagged adapter to sheared DNA, the individually labeled strands are PCR amplified from the asymmetric primer sites on the adapter tail and subjected to paired-end sequencing. In some embodiments, a sample (e.g., a DNA or RNA sample) is divided into multiple fractions, for example, into different wells (e.g., wells of a WaferGen SmartChip). By dividing the sample into different fractions (e.g., at least 5, 10, 20, 50, 75, 100, 150, 200, or 300 fractions), the proportion of molecules with mutations is higher in some wells than in the overall sample, thereby increasing the sensitivity of the analysis. In some embodiments, each fraction contains less than 500, 400, 200, 100, 50, 20, 10, 5, 2, or 1 DNA or RNA molecule. In some embodiments, the molecules in each fraction are sequenced separately. In some embodime...

Claims

1. 1. A method for monitoring and detecting early recurrence or metastasis of breast, bladder or colon cancer, comprising: selecting a set of at least 8 or 16 patient-specific single nucleotide variant loci based on somatic mutations identified in tumor samples from patients diagnosed with breast, bladder or colon cancer; collecting one or more blood or urine samples longitudinally from the patient after the patient has been treated with surgery, first line chemotherapy and / or adjuvant therapy; generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon in the set of amplicons spans at least one single nucleotide variant locus in the set of patient-specific single nucleotide variant loci associated with breast, bladder or colon cancer; determining a sequence of at least one segment of each amplicon of the set of amplicons that comprises a patient-specific single nucleotide variant locus, wherein detection of one or more patient-specific single nucleotide variants from the blood or urine sample is indicative of early recurrence or metastasis of breast, bladder or colon cancer.

2. 1. A method for treating breast cancer, bladder cancer or colon cancer, comprising: treating a patient diagnosed with breast cancer, bladder cancer, or colon cancer with surgery, first-line chemotherapy, and / or adjuvant therapy; collecting one or more blood or urine samples longitudinally from said patient; generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon in the set of amplicons spans at least one single nucleotide variant locus of a set of at least 8 or 16 patient-specific single nucleotide variant loci associated with said breast, bladder or colon cancer and selected based on somatic mutations identified in a tumor sample of said patient; determining a sequence of at least one segment of each amplicon of the set of amplicons that comprises a patient-specific single nucleotide variant locus, wherein detection of one or more patient-specific single nucleotide variants from the blood or urine sample is indicative of early recurrence or metastasis of breast, bladder or colon cancer; and administering to said patient a compound known to be effective in treating breast, bladder or colon cancer having one or more of said single nucleotide variants detected in said blood or urine sample.

3. 1. A method for monitoring or predicting response to treatment of breast cancer, bladder cancer or colon cancer, comprising: collecting one or more blood or urine samples longitudinally from a patient undergoing treatment for breast cancer, bladder cancer, or colon cancer; generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from each blood or urine sample or fraction thereof, wherein each amplicon in the set of amplicons spans at least one single nucleotide variant locus of a set of at least 8 or 16 patient-specific single nucleotide variant loci associated with said breast, bladder or colon cancer and selected based on somatic mutations identified in a tumor sample of said patient; determining a sequence of at least one segment of each amplicon of the set of amplicons that comprises a patient-specific single nucleotide variant locus, wherein detection of one or more patient-specific single nucleotide variants from the blood or urine sample is indicative of a poor response to the breast, bladder or colon cancer treatment.

4. The method according to any one of claims 1 to 3, wherein the cancer is breast cancer.

5. The method according to any one of claims 1 to 3, wherein the cancer is bladder cancer.

6. The method according to any one of claims 1 to 3, wherein the cancer is colon cancer.

7. 4. The method of claim 3, wherein the treatment is neoadjuvant therapy.

8. The method of claim 3, wherein the treatment is an adjuvant therapy.

9. 4. The method of claim 3, further comprising administering to the patient a compound known to be effective in treating breast, bladder or colon cancer having one or more of the single nucleotide variants detected in the blood or urine sample.

10. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in at least 85% of patients with early recurrence or metastasis of cancer.

11. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in at least 95% of patients with early recurrence or metastasis of HER2+ breast cancer.

12. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in at least 95% of patients with early recurrence or metastasis of triple-negative breast cancer.

13. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in at least 80% of patients with early recurrence or metastasis of HR+ / HER2- breast cancer.

14. 4. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in patients with early recurrence or metastasis of cancer at least 200 days prior to clinical recurrence or metastasis of cancer detectable by imaging and / or at least 100 days prior to elevated CA15-3 levels.

15. 4. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in patients with early recurrence or metastasis of HER2+ breast cancer at least 150 days prior to clinical recurrence or metastasis of HER2+ breast cancer detectable by imaging, and / or at least 100 days prior to elevated CA15-3 levels.

16. 4. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in patients with early recurrence or metastasis of triple-negative breast cancer at least 200 days prior to clinical recurrence or metastasis of triple-negative breast cancer detectable by imaging and / or at least 100 days prior to elevated CA15-3 levels.

17. 4. The method of any one of claims 1 to 3, wherein the method detects patient-specific single nucleotide variants in patients with early recurrence or metastasis of HR+ / HER2- breast cancer at least 250 days prior to clinical recurrence or metastasis of HR+ / HER2- breast cancer detectable by imaging, and / or at least 100 days prior to elevated CA15-3 levels.

18. The method of any one of claims 1 to 3, wherein the method does not detect patient-specific single nucleotide variants in at least 99% of patients who do not have early recurrence or metastasis of cancer.

19. The method of any one of claims 1 to 3, wherein the method does not detect patient-specific single nucleotide variants in at least 99% of patients free of early recurrence or metastasis of HER2+ breast cancer.

20. The method of any one of claims 1 to 3, wherein the method does not detect patient-specific single nucleotide variants in at least 99% of patients who do not have early recurrence or metastasis of triple-negative breast cancer.

21. The method of any one of claims 1 to 3, wherein said method does not detect patient-specific single nucleotide variants in at least 99% of patients free of early recurrence or metastasis of HR+ / HER2- breast cancer.

22. 4. The method of any one of claims 1 to 3, wherein the method has a specificity of at least 99.5% in detecting early recurrence or metastasis of cancer when two or more patient-specific single nucleotide variants are detected above a confidence threshold of 0.

97.

23. 4. The method of any one of claims 1-3, wherein the method has a specificity of at least 99.5% in detecting early recurrence or metastasis of HER2+ breast cancer when two or more patient-specific single nucleotide variants are detected above a confidence threshold of 0.

97.

24. 4. The method of any one of claims 1 to 3, wherein the method has a specificity of at least 99.5% in detecting early recurrence or metastasis of triple negative breast cancer when two or more patient-specific single nucleotide variants are detected above a confidence threshold of 0.

97.

25. 4. The method of any one of claims 1 to 3, wherein the method has a specificity of at least 99.5% in detecting early recurrence or metastasis of HR+ / HER2- breast cancer when two or more patient-specific single nucleotide variants are detected above a confidence threshold of 0.

97.

26. 1. A method for monitoring and detecting early recurrence or metastasis of breast, bladder or colon cancer, comprising: generating a set of amplicons by performing a multiplex amplification reaction on nucleic acids isolated from a blood or urine sample, or a fraction thereof, from a patient treated for breast, bladder or colon cancer, wherein each amplicon in the set of amplicons spans at least one single nucleotide variant locus in a set of patient-specific single nucleotide variant loci associated with said breast, bladder or colon cancer; determining a sequence of at least one segment of each amplicon of the set of amplicons that comprises a patient-specific single nucleotide variant locus, wherein detection of one or more patient-specific single nucleotide variants is indicative of early recurrence or metastasis of breast, bladder or colon cancer.

27. 27. The method of claim 26, wherein prior to determining the sequence of at least one segment of each amplicon of the set of amplicons for the blood or urine sample or fraction thereof, nucleic acid is isolated from the patient's tumor, somatic mutations are identified in the tumor for the set of patient-specific single nucleotide variant loci, and the single nucleotide variants.

