Differentiating methylation levels in complex biological samples

Abstract

The present invention relates to distinguishing methylation levels in complex biological samples. Provided herein is a method for distinguishing aberrant methylation levels of DNA from a first cell type comprising the steps of: (a) providing a test data set; (b) providing methylation status of the plurality of sites in reference genomic DNA from one or more reference individual organisms, (c) determining methylation difference between the test genomic DNA and the reference genomic DNA for each of the sites, thereby providing a normalized methylation difference for each site; and (d) weighting the normalized methylation difference for each of the sites by the coverage at each of the sites, thereby determining an aggregated coverage-weighted normalized methylation difference score. Also provided herein are sensitive methods for using genomic DNA methylation levels to distinguish cancer cells from normal cells and classify different cancer types according to their tissue of origin.

Description

[0001] This application is a divisional application of Chinese patent application 201680082163.6, filed on December 15, 2016, entitled "Distinguishing methylation levels in complex biological samples".

[0002] Cross-references to related applications

[0003] This application claims the benefit of U.S. Provisional Application Serial No. 62 / 401,591, filed September 29, 2016, and U.S. Provisional Application Serial No. 62 / 268,961, filed December 17, 2015, each of which is incorporated herein by reference. Background Technology

[0004] This disclosure relates to identifying methylation patterns in genomic DNA. Specific implementations relate to the prediction, diagnosis, prognosis, and monitoring of various conditions based on genomic methylation patterns.

[0005] Changes in cellular genetic information, such as mutations in gene sequences that can affect gene expression and / or protein sequence, are associated with many diseases and conditions. However, genes can also undergo changes that affect gene expression; changes caused by mechanisms other than genetic mutations. Epigenetics studies changes in gene expression caused by mechanisms other than underlying DNA sequence changes, and DNA methylation is one of these mechanisms. DNA methylation, such as the addition of a methyl group at the 5-position of the cytosine ring or the sixth nitrogen atom of the adenine ring, is common and plays a key role in regulating gene expression in the development and differentiation of diseases such as multiple sclerosis, diabetes, schizophrenia, aging, and cancer. In adult somatic cells, DNA methylation typically occurs in regions where cytosine nucleotides (C) and guanine nucleotides (G) are found adjacent to each other, with C and G linked by a phosphate group (p); the linear construct is called a “CpG” site. Methylation in specific gene regions (e.g., gene promoter regions) can increase or decrease the expression of these genes.

[0006] DNA methylation is ubiquitous and plays a crucial role in regulating gene expression in development, differentiation, and disease. Methylation in specific regions of genes (e.g., their promoter regions) can suppress the expression of these genes (Baylin and Herman (2000) DNA hypermethylation in tumorigenesis: epigenetics joins genetics. Trends Genet, 16, 168-174.; Jones and Laird (1999) Cancer epigenetics comes of age. Nat Genet, 21, 163-167.). Gene silencing effects of methylated regions have been shown to occur through the interaction of methylcytosine-binding proteins with other structural compounds in chromatin (Razin (1998) CpG methylation, chromatin structure and gene silencing - a three-way connection. Embo J, 17, 4905-4908.; Yan et al. (2001) Role of DNA methylation and histone acetylation in steroid receptor expression in breast cancer. J Mammary Gland Biol Neoplasia, 6, 183-192.). This, in turn, prevents DNA from reaching transcription factors through histone deacetylation and changes in chromatin structure (Bestor (1998) Gene silencing. Methylation meets acetylation. Nature, 393, 311-312.). Genomic imprinting, in which imprinted genes are preferentially expressed from maternal or paternal alleles, also involves DNA methylation. Imprinting disorders have been linked to several developmental disorders (Kumar (2000) Rett and ICF syndromes: methylation moves into medicine. J Biosci, 25, 213-214.; Sasaki et al. (1993) DNA methylation and genomic imprinting in mammals. Exs, 64, 469-486.; Zhong et al. (1996) A survey of FRAXE allelesizes in three populations. Am J Med Genet, 64, 415-419.). The references cited above are incorporated herein by reference.

[0007] In vertebrates, DNA methylation patterns are established early in embryonic development, and the distribution of 5-methylcytosine (5mC) along chromosomes is generally maintained throughout the lifespan of the organism (Razin and Cedar (1993) DNA methylation and embryogenesis. Exs, 64, 343-357; Reik et al. (2001) Epigenetic reprogramming in mammalian development. Science, 293, 1089-1093, each of which is incorporated herein by reference). Stable transcriptional silencing is important for normal development and is associated with several epigenetic modifications. If methylation patterns are not properly established or maintained, various conditions may occur, such as intellectual disability, immunodeficiency, and sporadic or hereditary cancers.

[0008] Changes in DNA methylation are considered one of the most common molecular alterations in human tumorigenesis. Hypermethylation of CpG sites in the promoter regions of tumor suppressor genes is a common mechanism of gene inactivation in cancer. Insufficient methylation of genomic DNA has been observed in tumor cells. Furthermore, associations between insufficient methylation and increased gene expression have been reported for many oncogenes. Monitoring overall changes in methylation patterns has been applied to the molecular classification of cancers; for example, gene hypermethylation has been associated with clinical risk groups in neuroblastoma and hormone receptor status in response to tamoxifen in breast cancer.

[0009] Besides playing a crucial role in cancer detection, a proper understanding of genetic methylation patterns has been used to detect other conditions. The initiation and maintenance of the inactivated X chromosome in female eutherians have been found to depend on methylation (Goto and Monk (1998) Regulation of X-chromosome inactivation in development in mice and humans. Microbiol Mol Biol Rev, 62, 362-378, cited herein). Rett syndrome (RTT) is an X-linked dominant disorder caused by mutations in the MeCP2 gene, further complicated by X-chromosome inactivation (XCI) patterns. Current models predict that MeCP2 represses transcription by binding methylated CpG residues and mediating chromatin remodeling (Dragich et al. (2000) Rett syndrome: a surprising result of mutation in MECP2. Hum Mol Genet, 9, 2365-2375, cited herein).

[0010] Several technical challenges hinder the development of methylation detection technologies into robust and cost-effective screening tools. For example, the accuracy and affordability of currently available technologies can be affected by impurities in the sample being tested. Therefore, cumbersome and expensive purification techniques are often employed to purify genomic samples from background nucleic acids. For instance, tumor biopsies are used to physically isolate tumor tissue from healthy tissue. Depending on the depth of the tissue within an individual, biopsies may require uncomfortable and risky collection procedures such as needle biopsies, endoscopy, bronchoscopy, colonoscopy, or surgery. The presence of circulating tumor DNA in the blood offers an attractive alternative to such biopsies. However, circulating tumor DNA is typically present in low amounts and exists against a background of relatively large amounts of non-tumor DNA.

[0011] Therefore, there is a need for methods to distinguish methylation patterns in complex genomic samples (e.g., tumor DNA) from a specific tissue of interest within a background of other genomic materials (e.g., circulating DNA) from other tissues. The methods and apparatus listed in this article meet this need and offer additional advantages. Summary of the Invention

[0012] This disclosure provides a method for distinguishing aberrant methylation levels of DNA from a first cell type. The method may include the steps of: (a) providing a test dataset comprising (i) methylation states at multiple sites of test genomic DNA from at least one test organism, and (ii) coverage at each of the methylation states for detecting the methylation states; (b) providing methylation states at the multiple sites of reference genomic DNA from one or more reference individual organisms; (c) determining a methylation difference between the test genomic DNA and the reference genomic DNA for each site, thereby providing a normalized methylation difference for each site; and (d) weighting the normalized methylation difference at each site by the coverage at each site, thereby determining a total coverage-weighted normalized methylation difference score.

[0013] A method for distinguishing anomalous methylation levels of DNA in a sample containing DNA from multiple different cell types is also provided, comprising the steps of: (a) providing a sample containing a mixture of genomic DNA from multiple different cell types from at least one test organism, thereby providing test genomic DNA; (b) detecting the methylation status of a plurality of sites in the test genomic DNA; (c) determining coverage at each of the sites for detecting the methylation status; (d) providing the methylation status of the plurality of sites in reference genomic DNA from at least one reference individual, wherein the at least one test organism and the reference individual are optionally of the same species; (e) determining the methylation difference between the test genomic DNA and the reference genomic DNA for each site, thereby providing a normalized methylation difference for each site; and (f) weighting the normalized methylation difference at each site by the coverage at each site, thereby determining a total coverage-weighted normalized methylation difference score.

[0014] In a specific implementation, this disclosure provides a method for detecting symptoms such as cancer. The method may include the following steps: (a) providing a mixture of genomic DNA from blood of an individual suspected of having the condition (e.g., cancer), wherein the mixture contains genomic DNA from multiple different cell types of the individual, thereby providing test genomic DNA; (b) detecting the methylation status of multiple sites in the test genomic DNA; (c) determining coverage at each of the sites for detecting the methylation status; (d) providing the methylation status of the multiple sites in reference genomic DNA from at least one reference individual, the reference individual being known to have the condition (e.g., cancer) or known not to have the condition (e.g., cancer); (e) determining the methylation difference between the test genomic DNA and the reference genomic DNA for each site, thereby providing a normalized methylation difference for each site; (f) weighting the normalized methylation difference at each site by the coverage at each site, thereby determining a total coverage-weighted normalized methylation difference score; and (g) determining whether the individual has or does not have the condition (e.g., cancer) based on the total coverage-weighted normalized methylation difference score.

[0015] This disclosure also provides an alternative sensitive method for distinguishing aberrant methylation levels in DNA from a first cell type. The method may include: a first stage of establishing a methylation baseline, comprising the steps of: (a) providing the methylation status of multiple sites in baseline genomic DNA from two or more normal individual organisms; and (b) determining the average methylation level and standard deviation of the methylation level of the baseline genomic DNA for each said site; and a second stage of determining a total methylation score of multiple training samples, comprising the steps of: (c) providing a training set of normal genomic DNA samples from two or more normal individual organisms, comprising (i) the methylation status of multiple sites in the training set of normal genomic DNA samples, and optionally (ii) a method for detecting the methylation level of said sites. (d) Coverage at each of the said sites in the methylation state; (e) Determining the methylation difference between each normal genomic DNA sample in the training set and the baseline genomic DNA for each said site, thereby providing the normalized methylation difference of each normal genomic DNA sample in the training set at each site; (f) Converting the normalized methylation difference of each normal genomic DNA sample in the training set at each site into a probability of observing such or greater normalized methylation difference, and optionally weighting the probability of such an event; (c) Determining the total methylation score of each normal genomic DNA sample in the training set to obtain the training set methylation score; (g) Calculate the average methylation score and standard deviation of the methylation scores in the training set; a third stage for determining the total methylation score of a given test sample, which may be performed before, after, or simultaneously with the second stage, includes the following steps: (h) providing a test dataset comprising (i) the methylation status of the plurality of sites of test genomic DNA from at least one test organism, and optionally (ii) coverage at each of the sites for detecting the methylation status; (i) determining the methylation difference between the test genomic DNA and the baseline genomic DNA for each of the sites, thereby providing normalized methylation of the test genomic DNA. The process includes: (j) converting the normalized methylation difference of the test genomic DNA at each said site into a probability of observing such or greater normalized methylation difference, and optionally weighting the probability of such an event; and (k) determining the total methylation score of the test genomic DNA; and a fourth stage comprising: (l) comparing the methylation score of the test genomic DNA with the mean methylation score and the standard deviation of the methylation score in the training set of normal genomic DNA to determine the number of standard deviations of the distribution of the methylation score of the test genomic DNA with respect to the methylation score in the training set of normal genomic DNA.

[0016] An alternative sensitive method is also provided for distinguishing abnormal methylation levels of DNA in samples containing DNA from multiple different cell types.The method may include: a first stage of establishing a methylation baseline, comprising the steps of: (a) providing methylation status at multiple sites in baseline genomic DNA from two or more normal individual organisms; and (b) determining the average methylation level and standard deviation of the methylation level of the baseline genomic DNA for each said site; and a second stage of determining a total methylation score of multiple training samples, comprising the steps of: (c) providing a training set of normal genomic DNA samples from two or more normal individual organisms, comprising (i) methylation status at multiple sites in the training set of normal genomic DNA samples, and optionally (ii) coverage at each said site for detecting the methylation status; (d) For each of the said sites, determine the methylation difference between each normal genomic DNA sample in the training set and the baseline genomic DNA, thereby providing the normalized methylation difference for each normal genomic DNA sample in the training set at each site; (e) convert the normalized methylation difference for each normal genomic DNA sample in the training set at each site into a probability of observing such or greater normalized methylation difference, and optionally weight the probabilities; (f) determine the total methylation score for each normal genomic DNA sample in the training set to obtain the training set methylation score; and (g) calculate the mean methylation score and standard deviation of the training set methylation score; determine the methylation score for a given test sample. The third stage of the total methylation fraction, which may be performed before, after, or simultaneously with the second stage, includes the following steps: (h) providing a mixture of genomic DNA from a test organism suspected of having symptoms associated with abnormal DNA methylation levels, wherein the mixture contains genomic DNA from multiple different cell types of the test organism, thereby providing test genomic DNA; (i) detecting the methylation status of the plurality of sites in the test genomic DNA, and optionally determining coverage of each of the sites used to detect the methylation status; (j) determining the methylation difference between the test genomic DNA and the baseline genomic DNA for each of the sites, thereby providing the total methylation fraction. The test genomic DNA is subjected to normalized methylation differences; (k) the normalized methylation differences of the test genomic DNA at each said site are converted into a probability of observing such or greater normalized methylation differences, and optionally the probabilities of such events are weighted; and (l) the total methylation fraction of the test genomic DNA is determined; and a fourth stage is included in which (m) the methylation fraction of the test genomic DNA is compared with the mean methylation fraction and the standard deviation of the methylation fraction in the training set of normal genomic DNA to determine the number of standard deviations of the distribution of the methylation fraction of the test genomic DNA with respect to the methylation fraction in the training set of normal genomic DNA.

[0017] In a specific implementation, this disclosure provides a method for detecting symptoms such as cancer.The method may include: a first stage of establishing a methylation baseline, comprising the steps of: (a) providing methylation status at multiple sites in baseline genomic DNA from at least one normal individual organism; and (b) determining the average methylation level and standard deviation of the methylation level of the baseline genomic DNA for each said site; and a second stage of determining a total methylation score of multiple training samples, comprising the steps of: (c) providing a training set of normal genomic DNA samples from two or more normal individual organisms, comprising (i) methylation status at multiple sites in the training set of normal genomic DNA samples, and optionally (ii) coverage at each said site for detecting the methylation status. (d) For each of the said sites, determine the methylation difference between each normal genomic DNA sample in the training set and the baseline genomic DNA, thereby providing the normalized methylation difference for each normal genomic DNA sample in the training set at each site; (e) convert the normalized methylation difference for each normal genomic DNA sample in the training set at each site into a probability of observing such or greater normalized methylation difference, and optionally weight the probability of such an event; (f) determine the methylation fraction for each normal genomic DNA sample in the training set to obtain the training set methylation fraction; and (g) calculate the mean methylation fraction and standard deviation of the training set methylation fraction; A third stage for determining the total methylation fraction of a given test sample, which may be performed before, after, or simultaneously with the second stage, includes the following steps: (h) providing a mixture of genomic DNA from a test organism suspected of having the condition, wherein the mixture contains genomic DNA from multiple different cell types of the test organism, thereby providing test genomic DNA; (i) detecting the methylation status of the plurality of sites in the test genomic DNA, and optionally determining coverage at each of the sites for detecting the methylation status; (j) determining the methylation difference between the test genomic DNA and the baseline genomic DNA for each of the sites, thereby providing the test genomic DNA. The test genomic DNA is subjected to normalized methylation differences; (k) the normalized methylation differences of the test genomic DNA at each said site are converted into a probability of observing such or greater normalized methylation differences, and optionally the probabilities of such events are weighted; and (l) the test genomic DNA is determined; and a fourth stage is included in which (m) the test genomic DNA is compared with the mean methylation fraction and the standard deviation of the methylation fraction in the training set of normal genomic DNA to determine the number of standard deviations of the distribution of the test genomic DNA's methylation fraction with respect to the methylation fraction in the training set of normal genomic DNA.

[0018] This disclosure provides a method for identifying or classifying a specific cancer type in an organism using methylation levels. The method may include: a first stage of identifying a specific cancer that can be used as a cancer type, comprising: (a) providing a dataset including the methylation status of multiple sites in genomic DNA from a clinical sample known to include the specific cancer; and a second stage of selecting overmethylated sites, comprising: (b) identifying overmethylated sites specific to the cancer type, the identification comprising (i) determining the average methylation level of each site in the methylated DNA of the clinical sample known to include the specific cancer, (ii) determining which sites satisfy a first threshold, a second threshold, or a combination thereof, wherein determining the first threshold comprises (1) determining an absolute value of the average methylation level of each site; (2) sorting the average methylation levels of each site from low to high; and (3) selecting those sites having an average methylation level greater than or equal to a first preselected value in a percentage order, and wherein determining the second threshold comprises (1) determining an absolute value of the average methylation level of each site; and (2) selecting those sites having an average methylation level greater than a second preselected value; and (iii) compiling the cancer... (a) A list of hypermethylation sites specific to the cancer type; and (b) repeating (a) and (b) for each specific cancer to generate multiple additional lists of hypermethylation sites specific to the cancer type; including a third stage of analyzing test genomic DNA samples from a test organism, which is performed by: (d) providing a test dataset comprising the methylation level of each hypermethylation site of test genomic DNA from an individual test organism, wherein the hypermethylation site is one of the list of cancer type-specific hypermethylation sites identified in steps (b) and (c); (e) averaging the methylation level of each of the hypermethylation sites to generate a single average methylation level of the test genomic DNA for the cancer type identified in steps (b) and (c); (f) repeating step (e) for each cancer type to generate an average methylation level for each cancer type; and (g) sorting the average methylation levels of each cancer type from low to high, wherein the cancer type corresponding to the highest average methylation level is the cancer present in the individual test organism. Attached Figure Description

[0019] Figure 1 Guidelines for designing probes for regions of the genome with methylation sites are shown.

[0020] Figure 2 The workflow for targeted circulating tumor DNA (ctDNA) methylation sequencing is shown.

[0021] Figure 3The total coverage-weighted normalized methylation differences (z-scores) for various cancer samples, as determined as described herein, are shown at various titration levels.

[0022] Figure 4 The coverage-weighted methylation fractions for colorectal cancer samples, as described herein, are shown at various titration levels.

[0023] Figure 5 The methylation scores of 66 samples from patients with advanced cancer and 25 samples from normal individuals are shown, demonstrating the ability of the methylation score algorithm to distinguish advanced cancer samples from normal samples.

[0024] Figure 6 Show Figure 5 A tabular summary of the methylation fractions shown.

[0025] Figure 7 The correlation between methylation profiles of plasma and tissue DNA samples from cancer patients is shown.

[0026] Figure 8 The results of cancer type classification for tumor tissue samples are shown, demonstrating the ability of the cancer type classification algorithm to identify most tumors with high accuracy based on DNA methylation data.

[0027] Figure 9 The results of cancer type classification based on plasma DNA samples from cancer patients demonstrate high clinical sensitivity for colorectal and breast cancer. Detailed Implementation

[0028] DNA methylation data can provide valuable information when evaluated independently or in combination with other information such as genotype or gene expression patterns. One aim of the methods outlined in this article is to determine, for example, whether one or more sites in the genome are differentially methylated in the test sample compared to a reference sample or dataset.

[0029] Specific implementation schemes can be used to detect, screen, monitor (e.g., for relapse, remission, or response to treatment), grade, classify (e.g., to help select the most appropriate form of treatment) and predict cancer using methylation analysis of circulating plasma / serum DNA.

[0030] Cancer DNA is known to demonstrate aberrant DNA methylation (see, for example, Herman et al. 2003 N Engl J Med 349:2042-2054, which is incorporated herein by reference). For instance, the CpG site promoters of genes such as tumor suppressor genes are hypermethylated compared to non-cancer cells, while the CpG sites in the gene body are hypomethylated. In specific embodiments of the methods described herein, methylation patterns detected from the blood of individuals suspected of having cancer indicate the methylation status of potentially cancerous tissue, such that the patterns are expected to be different between individuals with cancer when compared to healthy individuals without cancer or when compared to individuals whose cancer has been cured.

[0031] Because aberrant methylation occurs in most cancers, the methods described herein can be applied to detect any of a variety of malignant tumors exhibiting aberrant methylation, such as those in the lungs, breast, colorectal, prostate, nasopharynx, stomach, testes, skin, nervous system, bone, ovaries, liver, blood tissues, pancreas, uterus, kidneys, and lymphoid tissues. Malignant tumors can have various histological subtypes, such as carcinoma, adenocarcinoma, sarcoma, fibroadenocarcinoma, neuroendocrine, or undifferentiated tumors.

[0032] In specific implementations, methods for determining methylation patterns can be used to monitor fetal development (e.g., to determine the presence or absence of developmental abnormalities) or to identify the presence of specific diseases or conditions. In such cases, the methods can be performed using samples obtained from pregnant women (e.g., blood, tissue, or amniotic fluid), and these samples can be used to assess the methylation levels of fetal nucleic acids. DNA methylation profiles of placental tissue can be used to assess the pathophysiology of pregnancy-related or developmental disorders such as preeclampsia and intrauterine growth retardation. Conditions in genomic imprinting are associated with developmental disorders such as Prader-Willi syndrome and Angelman syndrome, and can be identified or assessed using the methods of this disclosure. Genomic imprinting and altered profiles of overall DNA methylation in placental and fetal tissues have been observed in pregnancies resulting from assisted reproductive technologies (see, for example, Hiura et al. 2012 Hum Reprod; 27:2541-2548, incorporated herein by reference) and can be detected using the methods listed herein. Exemplary methods that can be modified for use with the methods of this disclosure are listed in U.S. Patent Application Publication Nos. 2013 / 0189684 A1 or 2014 / 0080715 A1, each of which is incorporated herein by reference.

[0033] The ability to determine placental or fetal methylation patterns based on maternal plasma provides a non-invasive method for identifying, detecting, and monitoring pregnancy-related conditions such as preeclampsia, intrauterine growth restriction, preterm birth, and other conditions. For example, detecting disease-specific abnormal methylation signatures allows for the screening, diagnosis, and monitoring of such pregnancy-related conditions.

[0034] In addition, the methods listed in this article provide diagnostic or prognostic information for other conditions. For example, liver tissue can be analyzed to determine liver-specific methylation patterns that can be used to identify liver lesions. Other tissues that can also be analyzed include brain cells, bone, lungs, heart, muscles, and kidneys. DNA can be obtained from blood samples and analyzed using the methods listed in this article to determine the status of any of the various tissues that contribute DNA to the bloodstream.

[0035] Furthermore, the methylation pattern of transplanted organs can be determined based on the plasma DNA of the organ transplant recipient. Plasma-based graft analysis can be a synergistic technique based on plasma-based graft genome analysis, such as those listed below: Zheng et al. 2012 Clin Chem 58: 549-558; Lo et al. 1998 Lancet 351: 1329-1330; or Snyder et al. 2011 Proc Natl Acad Sci USA; 108: 6229-6234, each of which is incorporated herein by reference).

[0036] For example, methylation patterns in various tissues can change over time due to development, aging, disease progression (e.g., inflammation, cancer, or cirrhosis), or treatment. The dynamic nature of DNA methylation makes such analyses potentially very valuable for monitoring physiological and pathological processes. For instance, by detecting changes in an individual's plasma methylation pattern compared to baseline values ​​obtained when the individual was healthy, disease processes in organs that contribute to plasma DNA can be detected.

[0037] Unless otherwise stated, the terms used herein shall be understood to have their common meaning in the relevant art. Several terms used herein and their meanings are listed below.

[0038] As used herein, in relation to the use of DNA, the term "cell-free" is intended to mean DNA that has been removed from the body's cells. DNA removal can be a natural process, such as necrosis or apoptosis. Cell-free DNA is typically obtained from blood or portions of it, such as plasma. Cell-free DNA can also be obtained from other bodily fluids or tissues.

[0039] As used herein, the term "cell type" is intended to identify cells based on morphological, phenotypic, developmental origin, or other known or identifiable distinguishable cellular characteristics. Various distinct cell types can be obtained from a single organism (or from the same species of organism). Exemplary cell types include, but are not limited to, bladder, pancreatic epithelial cells, pancreatic α, pancreatic β, pancreatic endothelial cells, bone marrow lymphoblasts, bone marrow B lymphoblasts, bone marrow macrophages, bone marrow erythroblasts, bone marrow dendritic cells, bone marrow adipocytes, bone marrow osteocytes, bone marrow chondrocytes, promyelocytes, bone marrow megakaryocytes, bladder, brain B lymphocytes, brain glial cells, neurons, brain astrocytes, neuroectoderm, brain macrophages, brain microglia, brain epithelial cells, cortical neurons, brain fibroblasts, breast epithelial cells, colonic epithelial cells, colonic B lymphocytes, mammary epithelial cells, mammary myoepithelial cells, and mammary fibroblasts. Colonic epithelial cells, cervical epithelial cells, ovarian epithelial cells, ovarian fibroblasts, mammary duct epithelial cells, tongue epithelial cells, tonsillar dendritic cells, tonsillar B lymphocytes, peripheral blood lymphoblasts, peripheral blood T lymphoblasts, peripheral blood skin T lymphocytes, peripheral blood natural killer cells, peripheral blood B lymphoblasts, peripheral blood monocytes, peripheral blood myeloblasts, peripheral blood mononuclear cells, peripheral blood promyelocytes, peripheral blood macrophages, peripheral blood basophils, liver endothelial cells, liver mast cells, liver epithelial cells, liver B lymphocytes, spleen endothelial cells, spleen epithelial cells, spleen B lymphocytes, liver hepatocytes, liver Alexandrium cells. Alexander cells, liver fibroblasts, lung epithelial cells, bronchial epithelial cells, lung fibroblasts, lung B lymphocytes, lung Schwann cells, lung squamous cells, lung macrophages, lung osteoblasts, neuroendocrine cells, alveoli, gastric epithelial cells, and gastric fibroblasts. In some embodiments, two cells may be considered to be of the same type, even if one of the cells has been phenotypically or morphologically altered by a symptom or disease such as cancer. For comparative purposes, the first cell altered by the disease or symptom may be compared with the second cell based on the known or suspected state of the first cell before the alteration. For example, cancerous pancreatic ductal epithelial cells may be considered to be of the same type as non-cancer pancreatic ductal epithelial cells.

[0040] As used herein, when referring to the use of DNA, the term “circulatory” is intended to mean DNA that is currently or previously moving through the circulatory system of an organism, whether in a cell-free form or within circulating cells.

[0041] As used herein, when referring to genetic loci, the term "coverage" is intended to mean the number of detection events (e.g., sequence reads) that align with or "cover" a locus. In some embodiments, the term refers to the average number of detection events (e.g., sequence reads) that align with or "cover" multiple loci. Generally, the level of coverage obtained from sequencing methods is directly related to the confidence of the call accuracy (e.g., nucleotide type or methylation state) determined at a particular base position or genetic locus. At higher coverage levels, the locus is covered by a greater number of aligned sequence reads, so the call can be performed with higher confidence.

[0042] As used herein, the term "CpG site" is intended to refer to a location in a nucleic acid molecule or a sequence representation of the molecule in which a cytosine nucleotide and a guanine nucleotide occur, with the 3′ oxygen of the cytosine nucleotide covalently linked to the 5′ phosphate of the guanine nucleotide. The nucleic acid is typically DNA. The cytosine nucleotide may optionally contain a methyl moiety, a hydroxymethyl moiety, or a hydrogen moiety at the 5′ position of the pyrimidine ring.

[0043] As used herein, when referring to DNA, the term "derived" is intended to refer to the source from which the DNA is derived or the origin of DNA synthesis. In the case of biologically derived DNA, the term can be used to refer to an in vivo source from which the DNA is derived or an in vivo origin of DNA synthesis. Exemplary origins include, but are not limited to, cells, cell types, tissues, tissue types, organisms, or species of organisms. In the case of synthetically derived DNA, the term can be used to refer to an in vitro source from which the DNA is derived or an in vitro origin of DNA synthesis. However, DNA molecules derived from a particular source or origin may subsequently be copied or amplified. The sequence of the resulting copy or amplicon may be referred to as derived from said source or origin.

[0044] As used herein, when referring to collections of items, the term "each" is intended to identify an individual item in the collection, but not necessarily every item in the collection. Exceptions may occur if explicitly stated otherwise or if the context clearly indicates otherwise.

[0045] As used herein, the term "methylation difference" is intended to refer to a qualitative or quantitative marker that two nucleotides or nucleic acids do not have the same methylation state. Methylation difference can indicate nucleotides at alignment sites on different nucleic acids. In some cases, methylation difference can be the sum or aggregate of multiple alignment sites. When two or more nucleic acids are aligned, methylation difference can be the average across one or more alignment sites.

[0046] As used herein, when referring to a locus (e.g., a CpG site or a polynucleotide fragment) spanning several molecules containing this locus, the term "methylation state" refers to one or more characteristics of a locus associated with the presence or absence of a methylated portion. Non-limiting examples of such characteristics include whether any cytosine (C) base within the locus is methylated, the position of the methylated C base, the percentage of methylated C bases at a particular locus, and allelic differences in methylation due to, for example, differences in the origin of alleles. References to the methylation state of a particular CpG site in a nucleic acid molecule relate to the presence or absence of a methylated portion at position 5 of the pyrimidine ring of cytosine. The term may be applied to one or more cytosine nucleotides (or their representation, e.g., chemical formulas) or one or more nucleic acid molecules (or their representation, e.g., sequence representations). The term may also refer to the relative or absolute amount (e.g., concentration) of methylated or unmethylated C at a particular locus in a nucleic acid. Methylation state is sometimes hypermethylated and sometimes undermethylated. For example, if all or most of the C bases within a locus are methylated, the methylation state may be described as "hypermethylated." In another instance, if all or most of the C bases within a locus are not methylated, the methylation state may be described as "undermethylated." Similarly, if all or most of the C bases within a locus are methylated compared to a reference, the methylation state is considered hypermethylated compared to the reference. Alternatively, if all or most of the C bases within a locus are not methylated compared to a reference, the methylation state is considered undermethylated compared to the reference.

[0047] A "methylation site" is a locus in a nucleic acid where methylation has occurred or is likely to occur. Methylation sites are sometimes C bases or multiple C bases in a region, and sometimes they are CpG sites within a locus. Each methylation site in a locus may or may not be methylated. Methylation sites are susceptible to methylation due to naturally occurring events in vivo or due to events that chemically methylate nucleotides in vitro.

[0048] As used herein, when referring to two or more components, the term "mixture" is intended to mean two or more components present simultaneously in a fluid or container. The components are typically able to come into contact with each other by diffusion or agitation. The components can be individual molecules (e.g., two or more nucleic acid fragments), or the components can be a portion of a single molecule (e.g., a sequence region on a long nucleic acid molecule).

[0049] As used herein, the term "tissue" is intended to mean a collection or aggregation of cells that work together to perform one or more specific functions in an organism. Cells may optionally be morphologically similar. Exemplary tissues include, but are not limited to, the eye, muscle, skin, tendon, vein, artery, blood, heart, spleen, lymph nodes, bone, bone marrow, lung, bronchi, trachea, intestine, small intestine, large intestine, colon, rectum, salivary glands, tongue, gallbladder, appendix, liver, pancreas, brain, stomach, skin, kidney, ureter, bladder, urethra, gonads, testes, ovaries, uterus, fallopian tubes, thymus, pituitary gland, thyroid gland, adrenal gland, or parathyroid gland. Tissues can be derived from any of the various organs of the human body or other bodies.

[0050] Based on the above definitions, the following embodiments, which are listed and enumerated in the claims, can be understood.