28. 27. The method of claim 26, wherein the method comprises longitudinally collecting and sequencing blood or urine samples from the patient.

29. 27. The method of claim 26, wherein at least two or at least five SNVs are detected, and the presence of said at least two or at least five SNVs is indicative of early recurrence or metastasis of breast cancer, bladder cancer or colon cancer.

30. 27. The method of claim 26, wherein the breast, bladder or colon cancer is stage 1, 1b or 2a breast, bladder or colon cancer.

31. 27. The method of claim 26, wherein the individual has been treated with surgery prior to isolation of the blood or urine sample.

32. 27. The method of claim 26, wherein the individual has been treated with chemotherapy prior to isolation of the blood or urine sample.

33. 27. The method of claim 26, wherein the individual has been treated with radiation therapy prior to isolation of the blood or urine sample.

34. 27. The method of claim 26, wherein the individual has been treated with adjuvant therapy prior to isolation of the blood or urine sample.

35. 27. The method of claim 26, wherein the cancer is breast cancer.

36. 27. The method of claim 26, wherein the cancer is bladder cancer.

37. 27. The method of claim 26, wherein the cancer is colon cancer.

38. 27. The method of claim 26, further comprising administering to the individual a compound, said compound being known to be specifically effective in treating breast, bladder or colon cancer having one or more of said determined single nucleotide variants.

39. 27. The method of claim 26, wherein the method further comprises determining a variant allele frequency for each of the single nucleotide variants from the sequencing.

40. 40. The method of claim 39, wherein a treatment regimen for breast, bladder or colon cancer is identified based on said determining variant allele frequencies.

41. 40. The method of claim 39, further comprising administering to the individual a compound, wherein the compound is known to be specifically effective in treating breast, bladder or colon cancer in which one of the single nucleotide variants has a variable allele frequency at least half of the other determined single nucleotide variants.

42. 27. The method of claim 26, wherein the sequences are determined by high throughput DNA sequencing of a plurality of the single nucleotide variant loci.

43. 27. The method of claim 26, further comprising detecting a clonal single nucleotide variant in the breast cancer, bladder cancer or colon cancer by determining a variant allele frequency for each SNV locus based on sequences of multiple copies of a set of amplicons, wherein a relatively high allele frequency of the plurality of single nucleotide variant loci compared to the other single nucleotide variants is indicative of a clonal single nucleotide variant in the breast cancer, bladder cancer or colon cancer.

44. 44. The method of claim 43, further comprising administering to said individual a compound that targets said one or more clonal single nucleotide variants but not said other single nucleotide variants.

45. 44. The method of claim 43, wherein a variant allele frequency of greater than 1.0% is indicative of a clonal single nucleotide variant.

46. 27. The method of claim 26, further comprising forming an amplification reaction mixture by combining a polymerase, nucleotide triphosphates, nucleic acid fragments from a nucleic acid library generated from said sample, and a set of primers that each bind within 150 base pairs of a single nucleotide variant locus or a set of primer pairs that each span a region of 160 base pairs or less that includes a single nucleotide variant locus, and subjecting said amplification reaction mixture to amplification conditions to generate said set of amplicons.

47. A composition comprising a circulating tumor nucleic acid fragment comprising a universal adaptor, wherein the circulating tumor nucleic acid was derived from a breast cancer, a bladder cancer, or a colon cancer.

48. 48. The composition of claim 47, wherein the circulating tumor nucleic acid was derived from a blood or urine sample, or a fragment thereof, of an individual with breast cancer, bladder cancer, or colon cancer.

49. A composition comprising a solid support comprising a clonal population of a plurality of nucleic acids, said clonal population comprising amplicons generated from a sample of circulating free nucleic acid, wherein said circulating tumor nucleic acid was derived from breast cancer, bladder cancer or colon cancer.

50. 50. The composition of claim 49, wherein the circulating free nucleic acid was derived from a blood or urine sample, or a fragment thereof, of an individual with breast cancer, bladder cancer, or colon cancer.

51. The composition of any one of claims 47 to 50, wherein the nucleic acid fragments in different clonal populations comprise the same universal adaptors.