[0051] This disclosure provides a method for distinguishing aberrant methylation levels in DNA from a first cell type. The method may include the steps of: (a) providing a test dataset comprising (i) methylation status at multiple sites (e.g., CpG sites) of test genomic DNA from at least one test organism, and (ii) coverage at each site (e.g., CpG site) for detecting methylation status; (b) providing methylation status at multiple sites (e.g., CpG sites) of reference genomic DNA from one or more reference individual organisms; (c) determining methylation differences between the test genomic DNA and the reference genomic DNA for each site (e.g., CpG site), thereby providing a normalized methylation difference for each site (e.g., CpG site); and (d) weighting the normalized methylation difference for each site (e.g., CpG site) by coverage at each site (e.g., CpG site), thereby determining a total coverage-weighted normalized methylation difference score. Optionally, the sites derived from the test genomic DNA are derived from multiple different cell types from the individual test organism, and alternatively, the cell type from which each site is derived is unknown. In another optional embodiment, the individual test organism and one or more reference individual organisms are of the same species.

[0052] A method for distinguishing aberrant methylation levels of DNA in a sample containing DNA from multiple different cell types is also provided, comprising the steps of: (a) providing a sample containing a mixture of genomic DNA from multiple different cell types from at least one test organism, thereby providing test genomic DNA; (b) detecting the methylation status of multiple sites (e.g., CpG sites) in the test genomic DNA; (c) determining coverage at each site (e.g., CpG site) for detecting methylation status; (d) providing the methylation status of multiple sites (e.g., CpG sites) in reference genomic DNA from at least one reference individual, wherein the at least one test organism and the reference individual are optionally of the same species; (e) determining the methylation difference between the test genomic DNA and the reference genomic DNA for each site (e.g., CpG site), thereby providing a normalized methylation difference at each site (e.g., CpG site); and (f) weighting the normalized methylation difference at each site (e.g., CpG site) by coverage at each site (e.g., CpG site), thereby determining a total coverage-weighted normalized methylation difference score.

[0053] The present invention also provides an alternative sensitive method for distinguishing abnormal methylation levels of DNA from a first cell type.

[0054] The first phase of this method involves establishing a methylation baseline, which includes the following steps: (a) providing the methylation status of multiple sites (e.g., CpG sites) in the baseline genomic DNA from two or more normal individual organisms; and (b) determining the average methylation level and the standard deviation of the methylation level of the baseline genomic DNA for each site (e.g., CpG site). In some embodiments, the number of normal individual organisms providing the baseline genomic DNA is at least 3, at least 5, at least 10, at least 20, at least 50, or at least 100.

[0055] The second phase of this method involves determining the aggregate methylation score of multiple training samples, which includes the following steps: (c) providing a training set of normal genomic DNA samples from two or more normal individual organisms, comprising (i) the methylation status of multiple sites (e.g., CpG sites) in the training set of normal genomic DNA samples, and optionally (ii) coverage at each site (e.g., CpG site) used to detect methylation status; (d) determining the methylation difference between each normal genomic DNA sample in the training set and the baseline genomic DNA for each site (e.g., CpG site), thereby improving the overall methylation score. The training set includes: (e) providing normalized methylation differences for each normal genomic DNA sample at each site (e.g., CpG site); (f) converting the normalized methylation differences for each normal genomic DNA sample at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation differences, and optionally weighting the probabilities of such events; (g) determining the aggregate methylation score for each normal genomic DNA sample in the training set to obtain the training set methylation score; and (g) calculating the mean methylation score and standard deviation of the training set methylation score. In some embodiments, the number of normal individual organisms from which the genomic DNA of the training set is provided is at least 3, at least 5, at least 10, at least 20, at least 50, or at least 100.

[0056] The third phase of this method (which may be performed before, after, or concurrently with the second phase) involves determining the total methylation score of a given test sample, comprising the following steps: (h) providing a test dataset comprising (i) the methylation status of multiple sites (e.g., CpG sites) of test genomic DNA from at least one test organism, and optionally (ii) coverage at said sites (e.g., CpG sites) for detecting methylation status; (i) determining the methylation difference between the test genomic DNA and baseline genomic DNA for each site (e.g., CpG site), thereby providing a normalized methylation difference of the test genomic DNA; (j) converting the normalized methylation difference of the test genomic DNA at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation difference, and optionally weighting the probability of such an event; and (k) determining the total methylation score of the test genomic DNA.

[0057] The fourth stage of this method involves the following steps: (1) comparing the methylation fraction of the test genomic DNA with the mean methylation fraction and the standard deviation of the methylation fraction in a training set of normal genomic DNA to determine the number of standard deviations of the distribution of the methylation fraction of the test genomic DNA compared with that of the methylation fraction in the training set of normal genomic DNA. In cases where the number of standard deviations exceeds a predetermined threshold (e.g., 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, etc.), the test sample is considered to have an abnormal DNA methylation level.

[0058] Optionally, the methylation sites (e.g., CpG sites) from the test genomic DNA are derived from multiple different cell types from the individual test organism, and alternatively, the cell type from which each site (e.g., CpG site) is derived is unknown. In another optional embodiment, the individual test organism and one or more baseline individual organisms, training individual organisms, or combinations thereof are of the same species.

[0059] An alternative sensitive method is also provided for distinguishing abnormal methylation levels of DNA in samples containing DNA from multiple different cell types.

[0060] The first phase of this method involves establishing a methylation baseline, which includes the following steps: (a) providing the methylation status of multiple sites (e.g., CpG sites) in the baseline genomic DNA from two or more normal individual organisms; and (b) determining the average methylation level and the standard deviation of the methylation level of the baseline genomic DNA for each site (e.g., CpG site). In some embodiments, the number of normal individual organisms providing the baseline genomic DNA is at least 3, at least 5, at least 10, at least 20, at least 50, or at least 100.

[0061] The second phase of this method involves determining the aggregate methylation score of multiple training samples, which includes the following steps: (c) providing a training set of normal genomic DNA samples from two or more normal individual organisms, comprising (i) the methylation status of multiple sites (e.g., CpG sites) in the training set of normal genomic DNA samples, and optionally (ii) coverage at each site (e.g., CpG site) used to detect methylation status; and (d) determining the methylation difference between each normal genomic DNA sample in the training set and the baseline genomic DNA for each site (e.g., CpG site). The training set is provided with normalized methylation differences for each normal genomic DNA sample at each site (e.g., CpG site); (e) the normalized methylation differences for each normal genomic DNA sample at each site (e.g., CpG site) are converted into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation differences, and optionally said probabilities are weighted; (f) the total methylation score for each normal genomic DNA sample in the training set is determined to obtain the training set methylation score; and (g) the mean methylation score and standard deviation of the training set methylation score are calculated. In some embodiments, the number of normal individual organisms from which the genomic DNA of the training set is provided is at least 3, at least 5, at least 10, at least 20, at least 50, or at least 100.

[0062] The third phase of this method (which may be performed before, after, or concurrently with the second phase) involves determining the total methylation score of a given test sample, comprising the following steps: (h) providing a mixture of genomic DNA from a test organism suspected of having a symptom (e.g., cancer) associated with abnormal DNA methylation levels, wherein the mixture contains genomic DNA from multiple different cell types of the test organism, thereby providing test genomic DNA; (i) detecting the methylation status at multiple sites (e.g., CpG sites) in the test genomic DNA, and optionally determining coverage at each site (e.g., CpG site) used to detect the methylation status; (j) determining the methylation difference between the test genomic DNA and baseline genomic DNA for each site (e.g., CpG site), thereby providing normalized methylation difference of the test genomic DNA; (k) converting the normalized methylation difference of the test genomic DNA at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation difference, and optionally weighting the probability of such an event; and (i) determining the total methylation score of the test genomic DNA.

[0063] The fourth stage of this method involves the following steps: (m) comparing the methylation fraction of the test genomic DNA with the mean methylation fraction and the standard deviation of the methylation fraction in a training set of normal genomic DNA to determine the number of standard deviations of the distribution of methylation fractions of the test genomic DNA compared to that in the training set of normal genomic DNA. The test sample is considered to have an abnormal DNA methylation level when the number of standard deviations exceeds a predetermined threshold (e.g., 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, etc.).

[0064] The methods described herein can be implemented for any of a variety of test organisms. Exemplary organisms include, but are not limited to, eukaryotic (single-celled or multicellular) organisms. Exemplary eukaryotic organisms include mammals such as rodents, mice, rats, rabbits, guinea pigs, ungulates, horses, sheep, pigs, goats, cattle, cats, dogs, primates, and humans or non-human primates; plants such as Arabidopsis thaliana, maize (Zea mays), sorghum, oats (Oryza sativa), wheat, rice, rapeseed, or soybeans; algae such as Chlamydomonas reinhardtii; nematodes such as Caenorhabditis elegans; insects such as Drosophila melanogaster, mosquitoes, fruit flies, bees, or spiders; fish such as zebrafish (Daniorerio); reptiles; amphibians such as frogs or African toads (Xenopus laevis); and Dictyostelium. The methods disclosed herein can also be used to assess methylation in organisms such as prokaryotes, including bacteria, Escherichia coli, Staphylococcus aureus, or Mycoplasma pneumoniae; archaea; viruses, including hepatitis C virus or human immunodeficiency virus; or viroids.

[0065] When applied to multicellular organisms, specific embodiments of the methods listed herein offer advantages because they provide a determination of the methylation status of genomic DNA derived from a particular cell or tissue within a background of nucleic acids derived from other cells or tissues. Therefore, the methods listed herein can be particularly useful for mammals such as humans. In some cases, the methods can be performed on samples containing a mixture of nucleic acids from several different cell or tissue types, such as samples obtained from blood or other biological fluids from multicellular organisms. Furthermore, the methods listed herein can be advantageously used to assess the methylation status of genomic DNA obtained from somatic cells of pregnant female mammals (such as pregnant female humans) and / or from somatic cells of one or more prenatal offspring conceived by a female.

[0066] In some embodiments, the method can be implemented for a mixture of genomic DNA from several different cell types from a mixed organismal environment (e.g., metagenomic samples) (such as ecological samples (e.g., ponds, oceans, vents, etc.) or digestive system samples (e.g., mouth, intestine, colon, etc.)). Therefore, the method can be implemented for mixed organism samples in which individual species are not isolated or cultured.

[0067] As will be apparent from the several exemplary embodiments listed herein, CpG sites from the test genomic DNA evaluated in the methods of this disclosure may optionally be derived from a variety of different cell types from an individual test organism. Alternatively, the cell type from which each said CpG site is derived need not be known. This is often the case when the sample used for said methods is derived from blood or another biological fluid or metagenomic sample.

[0068] In specific implementations, test samples used for the methods listed herein may contain circulating tumor DNA and circulating non-tumor DNA. This may be the case when the test sample contains DNA obtained, for example, from the blood of an individual known or suspected of having cancer.

[0069] Specific implementations of the methods described herein can be carried out using the methylation status of multiple sites on the test genomic DNA from an individual test organism. In some cases, the data is provided to the individual or system implementing the method. Alternatively, implementations of the methods may include one or more steps for detecting the methylation status of multiple sites in the test genome.

[0070] Methylation of sites (such as CpG dinucleotide sequences) can be measured using any of the various techniques used in the field of analyzing such sites. For example, methylation can be measured using restriction enzyme-based techniques that utilize methylation-sensitive restriction endonucleases to distinguish between methylated and unmethylated cytosine. Restriction enzyme-based techniques include, for example, restriction digestion with a methylation-sensitive restriction enzyme followed by nucleic acid sequencing (e.g., massively parallel sequencing or next-generation sequencing), Southern blotting, real-time PCR, restriction marker genome scanning (RLGS), or differential methylation hybridization (DMH).

[0071] Restriction enzymes characteristically hydrolyze DNA when and / or after recognizing a specific sequence or a motif typically between 4 and 8 bases in length. Among such enzymes, methylation-sensitive restriction enzymes are distinguished by the fact that they either cleave DNA or do not, depending on the cytosine methylation state of the specific CpG sequence present in the recognized motif. In methods employing such methylation-sensitive restriction enzymes, the digested DNA fragments can be differentially segregated (e.g., based on size or hybridization affinity to complementary probes), differentially amplified (e.g., based on affinity for amplification primers), or differentially detected (e.g., by microarray detection techniques or nucleic acid sequencing techniques), thus allowing the deduction of the methylation state of the sequence.

[0072] In some embodiments employing methylation-sensitive restriction enzymes, a post-digestion PCR amplification step is added, where a set of two oligonucleotide primers (one on each side of the methylation-sensitive restriction site) is used to amplify the digested genomic DNA. PCR products are generated and unrestricted templates are detected (e.g., due to the presence of methylation restriction sites), whereas no PCR products are generated when digestion occurs at the opposing methylation-sensitive restriction enzyme site. Techniques for restriction enzyme-based analysis of genomic methylation are well known in the art and include the following: differential methylation hybridization (DMH) (Huang et al., 1999, Human Mol. Genet. 8, 459-70); non-I-based differential methylation hybridization (e.g., WO02 / 086163A1); restriction marker genomic scanning (RLGS) (Plass et al., 1999, Genomics 58: 254-62); methylation-sensitive arbitrary primer PCR (AP-PCR) (Gonzalgo et al., 1997, Cancer Res. 57: 594-599); and methylated CpG site amplification (MCA) (Toyota et al., 1999, Cancer Res. 59: 2307-2312). Other available methods for detecting genomic methylation are described, for example, in U.S. Patent Application Publication 2003 / 0170684 A1 or WO 04 / 05122. The references cited above are incorporated herein by reference.

[0073] Methylation of CpG dinucleotide sequences can also be measured using a cytosine-based conversion technique that relies on methylation-dependent chemical modification of the CpG sequence within isolated genomic DNA or fragments thereof, followed by DNA sequence analysis. Chemical reagents capable of distinguishing between methylated and unmethylated CpG dinucleotide sequences include hydrazine and bisulfite, which cleave nucleic acids. Bisulfite treatment, followed by alkaline hydrolysis, specifically converts unmethylated cytosine to uracil without modifying 5-methylcytosine, as described in Olek A., 1996, Nucleic Acids Res. 24: 5064-6 or Frommer et al., 1992, Proc. Natl. Acad. Sci. USA 89: 1827-1831, each of which is incorporated herein by reference. Bisulfite-treated DNA can then be analyzed using molecular techniques such as PCR amplification, sequencing, and detection including oligonucleotide hybridization (e.g., using nucleic acid microarrays).

[0074] Techniques for analyzing bisulfite-treated DNA may employ methylation-sensitive primers to analyze the CpG methylation status of isolated genomic DNA, for example, as described below: Herman et al., 1996, Proc. Natl. Acad. Sci. USA 93: 9821-9826 or U.S. Patent Nos. 5,786,146 or 6,265,171, each of which is incorporated herein by reference. Methylation-sensitive PCR (MSP) allows the detection of specific methylated CpG sites, for example, within regulatory regions of genes. The DNA of interest is treated such that methylated and unmethylated cytosines are differentially modified, for example, by bisulfite treatment, in a manner distinguishable by their hybridization behavior. PCR primers specific to each of the methylated and unmethylated states of the DNA are used in PCR amplification. The products of the amplification reaction are then detected, allowing deduction of the methylation status of CpG sites within the genomic DNA. Other methods for analyzing bisulfite-treated DNA include methylation-sensitive single nucleotide primer extension (Ms-SNuPE) (see, for example, Gonzalgo and Jones, 1997; Nucleic Acids Res. 25: 2529-2531 or U.S. Patent 6,251,594, each of which is incorporated herein by reference), or the use of real-time PCR-based methods, such as the well-known fluorescence-based real-time PCR technology MethyLight. TM (See, for example, Eads et al., 1999; Cancer Res. 59: 2302-2306, U.S. Patent No. 6,331,393, or Heid et al., 1996, Genome Res. 6: 986-994, each of which is incorporated herein by reference.) It should be understood that various methylation assays can be used to determine the methylation status of specific genomic CpG sites. Methods employing bisulfite conversion include, for example, bisulfite sequencing, methylation-specific PCR, methylation-sensitive single nucleotide primer extension (Ms-SnuPE), MALDI mass spectrometry, and methylation-specific oligonucleotide arrays, for example, as described in U.S. Patent No. 7,611,869 or International Patent Application WO2004 / 051224, each of which is incorporated herein by reference.

[0075] In specific embodiments, methylation of genomic CpG sites in a sample can be detected using an array of probes. In such embodiments, multiple distinct probe molecules may be attached to a substrate or otherwise spatially differentiated within the array. Exemplary arrays that may be used in this invention include, but are not limited to, slide arrays, silicon wafer arrays, liquid arrays, bead-based arrays, and other arrays known in the art or otherwise detailed herein. In preferred embodiments, the method of this invention can be practiced using a combination of miniaturized array platforms, high levels of assay reuse, and scalable automation of array techniques for sample handling and data processing. Particularly useful arrays are described in U.S. Patent Nos. 6,355,431; 6,429,027; 6,890,741; 6,913,884 or 7,582,420; or U.S. Patent Application Publication Nos. 2002 / 0102578 A1; 2005 / 0053980 A1; 2005 / 0181440 A1; or 2009 / 0186349 A1, each of which is incorporated herein by reference. Other examples of available arrays include those described in U.S. Patent Nos. 6,023,540, 6,200,737 or 6,327,410; or PCT Publication Nos. WO9840726, WO9918434 or WO9850782, each of which is incorporated herein by reference.

[0076] The complexity of the arrays used in this invention can vary depending on the probe composition and the desired use of the array. For example, the complexity of the nucleic acids (or CpG sites) detected in the array can be at least 10, 100, 1,000, 10,000, 0.1 million, 1 million, 10 million, 100 million, or more. Alternatively or additionally, the complexity can be selected to be no more than 100 million, 10 million, 1 million, 0.1 million, 10,000, 1,000, 100, or less. Of course, the complexity can be between one of the lower limits and one of the upper limits selected from the above ranges. Similar complexity ranges can be achieved using nucleic acid sequencing methods, such as those known in the art, such as next-generation sequencing or massively parallel sequencing.

[0077] Various commercially available bead-based products for detecting methylation can be used, including, for example, MethylationEPIC. TM BeadChip TM(Illumina, San Diego, CA) allows for the quantitative interrogation of over 850,000 methylation sites across the human genome at single-nucleotide resolution. Methylation microarrays are also available from Agilent (Santa Clara, CA) and other commercial suppliers of nucleic acid arrays. Array products can be customized for detecting a wide variety of methylation sites in the human genome or other genomes.

[0078] The detection of one or more nucleic acids obtained or generated in the techniques listed herein can be performed using sequencing procedures, such as sequencing-by-synthesis (SBS) or other techniques known in the art, such as large-scale sequencing or next-generation sequencing. In short, SBS can be initiated by contacting the target nucleic acid with one or more labeled nucleotides, DNA polymerase, etc. The target nucleic acid may be derived from methylation detection techniques, such as bisulfite conversion or restriction using a methyl-sensitive restriction endonuclease. Those features used as templates to extend the primers are incorporated into the detectable labeled nucleotides. Optionally, the labeled nucleotides may also include reversible termination properties that terminate further primer extension once the nucleotide is added to the primer. For example, a nucleotide analog with a reversible termination moiety may be added to the primer, preventing subsequent extension until a deblocking agent is delivered to remove the moiety. Thus, for embodiments using reversible termination, a deblocking agent may be delivered to a flow cell (before or after detection). Washing may be performed between the various delivery steps. The cycle may then be repeated n times to extend the primer by n nucleotides, thereby detecting a sequence of length n. Exemplary SBS procedures, fluid systems, and detection platforms readily applicable to the methods of this disclosure are described, for example, in Bentley et al., Nature 456:53-59 (2008), WO 04 / 018497; WO 91 / 06678; WO 07 / 123744; U.S. Patent Nos. 7,057,026; 7,329,492; 7,211,414; 7,315,019 or 7,405,281, or U.S. Patent Application Publication No. 2008 / 0108082A1, each of which is incorporated herein by reference.

[0079] Other sequencing procedures, such as pyrosequencing, can be used to detect large numbers of nucleic acids in parallel. Pyrosequencing detects the release of inorganic pyrophosphate (PPi) when a specific nucleotide is incorporated into the nascent nucleic acid chain (Ronaghi et al., Analytical Biochemistry 242(1), 84-9 (1996); Ronaghi, Genome Res. 11(1), 3-11 (2001); Ronaghi et al. Science 281(5375), 363 (1998); or U.S. Patent Nos. 6,210,891; 6,258,568 or 6,274,320, each of which is incorporated herein by reference). Ligation-while-sequencing reactions are also available, including, for example, those described in: Shendure et al., Science 309:1728-1732 (2005); or U.S. Patent Nos. 5,599,675 or 5,750,341, each of which is incorporated herein by reference. Some embodiments may include sequencing-while-hybridization procedures, such as those described, for example, in Bains et al., Journal of Theoretical Biology 135(3), 303-7 (1988); Drmanac et al., Nature Biotechnology 16, 54-58 (1998); Fodor et al., Science 251(4995), 767-773 (1995); or WO 1989 / 10977, each of which is incorporated herein by reference. Techniques employing fluorescence resonant energy transfer (FRET) and / or zero-mode waveguides may be used, such as those described in: Levene et al., Science 299, 682-686 (2003); Lundquist et al., Opt. Lett. 33, 1026-1028 (2008); or Korlach et al., Proc. Natl. Acad. Sci. USA 105, 1176-1181 (2008), the contents of which are incorporated herein by reference. Alternatively, sequencing techniques that detect protons released during the incorporation of nucleotides into the extension product may be used, such as those commercially available from Ion Torrent (Guilford, CT, a subsidiary of Life Technologies) or described in U.S. Patent Application Publication Nos. 2009 / 0026082 A1; 2009 / 0127589 A1; 2010 / 0137143 A1; or 2010 / 0282617 A1, each of which is incorporated herein by reference.

[0080] Particularly useful sequencing platforms that can be used include those commercially available from Illumina (San Diego, CA), such as the MiSeq nucleic acid sequencer. TM NextSeq TM or HiSeq TM Series; the 454 sequencing system is commercially available from Roche Life Sciences (Basel, Switzerland); the Ion Torrent sequencing system is available from Life Technologies (a subsidiary of Thermo Fisher Scientific (Waltham, MA); or the nanopore sequencing system is commercially available from Oxford Nanopore (Oxford, England). TruSeq TM DNA methylation kits are available from Illumina and can be used to generate bisulfite sequencing libraries detectable on Illumina sequencers. Commercially available products for preparing nucleic acid samples for detection on sequencing platforms from Illumina or other vendors include, for example, methylation analysis sample preparation products available from Thermo Fisher Scientific (Waltham, MA). Methylation sequencing DNA library kit (Swift Biosciences, Ann Arbor, MI), available from New England BioLabs (Beverley, MA). Methylated DNA enrichment kit, Pico Methyl-Seq available from Zymoresearch (Irvine, CA) TM Library preparation kits, or methylamphenicol available from EpiGentek (Farmingdale, NY) TM Universal methylated DNA preparation kit.

[0081] Specific implementations may include steps of manipulating nucleic acid samples to enrich desired nucleic acids. For example, samples provided for use with the methods listed herein may undergo targeted selection of a subset of genomic DNA fragments containing a pre-determined set of target CpG sites. Targeted selection may occur before or after treatment of the nucleic acid with bisulfite, a methyl-sensitive endonuclease, or other reagents used to distinguish methylated sites from unmethylated sites. Available targeted selection techniques are listed in Example I below.

[0082] Specific implementations of the methods listed herein will evaluate and / or use coverage determined for each site for which methylation status has been or will be determined. In some cases, coverage data is provided to the individual or system implementing the methods. Alternatively, implementations of the methods may include one or more steps for determining coverage at each site.

[0083] For implementations of sequencing technologies to detect methylation status, coverage can be considered as describing the average number of sequencing reads aligned with or “covering” a specific site (e.g., a CpG site). Next-generation sequencing coverage levels often determine whether a particular sequence or site can be characterized with a certain confidence. At higher coverage levels, each site is covered by a greater number of aligned sequence reads, so characterization can be performed with higher confidence. Available guidance for determining coverage is provided in Illumina Technical Specification “Estimating Sequencing Coverage” Publication No. 770-2011-022 (December 1, 2014), which is incorporated herein by reference. Similar coverage criteria can be applied to other detection technologies besides next-generation sequencing.

[0084] Specific embodiments of the invention may use at least 10x, 30x, 50x, 100x, 1,000x, 5,000x, 10,000x or more coverage at each site. Alternatively or additionally, coverage may be at most 10,000x, 5,000x, 1,000x, 100x, 50x, 30x, 10x or less. Given the number of sites evaluated and the amount of DNA used in the method, coverage may be selected based on the confidence level required to determine the methylation pattern.

[0085] As the number of loci assessed increases, the confidence level of the locus characterization also increases. This means that lower coverage may be acceptable. In specific implementations, the number of loci assessed may be at least 10, 100, 500, or 1x10^6 loci. 3 Individual sites, 5x10 3 Individual sites, 1x10 4 Individual sites, 1x10 5 Individual sites, 1x10 6 One or more sites. Alternatively or additionally, the number of sites evaluated may be up to 1x10. 6 Individual sites, 1x10 5 Individual sites, 1x10 4 Individual sites, 1x10 3 One site, 100 sites, or 10 sites.

[0086] The amount of DNA used in the methods listed herein will depend on several factors, such as the sample used and the analytical steps performed on the sample. A typical blood draw will yield 30 ng of circulating DNA. However, larger or smaller amounts of DNA can be provided by varying the volume of blood drawn, by using different types of samples (such as those exemplified elsewhere herein), and / or by utilizing sample extraction techniques with higher or lower yields. Therefore, the methods of the present invention can be implemented using amounts of at least 3 ng, 10 ng, 30 ng, 50 ng, 100 ng, 500 ng, or more of DNA. Alternatively or additionally, the amount of DNA can be up to 500 ng, 100 ng, 50 ng, 30 ng, 10 ng, or 3 ng.

[0087] Furthermore, in some embodiments, the DNA used in the method for assessing methylation status is a mixture of DNA from the target cell or tissue against a background of DNA from other cells or tissues (e.g., non-tumor DNA). The percentage of DNA from the target tissue or cells can be up to 90%, 50%, 25%, 10%, 1%, 0.1%, 0.01%, or lower. Alternatively or additionally, the percentage of DNA from the target tissue or cells can be at least 0.01%, 0.1%, 1%, 10%, 25%, 50%, 90%, or higher.

[0088] The parameters above, such as the amount of DNA, coverage, number of sites, and percentage of DNA from the target cell or tissue, can be adjusted, for example, within the range of the examples above, to suit the required confidence level for characterizing the methylation state of nucleic acids in the methods listed herein.

[0089] Specific embodiments of the methods listed herein include the step of providing the methylation status of multiple sites in a reference genomic DNA from one or more reference individual organisms. Optionally, the method may include one or more steps for detecting the methylation status of multiple sites in the reference genomic DNA from one or more reference individual organisms. In one aspect, the reference genomic DNA may include, for example, a baseline sample. Any of the methods listed herein for determining the methylation status of test DNA may be used to determine the methylation status of reference DNA.

[0090] The reference genomic DNA (such as a baseline sample) used in the methods of this disclosure may be derived from one or more organisms of the same species as the test organism. For example, when the test organism is an individual human, the reference genomic DNA may be derived from different human individuals. In some embodiments, the reference genomic DNA is derived from the same individual from which the test genomic DNA material was provided. For example, the test DNA may be derived from tissue suspected of having a specific symptom, while the reference DNA is derived from tissue known not to have said symptom. In a specific embodiment, the test DNA may be derived from a tumor sample obtained from an individual, while the reference DNA is derived from normal tissue obtained from the same individual. The tissue or cell types may be the same, but in fact, one of the tissue or cell types may have a symptom that the other tissue or cell type does not have. Alternatively, different tissue or cell types may be obtained from an individual, one of which provides the test DNA, and the other tissue or cell type provides the reference DNA. The reference genomic DNA may be obtained from a metagenomic sample (e.g., an environmental or community sample), for example, for comparison with the test metagenomic sample.

[0091] Test DNA may be derived from one or more test organisms at a different time than reference DNA (such as a baseline sample) derived from said one or more test organisms. For example, a reference DNA sample may be obtained from an individual at a time before the suspected presence of a disease or symptom, and then a test DNA sample may be obtained from said individual at a time after the suspected disease or symptom. In such embodiments, the test DNA and reference DNA may be obtained from similar tissues, communities, or cell types or from different tissues, communities, or cell types.

[0092] In one embodiment, the method of this disclosure may include the step of determining methylation differences between test genomic DNA and reference genomic DNA for multiple sites (e.g., CpG sites), thereby providing a normalized methylation difference for each site (e.g., CpG site). In a specific embodiment, the normalized methylation difference (also referred to as z-score) at a particular site (e.g., CpG site) is determined according to the following formula.

[0093]

[0094] Z i χ represents the normalized methylation difference at a specific site identified as i. i μ represents the methylation level at site i in the tested genomic DNA. i σ represents the average methylation level at site i in the reference genome, and σ i This represents the standard deviation of the methylation level at site i in the reference genomic DNA. The use of the formula for determining methylation differences is illustrated in Example I below.

[0095] The method of this disclosure may further include the step of determining a total coverage-weighted normalized methylation difference score by weighting the normalized methylation difference at each site (e.g., CpG site) by coverage at each site (e.g., CpG site). In a specific embodiment, the total coverage-weighted normalized methylation difference score (denoted as A) is determined according to the following formula.

[0096]

[0097] Where w i Let represent the coverage at site i, and k represent the total number of sites. The use of the formula for determining the total coverage-weighted normalized methylation difference score is illustrated in Example I below.

[0098] In specific implementations, the methods described herein can be used to identify changes in the methylation status of a test organism or to monitor such changes over time. Therefore, this disclosure provides a method comprising the steps of: (a) providing a test dataset comprising (i) methylation status at multiple sites of test genomic DNA from at least one test organism, and (ii) coverage at each site for detecting methylation status; (b) providing methylation status at multiple sites of reference genomic DNA from one or more reference individual organisms; (c) determining methylation differences between the test genomic DNA and the reference genomic DNA for each site, thereby providing a normalized methylation difference at each site; (d) weighting the normalized methylation differences at each site by coverage at each site, thereby determining a total coverage-weighted normalized methylation difference score; and (e) repeating steps (a) to (d) using a second test dataset and using the same reference genomic DNA from at least one reference individual, the second test dataset comprising (i) methylation status at multiple sites of a second test genomic DNA from an individual test organism, and (ii) coverage at each site for detecting methylation status; and (f) determining whether the total coverage-weighted normalized methylation difference score between the test genomic DNA and the second test genomic DNA has changed.

[0099] A method is also provided, comprising the steps of: (a) providing a sample containing a mixture of genomic DNA from multiple different cell types of at least one test organism, thereby providing test genomic DNA; (b) detecting the methylation status of multiple sites in the test genomic DNA; (c) determining coverage at each site for detecting methylation status; (d) providing the methylation status of multiple sites in reference genomic DNA from at least one reference individual, wherein the at least one test organism and the reference individual are optionally of the same species; (e) determining the methylation difference between the test genomic DNA and the reference genomic DNA for each site, thereby providing a normalized methylation difference for each site; (f) weighting the normalized methylation difference for each site by coverage at each site, thereby determining a total coverage-weighted normalized methylation difference score; (g) repeating steps (a) to (f) using a second test genomic DNA provided from a sample containing a mixture of genomic DNA from multiple different cell types of at least one test organism and using the same reference genomic DNA from at least one reference individual; and (h) determining whether the total coverage-weighted normalized methylation difference score between the test genomic DNA and the second test genomic DNA has changed.