52. 52. The composition of claim 51, wherein the clonal population of nucleic acids is derived from nucleic acid fragments from a set of samples from two or more individuals.

53. 53. The composition of Claim 52, wherein the nucleic acid fragment comprises one of a set of molecular barcodes corresponding to a sample in the set of samples.

54. 47. The method of any one of claims 26-46, wherein determining whether a single nucleotide variant is present in the sample comprises determining a confidence value for each allele call at each of the set of single nucleotide variant loci based at least in part on read depth for the loci.

55. 47. The method of any one of claims 26 to 46, wherein a single nucleotide variant call is made if the confidence value for the occurrence of a single nucleotide variant is greater than 90%.

56. 47. The method of any one of claims 26 to 46, wherein a single nucleotide variant call is made if the confidence value for the occurrence of a single nucleotide variant is greater than 95%.

57. 47. The method of any one of claims 26-46, wherein the set of single nucleotide variant loci comprises all of the single nucleotide variant loci identified in TCGA and COSMIC datasets for breast cancer, bladder cancer or colorectal cancer.

58. 47. The method of any one of claims 26 to 46, wherein the set of single nucleotide divergent sites comprises all of the single nucleotide divergent sites identified in the TCGA and COSMIC datasets for breast cancer, bladder cancer or colorectal cancer.

59. 47. The method of any one of claims 26-46, wherein the method is performed with a read depth for the set of at least 1,000 of said single nucleotide variant loci.

60. 47. The method of any one of claims 26-46, wherein the set of single nucleotide variant loci comprises between 25 and 1000 single nucleotide variant loci known to be associated with breast cancer, bladder cancer or colon cancer.

61. 47. The method of any one of claims 26-46, wherein a per cycle efficiency and error rate is determined for each amplification reaction of the multiplex amplification reaction of the single nucleotide variant loci, and the efficiency and error rate are used to determine whether single nucleotide variants at the set of single variant loci are present in the sample.

62. The method according to any one of claims 26 to 46, wherein the amplification reaction is a PCR reaction and the annealing temperature is 1 to 15°C higher than the melting temperature of at least 50% of the primers of the set of primers.

63. The method according to any one of claims 26 to 46, wherein the amplification reaction is a PCR reaction and the length of the annealing step during the PCR reaction is between 15 and 120 minutes.

64. The method according to any one of claims 26 to 46, wherein the amplification reaction is a PCR reaction and the length of the annealing step during the PCR reaction is between 15 and 120 minutes.

65. The method according to any one of claims 26 to 46, wherein the primer concentration in the amplification reaction is 1 to 10 nM.

66. The method of any one of claims 26 to 46, wherein the primers in the set of primers are designed to minimize primer-dimer formation.

67. 47. The method of any one of claims 26 to 46, wherein the amplification reaction is a PCR reaction, the annealing temperature is 1-15°C higher than the melting temperature of at least 50% of the primers in the set of primers, the length of the annealing step in the PCR reaction is 15-120 minutes, the primer concentration in the amplification reaction is 1-10 nM, and the primers in the set of primers are designed to minimize primer-dimer formation.

68. The method of any one of claims 26 to 46, wherein the multiplex amplification reaction is performed under limiting primer conditions.

69. 47. The method of any one of claims 44 to 46, wherein the method detects patient-specific single nucleotide variants in at least 90% of patients with early recurrence or metastasis of muscle invasive bladder cancer (MIBC).

70. 47. The method of any one of claims 44 to 46, wherein the method detects patient-specific single nucleotide variants in patients with early recurrence or metastasis of MIBC at least 100 days prior to clinical recurrence or metastasis of MIBC detectable by imaging.

71. 47. The method of any one of claims 44 to 46, wherein said method does not detect patient-specific single nucleotide variants in at least 95% of patients free of early recurrence or metastasis of MIBC.

72. 47. The method of any one of claims 44 to 46, wherein the method has a specificity of at least 99.5% in detecting early recurrence or metastasis of MIBC when two or more patient-specific single nucleotide variants are detected above a confidence threshold of 0.97.

Images