[0100] In another embodiment, the method is improved to take into account changes observed in the total DNA methylation within the normal population. The test genomic DNA is not directly compared to a reference genomic DNA; instead, an intermediate step is inserted to evaluate the changes in total DNA methylation within the normal population against the reference genomic DNA (referred to as the baseline sample in this embodiment). This involves calculating the “methylation score” for each member of the training set of normal genomic DNA samples, and determining the mean and standard deviation of the methylation scores for the training set, thereby generating information about the distribution of methylation scores in the normal population. In some embodiments, the number of normal individual organisms providing the genomic DNA for the training set is at least 3, at least 5, at least 10, at least 20, at least 50, or at least 100.

[0101] In this embodiment, the method may include determining the population-observed average methylation level (μ) for each CpG site i against a reference genomic DNA. i ) and the standard deviation of methylation level (σ) iThe first step is to take a population of reference or baseline genomic DNA samples. The selected genomic DNA can then be compared to the baseline DNA population to assess changes in methylation levels. More specifically, the methylation level at each site i (e.g., CpG site) in the selected genomic DNA can be compared to the population mean μ of the baseline sample. i A comparison is made to generate a methylation fraction of the selected genomic DNA. In one embodiment, the selected genomic DNA is a set of training controls, and in another embodiment, the selected genomic DNA is test genomic DNA. The methylation level can be determined by conventional methods known to those skilled in the art. For example, the methylation level can be calculated as the fraction of 'C' bases at the target CpG site in the 'C'+'U' bases after bisulfite treatment, or the fraction of 'C' bases at the target CpG site in the total 'C'+'T' bases after bisulfite treatment and subsequent nucleic acid amplification, as described herein.

[0102] The methylation fraction (MS) of the selected genomic DNA can be calculated by: determining the normalized methylation difference (z-score) at a specific site i (e.g., a CpG site) using a set of baseline samples; converting the z-score at each site into a probability of observing such or larger z-scores (e.g., a one-sided p-value); and combining the p-values ​​to form the final total methylation fraction. Optionally, the p-values ​​are weighted. Each of these steps is described in detail herein and immediately below.

[0103] The methylation fraction was initially determined for a training set of normal genomic DNA samples. First, the normalized methylation difference (z-score) at a specific site i (e.g., a CpG site) was determined according to the following formula.

[0104]

[0105] Z i χ represents the normalized methylation difference at a specific site identified as i. i μ represents the methylation level at site i in the training set of members of the normal genomic DNA. i σ represents the average methylation level at site i in the baseline sample, and σ i This represents the standard deviation of the methylation level at site i in the baseline sample.

[0106] Then the z-score (Z) of each CpG site i is calculated. i This is transformed into the probability of observing such or larger z-scores. On one hand, this probability is achieved by converting the z-score into a one-sided p-value (p...). iThis can be used to calculate the probability of a normal, t-, or binomial distribution. The statistical tools used for such calculations are well-known and readily available to those skilled in the art.

[0107] Next, according to Fisher's formula, the methylation fraction (MS) is determined by combining p-values, which is the sum of the probabilities of observed normalized methylation differences:

[0108]

[0109] Where p i represents the one-sided p-value at site i, and k represents the total number of sites. The methylation score is calculated for each member of the training set of normal genomic DNA.

[0110] Optionally, the p value at each CpG site can be obtained by taking the p value at each CpG site i (p i ) and weighting factor w i Multiply to weight, where w i This can correspond to the significance of CpG sites obtained from prior knowledge, the depth of coverage associated with the site, or any other ordination method. In this regard, according to Fisher's formula, the methylation fraction (denoted as MS) is determined by combining weighted p-values:

[0111]

[0112] Where p i Let represent the one-sided p-value at site i, k represent the total number of sites, and w i The significance of site i is indicated, for example, coverage. The use of this formula for determining the weighted methylation fraction of the training set for normal genomic DNA samples is shown in Example III.

[0113] Then, statistical analysis of the methylation fractions of the training set was performed. The mean methylation fraction (μ) of the training set of normal genomic DNA was calculated. MS ) and the standard deviation of methylation fraction (σ) MS This characterizes the distribution of methylation fractions in the normal population and can be used to determine whether the genomic DNA of a test genomic sample has abnormal methylation levels.

[0114] Then, the methylation fraction (MS) of the test genomic DNA is determined with reference to baseline samples (as described above for members of the training set) and compared with the distribution of methylation fractions determined for the training set of normal genomic DNA.

[0115] As described above in conjunction with the training set, the normalized methylation difference (z-score) at a specific site i (e.g., a CpG site) is first determined according to the following formula.

[0116]

[0117] Z i χ represents the normalized methylation difference at a specific site identified as i. i μ represents the methylation level at site i in the tested genomic DNA. i σ represents the average methylation level at site i in the baseline sample, and σ i This represents the standard deviation of the methylation level at site i in the baseline sample.

[0118] Then the z-score (Z) of each CpG site i is calculated. i This is transformed into the probability of observing such or larger z-scores. On one hand, this probability is achieved by converting the z-score into a one-sided p-value (p...). i This is used to calculate the probability of a normal, t-, or binomial distribution. Based on Fisher's formula, the methylation fraction (MS) of the tested genomic DNA is determined by combining p-values:

[0119]

[0120] Where p i Let represent the one-sided p-value at site i, and k represent the total number of sites.

[0121] Optionally, the p value at each CpG site can be obtained by taking the p value at each CpG site i (p i ) and weight w i Multiply to weight, where w i This can correspond to the significance of CpG sites obtained from prior knowledge, the depth of coverage associated with the site, or any other sequencing method. According to Fisher's formula, the methylation fraction (MS) of the test genomic DNA is determined by combining weighted p-values:

[0122]

[0123] Where p i Let represent the one-sided p-value at site i, k represent the total number of sites, and w i The significance of site i is indicated, for example, coverage. The use of this formula for determining the weighted methylation fraction of a test genomic DNA sample is shown in Examples II and III.

[0124] Finally, the average methylation fraction (μ) of the training set composed of normal genomic DNA was analyzed. MS ) and the standard deviation of methylation fraction (σ) MSThe distribution of methylation fractions determined for the training set is represented by ), which evaluates the methylation fraction of the test genomic DNA. The methylation fraction of the test genomic DNA is determined by the average methylation fraction (μ) of the training set and the normal genomic DNA using the following formula. MS The number of standard deviations

[0125]

[0126] Z MS The difference in normalized methylation fraction is represented by MS, which represents the methylation fraction of the test sample, in μ. MS σ represents the average methylation fraction of the training set of normal genomic DNA, and σ MS The standard deviation represents the methylation fraction of the training set of normal genomic DNA. The use of this formula to determine the difference in normalized methylation fractions is shown in Example III. Z-squared values ​​greater than 1.5, greater than 2, greater than 2.5, or greater than 3 standard deviations are... MS The value indicates an abnormal level of DNA methylation in the tested genomic DNA. In a preferred embodiment, a Z-value greater than 3 standard deviations... MS The value is used as an indicator of abnormal DNA methylation levels in genomic DNA.

[0127] In another embodiment, the methods listed herein can be used to identify changes in the methylation status of a test organism or to monitor such changes over time. Therefore, this disclosure provides a method comprising the steps of: (a) providing the methylation status of multiple sites (e.g., CpG sites) in baseline genomic DNA from two or more normal individual organisms; (b) determining, for each site (e.g., CpG site), the average methylation level and the standard deviation of the methylation level of the baseline genomic DNA; (c) providing a test dataset comprising (i) the methylation status of multiple sites (e.g., CpG sites) in a first test genomic DNA from at least one test organism, and optionally (ii) coverage at each site (e.g., CpG site) for detecting the methylation status; and (d) determining the methylation status between the first test genomic DNA and the baseline genomic DNA for each site (e.g., CpG site). (e) Normalizing the methylation difference of the first test genomic DNA to provide a normalized methylation difference; (e) converting the normalized methylation difference of the first test genomic DNA at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation difference, and optionally weighting the probability of such an event; (f) determining the methylation fraction of the first test genomic DNA; (g) repeating steps (c) to (f) using a second test genomic DNA provided from a sample containing a mixture of genomic DNA from multiple different cell types from at least one test organism and using the same baseline genomic DNA; and (h) determining whether the methylation fraction between the first test genomic DNA and the second test genomic DNA has changed.

[0128] Alternative methods for monitoring changes in DNA methylation over time include the following steps: (a) providing the methylation status of multiple sites (e.g., CpG sites) in baseline genomic DNA from two or more normal individual organisms; (b) determining the average methylation level and standard deviation of the methylation level of the baseline genomic DNA for each site (e.g., CpG site); (c) providing a mixture of genomic DNA from a test organism suspected of having a condition (e.g., cancer) associated with abnormal DNA methylation levels, wherein the mixture contains genomic DNA from multiple different cell types of the test organism, thereby providing a first test genomic DNA; (d) detecting the methylation status of multiple sites (e.g., CpG sites) in the first test genomic DNA, and optionally determining coverage at each site (e.g., CpG site) used for detecting methylation status; (e) (c) Determine the methylation difference between the first test genomic DNA and the baseline genomic DNA for each site (e.g., CpG site) to provide a normalized methylation difference of the first test genomic DNA; (f) convert the normalized methylation difference of the first test genomic DNA at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation differences, and optionally weight the probability of such events; (g) determine the methylation fraction of the first test genomic DNA; (h) repeat steps (c) to (g) using a second test genomic DNA provided from a sample containing a mixture of genomic DNA from multiple different cell types from at least one test organism and using the same baseline genomic DNA; and (i) determine whether the methylation fraction between the first test genomic DNA and the second test genomic DNA has changed.

[0129] The first and second test genomic DNA samples (or test datasets) compared in the methods listed herein may be derived from the same type of cells, communities, tissues, or fluids, but at different time points. Therefore, the methods listed herein can be used to identify or monitor changes over time. In some embodiments, the different time points may occur before, during, and / or after a particular treatment. For example, in the case of monitoring or predicting cancer, samples may be obtained from the individual before and after the start of treatment (such as surgery, chemotherapy, or radiation therapy). Furthermore, multiple samples may be obtained at different time points during treatment. For example, samples may be obtained and evaluated at time points throughout surgery (e.g., to assess whether surrounding cancerous tissue has been cleared) or at different time points throughout chemotherapy or radiation therapy. Different samples may be obtained from the individual and tested after treatment, for example, to test for recurrence and remission.

[0130] In another example, gut metagenomic samples can be obtained before and after treatment (e.g., for a digestive condition). The methylation status of the samples can be assessed and compared to identify changes in the gut microbiota caused by treatment. These changes can then be used to monitor treatment and determine the prognosis of the treated individual.

[0131] Any of the various sample types containing tumor DNA listed herein or known in the art can be used in methods for identifying or monitoring changes in an individual's methylation status. Observed changes can provide a basis for diagnosis, prognosis, or screening of individuals with specific conditions such as cancer.

[0132] The methods listed herein can also be used to screen or test candidate treatments, such as in experimental cell cultures, tissues, or organisms. Therefore, the methods listed herein can be used to identify or monitor changes over time in cell cultures, tissues, or organisms being tested in a clinical or laboratory setting. In some embodiments, different time points can occur before, during, and / or after a particular candidate treatment. For example, samples can be obtained from a test organism before and after the initiation of a candidate treatment (such as surgery, chemotherapy, or radiation therapy). Furthermore, multiple samples can be obtained at different time points during the candidate treatment. For example, samples can be obtained and evaluated at time points throughout the surgical procedure (e.g., to assess whether surrounding cancerous tissue has been removed) or at different time points throughout the candidate chemotherapy or radiation therapy. Different samples can be obtained from the test organism and tested after the candidate treatment, for example, to assess recurrence and remission. Control organisms that have not received the candidate treatment and / or do not have a specific condition can also be tested using similar methods. Comparison of results between samples treated with the candidate treatment and controls can be used to determine the efficacy and / or safety of a particular candidate treatment.

[0133] Any of the various sample types containing tumor DNA listed herein or known in the art can be used in methods for identifying or screening candidate treatments. Variations can be used to assess the efficacy and / or safety of a particular candidate treatment, regardless of whether they are compared to a specific control.

[0134] In a specific embodiment, this disclosure provides a method for detecting a condition such as cancer. The method may include the following steps: (a) providing a mixture of genomic DNA from an individual suspected of having the condition (e.g., cancer), wherein the mixture contains genomic DNA from multiple different cell types of the individual, thereby providing test genomic DNA; (b) detecting the methylation status of multiple sites (e.g., CpG sites) in the test genomic DNA; (c) determining coverage at each site (e.g., CpG site) used to detect the methylation status; and (d) providing the methylation status of multiple sites (e.g., CpG sites) in reference genomic DNA from at least one reference individual known to have the condition. (e) or known not to have the condition (e.g., cancer); (e) determining the methylation difference between the test genomic DNA and the reference genomic DNA for each site (e.g., CpG site), thereby providing a normalized methylation difference for each site (e.g., CpG site); (f) weighting the normalized methylation difference for each site (e.g., CpG site) by coverage at each site (e.g., CpG site), thereby determining a total coverage-weighted normalized methylation difference score; and (g) determining whether the individual has or does not have the condition (e.g., cancer) based on the total coverage-weighted normalized methylation difference score. In some embodiments, the sample is blood, and the DNA may include, for example, cell-free DNA from blood.

[0135] A method for identifying changes in a symptom such as cancer is also provided. The method may include the following steps: (a) providing a mixture of genomic DNA from an individual suspected of having the symptom (e.g., cancer), wherein the mixture contains genomic DNA from multiple different cell types of the individual, thereby providing test genomic DNA; (b) detecting the methylation status of multiple sites (e.g., CpG sites) in the test genomic DNA; (c) determining coverage at each site (e.g., CpG site) used to detect the methylation status; (d) providing the methylation status of multiple sites (e.g., CpG sites) in reference genomic DNA from at least one reference individual, the reference individual being known to have the symptom (e.g., cancer) or known not to have the symptom (e.g., cancer); and (e) determining the methylation status of the test genomic DNA relative to the reference genomic DNA for each site (e.g., CpG site). (a) methylation differences between sites, thereby providing a normalized methylation difference for each site (e.g., CpG site); (f) weighting the normalized methylation difference for each site (e.g., CpG site) by coverage at each site (e.g., CpG site) to determine a total coverage-weighted normalized methylation difference score; and (g) repeating steps (a) to (f) using a second mixture of genomic DNA from an individual suspected of having the condition (e.g., cancer) and using the same reference genomic DNA from at least one reference individual, and (h) determining whether the total coverage-weighted normalized methylation difference score of the second test genomic DNA has changed compared to the test genomic DNA, thereby determining whether the condition (e.g., cancer) has occurred or not changed based on the total coverage-weighted normalized methylation difference score.

[0136] In a specific embodiment, this disclosure provides a method for detecting symptoms such as cancer. The method may include the following steps: (a) providing the methylation status of multiple sites (e.g., CpG sites) in baseline genomic DNA from at least one normal individual organism; (b) determining the average methylation level and standard deviation of the methylation level of the baseline genomic DNA for each site (e.g., CpG site); (c) providing a training set of normal genomic DNA samples from two or more normal individual organisms, comprising (i) the methylation status of multiple sites (e.g., CpG sites) in the training set of normal genomic DNA samples, and optionally (iI) coverage at each site (e.g., CpG site) for detecting methylation status; (d) For each site (e.g., CpG site), determine the methylation difference between each normal genomic DNA sample in the training set and the baseline genomic DNA, thereby providing the normalized methylation difference for each normal genomic DNA sample in the training set at each site (e.g., CpG site); (e) convert the normalized methylation difference for each normal genomic DNA sample in the training set at each site (e.g., CpG site) into the probability of observing such or greater normalized methylation differences, and optionally weight the probability of such events; (f) determine the methylation fraction for each normal genomic DNA sample in the training set to obtain the training set methylation fraction; (g) calculate the training... The test genomic DNA is prepared by: (i) collecting the average methylation fraction and standard deviation of the methylation fraction; (h) providing a mixture of genomic DNA from a test organism suspected of having the condition (e.g., cancer), wherein the mixture contains genomic DNA from multiple different cell types of the test organism; (i) detecting the methylation status at multiple sites (e.g., CpG sites) in the test genomic DNA, and optionally determining the coverage at each site (e.g., CpG site) used to detect the methylation status; and (j) determining the methylation difference between the test genomic DNA and baseline genomic DNA for each site (e.g., CpG site), thereby providing a test genomic DNA. (k) Normalized methylation difference of genomic DNA; (k) Converting the normalized methylation difference of the test genomic DNA at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation difference, and optionally weighting the probability of such an event; (l) Determining the methylation fraction of the test genomic DNA; and (m) Comparing the methylation fraction of the test genomic DNA with the mean methylation fraction and the standard deviation of the methylation fraction in the training set of normal genomic DNA to determine the number of standard deviations of the distribution of the methylation fraction of the test genomic DNA with that of the methylation fraction in the training set of normal genomic DNA.In cases where the number of standard deviations exceeds a predetermined threshold (e.g., 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, etc.), the test sample is considered to have an abnormal level of DNA methylation.

[0137] Optionally, the sites from the test genomic DNA are derived from multiple different cell types from the individual test organism, and alternatively, the cell type from which each site (e.g., a CpG site) is derived is unknown. In another optional embodiment, the individual test organism and one or more baseline individual organisms, training individual organisms, or combinations thereof are of the same species. In some embodiments, the mixture of genomic DNA from an individual suspected of having said disease is blood, and the DNA may, for example, include cell-free DNA (cfDNA) or circulating tumor DNA (ctDNA) from blood.

[0138] This document also provides a method for identifying changes in a symptom such as cancer over time. The method may include the following steps: (a) providing the methylation status of multiple sites (e.g., CpG sites) in baseline genomic DNA from at least one normal individual organism; (b) determining the mean methylation level and standard deviation of the methylation level of the baseline genomic DNA for each site (e.g., CpG site); (c) providing a first mixture of genomic DNA from a test organism suspected of having the symptom (e.g., cancer), wherein the first mixture contains genomic DNA from multiple different cell types of the test organism, thereby providing a first test genomic DNA; (d) detecting the methylation status of multiple sites (e.g., CpG sites) in the first test genomic DNA, and optionally determining coverage at each site (e.g., CpG site) used for detecting the methylation status; and (e) determining the first test genomic DNA for each site (e.g., CpG site). (c) methylation differences between the first test genomic DNA and the baseline genomic DNA, thereby providing normalized methylation differences of the first test genomic DNA; (f) converting the normalized methylation differences of the first test genomic DNA at each site (e.g., CpG site) into a probability (e.g., a one-sided p-value) of observing such or greater normalized methylation differences, and optionally weighting the probability of such events; (g) determining the methylation fraction of the first test genomic DNA; (h) repeating steps (c) to (g) using a second mixture of genomic DNA from a test organism suspected of having the condition (e.g., cancer), wherein the second mixture contains the second test genomic DNA; and (i) determining whether the methylation fraction of the second test genomic DNA has changed compared to the first test genomic DNA, thereby determining whether the condition (e.g., cancer) has occurred or not based on the change in methylation fraction.

[0139] The methylation status determined using the methods listed in this paper can be used for molecular classification and cancer prediction using guidelines developed for gene expression and other genomic data (see, for example, Golub et al. (1999) Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science, 286, 531-537.). Other classification systems that may be used include those developed for molecular classification in breast cancer based on relevant overall changes in methylation patterns (see, for example, Huang et al. (1999) Methylation profiling of CpG sites in human breast cancer cells. Hum Mol Genet, 8, 459-470) or those developed for relevant methylation patterns of tumor suppressor genes (e.g., p16, a cyclin-dependent kinase inhibitor) in certain human cancer types (see, for example, Herman et al. (1995) Inactivation of the CDKN2 / p16 / MTS1 gene is frequently associated with aberrant DNA methylation in all common human cancers. Cancer Res, 55, 4525-4530.; Otterson et al. (1995) CDKN2 gene silencing in lung cancer by DNA hypermethylation and kinetics of p16INK4 protein induction by 5-aza 2′deoxycytidine. Oncogene, 11, 1211-1216.). The above references are incorporated into this paper by way of citation.

[0140] In some applications of the methylation analysis methods listed herein, models can be developed to predict disease types without prior pathological diagnosis. Therefore, in some embodiments, the methods listed herein are used to determine the methylation pattern of staged tumor samples relative to matched normal tissue from the same patient. The determined differences in methylation patterns between tumor and normal tissue can be used to build models for predicting, diagnosing, or monitoring cancer. For example, the methylation pattern determined for a test sample can be compared with methylation patterns from known normal tissue and / or from known tumors, and a diagnosis can be made based on the degree of similarity between the test sample and one or both of these references.

[0141] Furthermore, the methods described in this article can facilitate tumor identification, classification, and prognostic assessment. This information can then be used to identify subgroups of tumors with relevant characteristics. This classification can be used to determine the etiology of various cancer types and predict their clinical behavior.

[0142] In specific embodiments of the method of the present invention, cancer is predicted, detected, identified, classified, or monitored from cell-free DNA of cancer patients. For example, determining the methylation pattern from a plasma sample can be used for cancer screening. When the methylation pattern of a plasma sample is abnormal compared to a healthy reference, cancer may be suspected. Further confirmation and assessment of the cancer type or tissue origin of the cancer can then be performed by determining a plasma profile of methylation at different genomic loci or by detecting tumor-related copy number aberrations, chromosomal translocations, and single nucleotide variants through plasma genomic analysis. Alternatively, radiological and imaging investigations (e.g., computed tomography, magnetic resonance imaging, positron emission tomography) or endoscopy (e.g., upper gastrointestinal endoscopy or colonoscopy) can be used to further investigate individuals suspected of having cancer based on plasma methylation level analysis.

[0143] In one aspect of the invention, this document provides a method for identifying or classifying specific cancer types in a test organism, preferably a mammalian organism, and more preferably a human, using methylation levels. In this aspect, the methylation level of the test genomic DNA is evaluated for a subset of preselected methylation sites (referred to herein as “hypermethylated” sites) associated with known cancer types, and then sorted from low to high. The cancer type corresponding to the highest average methylation level is considered to be associated with the test genomic DNA, i.e., said cancer type is considered to be present in the test organism.

[0144] As a starting point, the method may include identifying a specific cancer that can be used as a cancer type in an identification or classification algorithm according to this aspect of the invention. A cancer type is a member of a specific cancer group that can be used to determine whether a test organism has a specific cancer type, such as invasive breast cancer, colon adenocarcinoma, lung adenocarcinoma, etc.

[0145] Determining whether a cancer is suitable for use as a cancer type in the methods of this invention includes obtaining genomic DNA sequence data from clinical samples. Genomic DNA sequence data suitable for use herein can be readily obtained from known databases characterizing genomic and epigenomic changes (such as changes in methylation status) in different cancer types. The larger the number of clinical samples of cancer in the database, the more likely said cancer is to be used as a cancer type. Cancer types suitable for use in the methods of this invention can be defined using genomic DNA sequence data from at least 10, at least 15, at least 20, at least 25, at least 30, at least 40, at least 50, at least 75, or at least 100 clinical samples of a particular cancer.

[0146] Once the appropriate cancer type groups are defined, a list of so-called “hypermethylated” sites specific to each cancer type is compiled. In some embodiments, available methylation sites that can be assessed against methylation status include selected CpG sites from pan-cancer groups listed in Table I (methylation sites listed from Genome Construction 37) and / or listed in Table II (methylation sites listed from Genome Construction 37). In other embodiments, among others, available methylation sites that can be assessed against methylation status include those present in the Cancer Genome Atlas (see, for example, Cancer Genome Atlas Research Network et al., Nature Genetics 45:1113-1120 (2013)), CpG sites described in Worthley et al., Oncogene 29, 1653-1662 (2010) for identifying or monitoring colorectal cancer, and methylation markers listed in U.S. Patent Application Publication No. 2008 / 0166728A1 for detecting ovarian cancer. All cited references are incorporated herein by reference in their entirety. All or a subset of the references listed or included in this paper may be used for the identification or classification methods outlined herein. For example, at least 100, 1x10 3 1 x 10 4 1 x 10 5 1 x 10 6 One or more methylation sites can be used as starting points. In some embodiments, the entire methylome (i.e., the complete set of methylation sites in the genome of the test organism) can be used to select hypermethylation sites suitable for the methods of the present invention.

[0147] Table I: Pan-Cancer Group

[0148]

[0149]

[0150]

[0151]

[0152]

[0153]

[0154]

[0155]

[0156]

[0157]

[0158]

[0159]

[0160]

[0161]

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174]

[0175]

[0176]

[0177]

[0178]

[0179]

[0180]

[0181]

[0182]

[0183]

[0184]

[0185]

[0186]

[0187]

[0188]

[0189]

[0190]

[0191]

[0192]

[0193]

[0194]

[0195]

[0196]

[0197]

[0198]

[0199]

[0200]

[0201]

[0202]

[0203]

[0204]

[0205]

[0206]

[0207]

[0208] Table II: CRC Groups

[0209]

[0210]

[0211]

[0212]

[0213]

[0214]

[0215]

[0216] The selection of hypermethylation sites according to the method of the present invention is defined as follows. For each clinical sample of a specific cancer type in a database (e.g., a cancer genome atlas), the methylation level is determined for each methylation site i from a starting set of sites as described in the preceding paragraph. For example, for each clinical sample from a set of colon adenocarcinoma samples in a cancer genome atlas, the methylation status at each CpG site listed in Tables I and II is determined, and the average methylation level at each site i is calculated as described elsewhere in this application. In some embodiments, the methylation level may be determined as the fraction of 'C' bases in the total 'C'+'U' bases at the target CpG site i after bisulfite treatment. In other embodiments, the methylation level may be determined as the fraction of 'C' bases in the total 'C'+'T' bases at site i after bisulfite treatment and subsequent nucleic acid amplification. The average methylation level at each site is then evaluated to determine whether one or more thresholds are met. In some embodiments, the thresholds are selected from those sites that have the highest ranked average methylation values ​​for a specific cancer type. For example, the threshold could be those sites that have an average methylation level in the top 50%, top 40%, top 30%, top 20%, top 10%, top 5%, top 4%, top 3%, top 2%, or top 1% of the average methylation level across all sites i tested for a specific cancer type (e.g., colon adenocarcinoma). Alternatively, the threshold could be those sites that have an average methylation level greater than or equal to 50, 60, 70, 80, 90, 95, 96, 97, 98, or 99 percentiles. In other embodiments, the threshold could be based on the absolute value of the average methylation level. For example, the threshold could be those sites with an average methylation level greater than 99%, 98%, 97%, 96%, 95%, 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, or 2%. Relative and absolute thresholds can be applied individually or in combination to the average methylation level at each site i. As an illustration of combined threshold application, a subset of sites can be selected from the top 3% of all sites tested for average methylation level and also have an absolute average methylation level greater than 6%. The result of this selection process is multiple lists of specific hypermethylated sites (e.g., CpG sites) considered most informative for this cancer type, each list for each cancer type. These lists are then used to identify or classify test genomic DNA samples from the test organism, i.e., to determine whether the test organism has a specific cancer type.

[0217] In the next step of the method of the present invention, a test genomic DNA sample from a test organism is analyzed by determining the methylation level at each site i on a list of hypermethylation sites for each cancer type, and then averaging these methylation levels at each site to calculate the average methylation level across hypermethylation sites for each cancer type. For example, for each hypermethylation site i of colon adenocarcinoma, the methylation level at each site i on a list of hypermethylation sites for colon adenocarcinoma is determined, and these methylation levels are then averaged to provide a single average methylation level. This process is repeated using the results of a previously defined list of hypermethylation sites for each cancer type and a set of average methylation levels, each corresponding to a different cancer type. The average methylation levels are then sorted from low to high. The cancer type corresponding to the highest average methylation level is considered to be associated with the test genomic DNA, i.e., said cancer type is considered to be present in the test organism. It should be understood that normalized methylation difference or z-score may also be used in place of the methylation level at each CpG site in the method of the present invention.

[0218] For cancer screening or detection, determining the methylation level of a plasma (or other biological) sample can be used in conjunction with other forms of measurement for cancer screening or detection, such as prostate-specific antigen (PSA) measurements (e.g., for prostate cancer), carcinoembryonic antigen (CEA) (e.g., for colorectal cancer, gastric cancer, pancreatic cancer, lung cancer, breast cancer, medullary thyroid carcinoma), alpha-fetoprotein (AFP) (e.g., for liver cancer or germ cell tumors), and CA19-9 (e.g., for pancreatic cancer).

[0219] Available methylation sites that can be detected in the methods listed herein to assess cancer include those present in the Cancer Genome Atlas (see, for example, Cancer Genome Atlas Research Network et al., Nature Genetics 45: 1113-1120 (2013)) or selected CpG sites from the pan-cancer group listed in Table I (methylation sites listed from Genome Construction 37). Further examples of CpG sites that can be used, for example, to identify or monitor colorectal cancer are described in Worthley et al., Oncogene 29, 1653-1662 (2010) or listed in Table II (methylation sites listed from Genome Construction 37). Available methylation markers for detecting ovarian cancer are listed in U.S. Patent Application Publication No. 2008 / 0166728 A1, which is incorporated herein by reference. All or a subset of the markers listed herein and / or listed in the foregoing references may be used in the methods listed herein. For example, at least 10, 25, 50, 100, or 1x10 CpG sites may be used. 3 1 x 10 4 One or more markers.

[0220] Analysis of methylation, prognostic, or diagnostic information derived from the methods listed herein can be conveniently performed using a variety of computer-executable algorithms and programs. Therefore, some embodiments employ processing involving data stored in or transmitted through one or more computer systems or other processing systems. Embodiments of the invention also relate to apparatus for performing these operations. This apparatus may be specifically configured for the claimed purpose, or it may be a general-purpose computer (or a group of computers) selectively activated or reconfigured by computer programs and / or data structures stored in a computer. In some embodiments, a group of processors performs some or all of the listed analytical operations collaboratively (e.g., via a network or cloud computing) and / or in parallel. The processor or group of processors used to perform the methods described herein can be of various types, including microcontrollers and microprocessors such as programmable devices (e.g., CPLDs and FPGAs) and non-programmable devices such as gate arrays (ASICs) or general-purpose microprocessors.

[0221] Additionally, some implementations involve tangible and / or non-transitory computer-readable media or computer program products, including program instructions and / or data (including data structures) for performing various computer-implemented operations. Examples of computer-readable media include, but are not limited to, semiconductor memory devices, magnetic media such as disk drives and magnetic tapes, optical media such as CDs, magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as read-only memory devices (ROM) and random access memory (RAM). The computer-readable medium may be directly controlled by an end user, or the medium may be indirectly controlled by an end user. Examples of directly controlled media include media located at a user facility and / or media not shared with other entities. Examples of indirectly controlled media include media that can be indirectly accessed by a user via an external network and / or through services providing shared resources (such as the “cloud”). Particularly useful clouds are those configured and managed to store and analyze genetic data, such as BaseSpace. TM Services (Illumina, Inc., San Diego, CA) or cloud services described in U.S. Patent Application Publication Nos. 2013 / 0275486 A1 or 2014 / 0214579 A1 (each of which is incorporated herein by reference). Examples of program instructions include both machine code generated by a compiler and files containing high-level code that can be executed by a computer using an interpreter.

[0222] In some implementations, the data or information used in the disclosed methods and apparatus is provided in electronic format. Such data or information may include reads derived from nucleic acid samples, reference sequences, methylation status, patterns of methylation status, methylation differential scores, normalized methylation differential scores, total coverage-weighted normalized methylation differential scores, methylation scores, coverage-weighted methylation scores, consultation recommendations, diagnoses, etc. As used herein, data or other information provided in electronic format can be stored on a machine and transferred between machines. Typically, data in electronic format is provided digitally and can be stored in various data structures, lists, databases, etc., in bits and / or bytes. Data can be represented electronically, optically, etc.

[0223] Additionally, some implementations involve tangible and / or non-transitory computer-readable media or computer program products, which include instructions and / or data (including data structures) for performing various computer-implemented operations. One or more steps of the methods listed herein may be implemented by a computer program residing in a tangible and / or non-transitory computer-readable medium or using computer hardware.

[0224] For example, a computer program product is provided, and it includes a non-transitory computer-readable medium thereon providing program instructions for the following steps: (a) obtaining a test dataset comprising (i) methylation states at multiple sites of test genomic DNA from at least one test organism, and (ii) coverage at each site for detecting methylation states; (b) obtaining methylation states at multiple sites of reference genomic DNA from one or more reference individual organisms; (c) determining methylation differences between the test genomic DNA and the reference genomic DNA for each site, thereby providing a normalized methylation difference for each site; and (d) weighting the normalized methylation differences at each site by coverage at each site, thereby determining a total coverage-weighted normalized methylation difference score; and (e) storing or transmitting the total coverage-weighted normalized methylation difference score.

[0225] The methods disclosed herein can also be performed using a computer processing system adapted or configured to perform methods for identifying the methylation state or other characteristics of nucleic acids. Therefore, in one embodiment, the present invention provides a computer processing system adapted or configured to perform the methods described herein. In one embodiment, the device includes a nucleic acid detection apparatus, such as a nucleic acid sequencing apparatus, adapted or configured to determine the methylation state and / or other characteristics of nucleic acids. The device may also include components for processing samples from a test organism and / or a reference organism. Such components are described elsewhere herein.

[0226] Nucleic acid sequences, methylation states, methylation patterns, or other data may be input into a computer or stored directly or indirectly on a computer-readable medium. In one embodiment, the computer system is directly connected to a nucleic acid detection device (e.g., a sequencing device) that determines the methylation state of nucleic acids from a sample. Data or other information from such tools is provided through an interface in the computer system. Alternatively, methylation data processed by the system is provided from a data storage source such as a database or other repository. Once available for processing, a storage device or mass storage device at least temporarily acquires or stores the methylation state or other characteristics of the nucleic acids. Additionally, as described herein, the storage device may store methylation differences, normalized methylation differences, total weighted normalized methylation differences, methylation scores, or coverage-weighted methylation scores. The storage may also store various routines and / or programs for analyzing or presenting such information. Such programs / routines may include programs for performing statistical analyses, etc.

[0227] In one instance, a user provides a sample to a nucleic acid sequencing device. Data is collected and / or analyzed by the sequencing device connected to a computer. Software on the computer performs the data collection and / or analysis. The data may be stored, displayed (e.g., via a monitor or other similar device), and / or transmitted to another location. The computer may be connected to the Internet for transmitting data to a handheld device and / or cloud environment used by a remote user (e.g., a physician, scientist, or analyst). It should be understood that the data may be stored and / or analyzed prior to transmission. In some embodiments, raw data is collected and transmitted to a remote user or device for analysis and / or storage. Transmission may occur via the Internet, but may also be via satellite or other connections. Alternatively, the data may be stored on a computer-readable medium, and said medium may be transmitted to the end user (e.g., via mail). Remote users may be located in the same or different geographical locations, including but not limited to buildings, cities, states, countries, or continents.

[0228] In some embodiments, the method further includes acquiring data on multiple polynucleotide sequences (e.g., reads, tags, and / or methylation states) and transmitting the data to a computer or other computing system. For example, the computer may be connected to laboratory equipment such as sample acquisition devices, nucleotide amplification devices, nucleotide sequencing devices, or hybridization devices. The computer can then acquire the applicable data collected by the laboratory apparatus. Data may be stored on the computer at any step, e.g., during real-time acquisition, before transmission, during transmission, or with binding transmission, or after transmission. Data may be stored on a computer-readable medium retrievable from the computer. Acquired or stored data may be transferred from the computer to a remote location, for example, via a local network or a wide area network such as the Internet. As described below, various operations can be performed on the transmitted data at the remote location.

[0229] The types of electronically formatted data that can be stored, transmitted, analyzed, and / or manipulated in the systems, apparatus, and methods disclosed herein are as follows: reads obtained by sequencing nucleic acids in a test sample, methylation status of sites in nucleic acids, one or more reference genomes or sequences, methylation differential scores, normalized methylation differential scores, total coverage-weighted normalized methylation differential scores, methylation scores, or coverage-weighted methylation scores, as described herein.

[0230] These diverse data types can be acquired, stored, transmitted, analyzed, and / or manipulated at one or more locations using various devices. Processing options span a wide range. At one end of the range, all or many of this information are stored and used at the location where the test sample is processed (e.g., a physician's office or other clinical setting). At the other extreme, a sample is acquired at one location, processed at a second location (e.g., preparation, detection, or sequencing), the data is analyzed at a third location (e.g., aligning sequencing reads and determining methylation characteristics), and a diagnosis, recommendation, and / or planning is made at a fourth location (or the location where the sample was acquired).

[0231] In various implementations, methylation data is generated on a nucleic acid detection device (e.g., a sequencing device) and then transmitted to a remote site where it is processed to determine methylation characteristics. At this remote location, for example, a methylation differential score, a normalized methylation differential score, a total coverage-weighted normalized methylation differential score, a methylation score, or a coverage-weighted methylation score can be determined. Also at this remote location, methylation characteristics can be evaluated to make a prognostic or diagnostic determination.

[0232] Any one or more of these operations can be automated, as described elsewhere in this document. Typically, nucleic acid detection and sequence data analysis are performed computationally. Other operations may be performed manually or automatically.

[0233] Examples of locations where sample collection can be performed include healthcare worker offices, clinics, patients' homes (where sample collection tools or kits are provided), and mobile healthcare vehicles. Examples of locations where sample processing prior to methylation testing can be performed include healthcare worker offices, clinics, patients' homes (where sample processing equipment or kits are provided), mobile healthcare vehicles, and facilities of nucleic acid analysis providers. Examples of locations where nucleic acid testing (e.g., sequencing) can be performed include healthcare worker offices, clinics, healthcare worker offices, clinics, patients' homes (where sample sequencing equipment and / or kits are provided), mobile healthcare vehicles, and facilities of nucleic acid analysis providers. Locations where nucleic acid testing occurs may be equipped with dedicated network connections for transmitting sequence data in electronic format (typically reads). This connection may be wired or wireless and may be configured to send data to sites where the data can be processed and / or aggregated before being transmitted to processing sites. Data aggregators may be maintained by health organizations such as health maintenance organizations (HMOs).

[0234] The analytical operations can be performed at any of the aforementioned locations or alternatively at another site dedicated to computation and / or services for analyzing nucleic acid sequence data. Such locations include, for example, clusters such as general-purpose server clusters, facilities for genetic analysis services, etc. In some embodiments, the computing equipment used to perform the analysis is leased or rented. Computing resources can be part of an internet-accessible collection of processors, such as processing resources commonly known as the “cloud,” instances of which are provided elsewhere herein. In some cases, computation is performed by parallel or massively parallel processor groups that are related to or independent of each other. This processing can be implemented using distributed processing such as cluster computing, grid computing, etc. In such embodiments, clusters or grids of computing resources collectively form a supercomputer consisting of multiple processors or computers working together to perform the analyses and / or derivations described herein. These techniques, as well as more conventional supercomputers, can be used to process sequence data as described herein. Each takes the form of processor- or computer-dependent parallel computation. In the case of grid computing, these processors (often an entire computer) are connected via a network (private network, public network, or the internet) using conventional network protocols such as Ethernet. In contrast, supercomputers have many processors connected via a local high-speed computer bus.

[0235] In some implementations, a diagnosis (e.g., determining that a patient has a specific type of cancer) is generated at the same location as the analysis operation. In other implementations, it is performed at a different location. In some embodiments, the diagnosis is reported at the sampling location, although this is not always the case. Examples of locations where diagnoses can be generated or reported and / or where planning can be performed include a healthcare professional's office, clinic, computer-accessible internet site, and handheld devices with wired or wireless network connectivity, such as cellular phones, tablets, smartphones, etc. Examples of locations where consultations can be performed include a healthcare professional's office, clinic, computer-accessible internet site, handheld devices, etc.

[0236] In some implementations, sample collection, sample processing, and methylation status detection are performed at a first location, and analysis and derivation are performed at a second location. However, in some cases, sample collection is performed at one location (e.g., a healthcare professional's office or clinic), and sample processing and methylation status detection are performed at a different location, optionally the same location where analysis and derivation occur.

[0237] In various implementations, a series of operations listed above can be triggered by a user or entity initiating sample acquisition, sample processing, and / or methylation status detection. Following the initiation of one or more of these operations, other operations may naturally occur sequentially. For example, a nucleic acid sequencing operation may result in reads being automatically acquired and sent to a processing device, which then often automatically and possibly without additional user intervention performs methylation status analysis and determines methylation differential scores, normalized methylation differential scores, total coverage-weighted normalized methylation differential scores, methylation scores, or coverage-weighted methylation scores. In some specific implementations, the results of this processing operation may then be automatically delivered to processing information or reported to system components or entities, in a reformulated diagnostic format, to health professionals and / or patients. As explained, such information may also be automatically processed to generate treatment, testing, and / or monitoring plans, possibly along with consultation information. Thus, initiating early-stage operations can trigger end-to-end processing, where health professionals, patients, or other relevant parties are provided with diagnostic, planning, consultation, and / or other information that can be used to take action on the patient's condition. This can be achieved even if parts of the system are physically separated and may be located far from places such as sample collection and nucleic acid testing equipment.

[0238] In some implementations, the results of the methods listed herein will be communicated to the individual by a genetic counselor, physician (e.g., attending physician, obstetrician, etc.), or other qualified medical expert. In some implementations, counseling is provided in person; however, it should be recognized that in some cases, counseling may be provided remotely (e.g., via text, cellular phone, cellular phone application, tablet application, internet, etc.).

[0239] In some implementations, results may be delivered to medical experts or patients via computer systems. For example, a “smart recommendation” system may be provided that offers genetic counseling information in response to test results, instructions from a healthcare provider, and / or in response to a query (e.g., from a patient). In some implementations, the information will be specific to clinical information provided by physicians, healthcare systems, and / or patients. In some implementations, information may be provided iteratively. Thus, for example, a patient may provide a “hypothesis” query, and the system may return information such as diagnostic options, risk factors, timing, and the meaning of various outcomes.

[0240] In specific implementations, the results or other information generated by the methods listed herein may be provided in a temporary manner (e.g., presented on a computer screen). In some implementations, the information may be provided in a non-temporary manner. Thus, for example, the information may be printed out (e.g., as a list of options and / or suggestions optionally associated with time) and / or stored on a computer-readable medium (e.g., magnetic media such as local hard drives, servers, etc., optical media, flash memory, etc.).

[0241] It should be understood that such systems are typically configured to provide sufficient security to maintain patient privacy, for example, in accordance with general standards in the medical field.

[0242] The foregoing discussion of genetic counseling is intended to be illustrative and non-limiting. Genetic counseling is a well-established branch of medical science, and the methods described herein are incorporated into the counseling component within the scope and skill level of practitioners. Furthermore, it should be recognized that the nature of genetic counseling and related information and advice may change as the field progresses.

[0243] The present invention also relates to the following embodiments:

[0244] 1. A method for distinguishing aberrant methylation levels of DNA in a sample containing DNA from multiple different cell types, comprising:

[0245] (a) Provide a sample containing a mixture of genomic DNA from multiple different cell types of an individual test organism, thereby providing test genomic DNA;

[0246] (b) Detect the methylation status of multiple CpG sites in the test genomic DNA;

[0247] (c) Determine the coverage at each of the CpG sites used to detect the methylation state;

[0248] (d) Provide the methylation status of the plurality of CpG sites in reference genomic DNA from at least one reference individual;

[0249] (e) Determine the methylation difference between the test genomic DNA and the reference genomic DNA for each CpG site, thereby providing a normalized methylation difference for each CpG site; and

[0250] (f) The normalized methylation difference at each CpG site is weighted by the coverage at each CpG site to determine the total coverage-weighted normalized methylation difference score.

[0251] 2. The method of embodiment 1, wherein the sample comprises circulating tumor DNA and circulating non-tumor DNA.

[0252] 3. The method of embodiment 1, wherein the sample comprises cell-free DNA from blood.

[0253] 4. The method of any one of embodiments 1 to 3, wherein the individual test organism is a pregnant woman, and the test genomic DNA comprises genomic DNA derived from the woman's somatic cells and genomic DNA derived from the somatic cells of the woman's prenatal offspring.

[0254] 5. The method of any one of embodiments 1 to 4, wherein providing the sample in step (a) comprises targeting a subset of genomic DNA fragments containing a predetermined set of target CpG sites.

[0255] 6. The method of embodiment 5, wherein providing the sample in step (a) further includes treating the subset of genomic DNA fragments with bisulfite.

[0256] 7. The method of any one of embodiments 1 to 6, wherein the detection in step (b) comprises sequencing technology for continuously separating nucleotides in the test genomic DNA.

[0257] 8. The method as described in any one of embodiments 1 to 7, comprising:

[0258] (I) Steps (a) to (f) are repeated using a second test genomic DNA provided from a sample containing a mixture of genomic DNA from multiple different cell types of the individual test organism and using the same reference genomic DNA from the at least one reference individual.

[0259] (II) Determine whether the total coverage-weighted normalized methylation difference score between the test genomic DNA and the second test genomic DNA has changed.

[0260] 9. The method as described in any one of embodiments 1 to 8, wherein the normalized methylation difference at a specific CpG site is determined according to the following:

[0261]

[0262] Z i χ represents the normalized methylation difference at a specific CpG site identified as i. i The methylation level at CPG site i in the tested genomic DNA is represented by μ. i This represents the average methylation level at CpG site i in the reference genome, and σ i The standard deviation represents the methylation level at CpG site i in the reference genomic DNA.

[0263] 10. The method of embodiment 9, wherein the total coverage-weighted normalized methylation difference score (denoted as A) is determined according to the following:

[0264]

[0265] Where w i This represents the coverage at CpG site i, and k represents the total number of CpG sites.

[0266] 11. A method for distinguishing aberrant methylation levels of DNA from a first cell type, the method comprising:

[0267] (a) Provide the mean methylation level and standard deviation of the methylation level for each CpG site in the baseline genomic DNA of two or more normal individual baseline organisms;

[0268] (b) Provide a test dataset, which includes:

[0269] The methylation status of the plurality of CpG sites in the first test genomic DNA from an individual test organism, wherein the CpG sites are derived from the sample.

[0270] (c) Determine the methylation difference between the first test genomic DNA and the baseline genomic DNA for each CpG site, thereby providing a normalized methylation difference for each CpG site; and

[0271] (d) The normalized methylation difference at each CpG site is converted into a one-sided p value;

[0272] (e) Determine the total methylation fraction of the combination of one-sided p values ​​for each CpG site of the first test genomic DNA.

[0273] 12. The method of embodiment 11, wherein (a) includes providing the methylation status of the plurality of CpG sites in the baseline genomic DNA from the two or more normal individual organisms, and determining the average methylation level and the standard deviation of the methylation level of the baseline genomic DNA for each CpG site.

[0274] 13. The method as described in embodiment 11, further comprising:

[0275] A second test dataset is provided, which includes:

[0276] The methylation status of the plurality of CpG sites in the second test genomic DNA from the individual test organism, wherein the CpG sites are derived from the sample;

[0277] For each CpG site, the methylation difference between the second test genomic DNA and the baseline genomic DNA is determined, thereby providing the normalized methylation difference for each CpG site of the second test genomic DNA; and

[0278] The normalized methylation differences at each CpG site of the second test genomic DNA are converted into one-sided p-values;

[0279] The total methylation fraction of the combination of one-sided p-values ​​for each CpG site in the second test genomic DNA was determined; and

[0280] The total methylation fractions of the first test genomic DNA and the second test genomic DNA are compared to determine whether the total methylation fraction between the first and second test genomic DNAs has changed.

[0281] 14. The method of embodiment 11, further comprising:

[0282] (f) Provide a training dataset, which includes:

[0283] The methylation status of the plurality of CpG sites in the training genomic DNA of two or more normal individual training organisms, wherein the CpG sites are derived from a variety of different cell types from the normal individual training organisms;

[0284] (g) For each CpG site, determine the methylation difference between each training genomic DNA from the normal individual training organism and the baseline genomic DNA, thereby providing the normalized methylation difference for each CpG site of each training genomic DNA;

[0285] (h) Convert the normalized methylation differences at each CpG site into one-sided p-values; and

[0286] (i) Determine the total methylation fraction of the combination of one-sided p values ​​for each CpG site in each training genomic DNA;

[0287] (j) Using the total methylation fraction of each training genomic DNA, calculate the mean total methylation fraction and the standard deviation of the total methylation fraction of the training genomic DNA to obtain the distribution of the total methylation fraction of the training genomic DNA; and

[0288] (k) Evaluate the total methylation score of the first test genomic DNA for the distribution of the total methylation score of the training genomic DNA.

[0289] 15. The method as described in any one of embodiments 11 to 14, wherein the normalized methylation difference at a specific CpG site is determined according to the following formula:

[0290]

[0291] Z i χ represents the normalized methylation difference at a specific CpG site identified as i. i The methylation level at CPG site i in the first test genomic DNA or the training genomic DNA is represented by μ. i This represents the average methylation level at CpG site i in the baseline genome, and σ i The standard deviation represents the methylation level at CpG site i in the baseline genomic DNA.

[0292] 16. The method as described in any one of embodiments 14 to 15, wherein the evaluation in step (k) determines the difference in normalized methylation fraction according to the following formula:

[0293]

[0294] ZMS The difference in normalized methylation fraction is represented by MS, which represents the total methylation fraction of the first tested genomic DNA, in μ. MS σ represents the average methylation fraction of the training set of normal genomic DNA. MS The standard deviation represents the total methylation fraction of the training set of normal genomic DNA.

[0295] 17. The method as described in any one of embodiments 11 to 16, wherein the total methylation fraction (MS) is determined according to the following formula:

[0296]

[0297] Where p i denoted by , where p represents the one-sided p value at site i, and k represents the total number of CpG sites.

[0298] 18. The method as described in any one of embodiments 11 to 17, wherein the total methylation fraction (MS) is determined according to the following formula:

[0299]

[0300] Where p i Let represent the one-sided p-value at site i, k represent the total number of CpG sites, and w i This indicates the coverage of site i.

[0301] 19. The method of any one of embodiments 11 to 18, wherein the sample from the individual test organism comprises circulating tumor DNA and circulating non-tumor DNA.

[0302] 20. The method of any one of embodiments 11 to 19, wherein the sample comprises cell-free DNA from blood.

[0303] 21. The method of any one of embodiments 11 to 20, wherein the individual test organism is a pregnant woman, and the first test genomic DNA comprises genomic DNA derived from the woman's somatic cells and genomic DNA derived from the somatic cells of the woman's prenatal offspring.

[0304] 22. The method of any one of embodiments 11 to 21, wherein providing the sample in step (b) comprises targeting a subset of genomic DNA fragments containing a predetermined set of target CpG sites.

[0305] 23. The method of any one of embodiments 11 to 22, wherein providing the sample in step (b) further comprises treating the subset of genomic DNA fragments with bisulfite.

[0306] 24. The method of any one of embodiments 11 to 23, wherein providing the sample in step (b) includes detecting the methylation status of the CpG site.

[0307] 25. The method of any one of embodiments 11 to 24, wherein the detection of the methylation status of the CpG site comprises sequencing technology that sequentially identifies nucleotides in the first test genomic DNA.

[0308] 26. The method as described in implementation scheme 16, wherein if Z Ms If the value is greater than 3, the first tested genomic DNA is classified as having an abnormal methylation level.

[0309] 27. A method for identifying cancer present in an individual organism, comprising:

[0310] (a) Provide a dataset, which includes:

[0311] Methylation status of multiple CpG sites i in genomic DNA from clinical samples known to include specific cancers;

[0312] (b) Identifying cancer type-specific hypermethylated CpG sites i, including:

[0313] (i) Determine the average methylation level of each CpG site i in the genomic DNA of the clinical sample known to include the specific cancer.

[0314] (ii) Determine which CpG sites i satisfy the first threshold, the second threshold, or a combination thereof.

[0315] Determining the first threshold includes (1) determining the absolute value of the average methylation level for each CpG site i; (2) sorting the average methylation levels of each CpG site i from low to high; and (3) selecting those CpG sites i that have an average methylation level greater than or equal to a first preselected value in the percentage order.

[0316] Determining the second threshold includes (1) determining the absolute value of the average methylation level for each CpG site i; and (2) selecting those CpG sites i that have an average methylation level greater than the second preselected value.

[0317] (iii) Compile a list of hypermethylation sites specific to the stated cancer type.

[0318] (c) Repeat (a) and (b) for each specific cancer to generate multiple lists of additional cancer type-specific hypermethylation sites;

[0319] (d) Provide a test dataset, which includes:

[0320] The methylation level of each hypermethylation site from the test genomic DNA of the individual test organism, wherein the hypermethylation site is one of the list of cancer type-specific hypermethylation sites identified in steps (b) and (c);

[0321] (e) The methylation level of each of the hypermethylation sites is averaged to produce a single average methylation level of the test genomic DNA for the cancer type identified in steps (b) and (c);

[0322] (f) Repeat step (e) for each cancer type to generate the average methylation level for each cancer type;

[0323] (g) The average methylation levels of each cancer type are sorted from low to high, wherein the cancer type corresponding to the highest average methylation level is the cancer present in the individual test organism.

[0324] 28. The method of embodiment 27, wherein the first preselected value is a percentile rank greater than or equal to 97.

[0325] 29. The method as described in any one of embodiments 27 to 28, wherein the second preselected value is greater than 6%.

[0326] 30. The method of any one of embodiments 27 to 29, wherein the test genomic DNA from the individual test organism comprises circulating tumor DNA and circulating non-tumor DNA.

[0327] 31. The method of any one of embodiments 27 to 30, wherein the tested genomic DNA comprises cell-free DNA from blood.

[0328] 32. The method of any one of embodiments 27 to 31, wherein providing the test dataset in step (d) comprises targeting a subset of genomic DNA fragments containing a pre-determined set of target CpG sites.

[0329] 33. The method of any one of embodiments 27 to 32, wherein providing the test dataset in step (d) further includes the subset of genomic DNA fragments treated with bisulfite.

[0330] 34. The method of any one of embodiments 27 to 33, wherein providing the test dataset in step (d) includes detecting the methylation state of the CpG sites.

[0331] 35. The method of any one of embodiments 27 to 34, wherein the detection of the methylation status of the CpG site comprises sequencing technology that sequentially identifies nucleotides in the first test genomic DNA.

[0332] Example I

[0333] Analytical sensitivity of ctDNA methylation-based cancer detection using aggregated normalized coverage weighted methylation differentials

[0334] This embodiment describes a highly sensitive assay for detecting methylation in circulating tumor DNA (ctDNA). Aberrant DNA methylation is a common phenomenon in cancer and can be one of the earliest changes to occur during tumorigenesis. The assay described in this embodiment can be used for cancer screening.

[0335] The general approach used in this application includes targeted methylation sequencing of multiple CpG sites affected in cancer.

[0336] The technical challenges addressed by the method include providing ultra-high sensitivity and specificity beneficial for screening applications, providing protocols for targeted methylation sequencing from low-input ctDNA, and providing bioinformatics algorithms for analyzing methylation levels across a large number of target sites.

[0337] Targeted capture probe design

[0338] Two targeted methylation groups were developed. The pan-cancer group targets 9,921 CpG sites affected in 20 major cancer types, selected from the Cancer Genome Atlas database. The CpG sites included in the pan-cancer group are listed in Table I. The CRC group targets 1,162 CpG sites affected in colorectal cancer. The CpG sites included in the CRC group are listed in Table II. The CpG sites listed in Tables I and II refer to genome construct 37.

[0339] The probe sequences for CpG sites were selected from an Infinium HM450 array (Illumina, San Diego, CA). The probe design principle is based on... Figure 1 As shown in the diagram. Two probes are used for targets with more than 4 CpG sites, including a fully methylated probe (with a G nucleotide complementary to the C position of each CpG site) and a fully unmethylated probe (with an A nucleotide complementary to the U nucleotide expected to be generated by the bisulfite conversion of each C position of the CpG site), as shown in the diagram. Figure 1 As shown. In contrast, only one probe is used for targets with 4 or fewer CpG sites (the probe includes a degenerate nucleotide R complementary to U or C at the C position of each CpG site).

[0340] cfDNA isolated and extracted from plasma

[0341] Plasma samples were obtained from human blood. Cell-free DNA (cfDNA) was extracted using the QIAamp Cyclic Nucleic Acid Kit (Qiagen, Hilden, Germany). Figure 2 As shown, and in an environment for evaluating titration and detection sensitivity, the workflow below is used to perform targeted ctDNA methylation sequencing.

[0342] Titration and detection sensitivity

[0343] Genomic DNA for NA12878 was purchased from the Coriell Institute (Coriell Institute, Camden, NJ), and genomic DNA for LS1034 was purchased from ATCC (ATCC, Manassas, VA). Genomic DNA fragmentation was performed using a Covaris M200 (Covaris, Woburn, MA), and the fragment size was selected to 130-250 bp using BluePippin (Sage Science, Beverly, MA) to mimic the size distribution of cfDNA. Quant-iT was used... TM DNA quantification was performed using a dsDNA assay kit (ThermoFisher Scientific, Grand Island, NY). 10%, 1%, or 0.1% LS1034 DNA was added to the NA12878 DNA background to prepare DNA mixtures. 30 ng of each mixture, 100% NA12878, or 100% LS1034 DNA was used in library preparation. Three replicate libraries were generated for each titration level. A set of six replicates from NA12878 was used as a baseline reference genome.

[0344] Extracted cfDNA or cleaved and size-selected genomic DNA was treated with bisulfite and purified using the EZ DNA Methylation-Lightning kit (Zymo Research, Irvine, CA).

[0345] use Bisulfite sequencing libraries were prepared using the Methylation Sequencing DNA Library Kit (Swift Biosciences, Ann Arbor, MI).

[0346] Targeted capture of bisulfite sequencing libraries was performed using probes complementary to fragments having CpG sites listed in Table I or Table II. The capture probes were synthesized and biotinylated at Illumina. Illumina TruSight was used according to the manufacturer's instructions. TM The rapid capture kit performs target capture, but with the difference of using a custom capture probe and performing hybridization and washing steps at 48°C.

[0347] According to the manufacturer's instructions, the products of the capture step were sequenced on an Illumina HiSeq 2500 sequencer using 2 x 100 cycles, with four samples in fast run mode.

[0348] Bioinformatics Analysis

[0349] The FASTQ sequences were demultiplexed and then computer-demethylated, whereby all Cs in read 1 were converted to Ts and all Gs in read 2 were converted to As. These “demethylated” FASTQ sequences were then aligned with an index containing the “demethylated” hg19 genome using BWA (v 0.7.10-r789). BWA alignment is described in Li and Durbin (2010), Fast and accurate long-read alignment with Burrows-Wheeler Transform. Bioinformatics, Epub. [PMID: 20080505], which is incorporated herein by reference. After alignment, the “demethylated” FASTQ sequences were replaced with the original FASTQ sequences. Methylation levels were calculated as the fraction of 'C' bases at target CpG sites in the total 'C'+'T' bases.

[0350] After calculating the methylation level at each CpG site for each sample and replicate, the total coverage-weighted normalized methylation difference z-score was calculated as follows.

[0351] (1) The methylation level at each CpG site was normalized by subtracting the mean methylation level from the baseline and dividing by the standard deviation of the methylation level at the baseline to obtain the z-score for each site. Specifically, the normalized methylation difference at each CpG site was determined according to the following formula:

[0352]

[0353] Z i χ represents the normalized methylation difference at a specific site identified as i. i μ represents the methylation level at site i in the tested genomic DNA.i σ represents the average methylation level at site i in the reference genome, and σ i This represents the standard deviation of the methylation level at site i in the reference genomic DNA.

[0354] (2) Multiply the z-score at each CpG site by the coverage observed at the CpG site, and then sum the coverage-weighted z-scores across all CpG sites and then divide by the sum of the squares of coverage at each CpG site. More specifically, determine the total coverage-weighted methylation difference z-score according to the following formula (example A of total coverage-weighted normalized methylation difference score):

[0355]

[0356] Where w i represents the coverage at site i, and k represents the total number of sites.

[0357] result

[0358] Titration experiments were performed using the colorectal cancer cell line LS1034 and the normal cell line NA12878 to demonstrate analytical sensitivity. Specifically, LS1034 was titrated into NA12878 at 0.1%, 1%, and 10% using both pan-cancer and CRC groups, and targeted ctDNA methylation sequencing was performed on triplicate copies of both pure LS1034 and pure NA12878. For each of the 15 sample replicates, the total coverage-weighted methylation differential z-score was calculated using the normal NA12878 sample as a baseline. Figure 3 The results indicate that the variation within the sample is much smaller than the variation between titration levels. Specifically, the significant difference between the LS1034 titrated at 0.1% in NA12878 and the NA12878 sample indicates that the determination and the associated total coverage-weighted methylation difference z-score achieve a detection limit of 0.1%.

[0359] The results obtained using the method of this embodiment provide a highly sensitive assessment of the cumulative effects of multiple affected CpG sites across the genome. By providing a method for detecting methylation patterns, the method of this embodiment can provide improved cancer diagnosis compared to methods that rely on somatic mutation detection, as demonstrated by the improved consistency between alternations between CRC tissue and corresponding plasma when assessing DNA methylation markers compared to somatic mutations (see, for example, Danese et al., “Comparison of Genetic and Epigenetic Alterations of Primary Tumors and Matched Plasma Samples in Patients with Colorectal Cancer” PLoS ONE 10(5): e0126417. doi: 10.1371 / journal.pone.0126417 (2015), which is incorporated herein by reference). The method described in this embodiment also provides identification of the tissue origin of cancer. Specifically, tissue-specific methylation markers have been shown to be useful for tracing the tissue origin of specific ctDNA sequences (see, for example, Sun et al., “Plasma DNAtissue mapping by genome-wide methylation sequencing for noninvasive prenatal, cancer, and transplantation assessments” Proc. Natl. Acad. Sci., USA 112(40)E5503-E5512(2015), which is incorporated herein by reference).

[0360] Example II

[0361] Analytical Sensitivity of ctDNA Methylation-Based Cancer Detection Using Coverage-Weighted Methylation Fractions

[0362] This embodiment describes an alternative, highly sensitive assay for detecting methylation in circulating tumor DNA (ctDNA). The assay described in this embodiment can also be used for cancer screening, monitoring disease progression, or assessing a patient's response to therapeutic treatment.

[0363] Targeted capture probe design

[0364] For this study, the two targeted methylation groups described in Example I will be combined. The pan-cancer group targets 9,921 CpG sites affected in 20 major cancer types, as selected from the Cancer Genome Atlas database. The CpG sites included in the pan-cancer group are listed in Table I. The CRC group targets 1,162 CpG sites affected in colorectal cancer. The CpG sites included in the CRC group are listed in Table II. The combined CpG sites listed in Tables I and II refer to Genome Construction 37.

[0365] The probe sequences for CpG sites were selected from an Infinium HM450 array (Illumina, San Diego, CA). The probe design principle is based on... Figure 1 As shown in the diagram. Two probes are used for targets with more than 4 CpG sites, including a fully methylated probe (with a G nucleotide complementary to the C position of each CpG site) and a fully unmethylated probe (with an A nucleotide complementary to the U nucleotide expected to be generated by the bisulfite conversion of each C position of the CpG site), as shown in the diagram. Figure 1 As shown. In contrast, only one probe is used for targets with 4 or fewer CpG sites (the probe includes a degenerate nucleotide R complementary to U or C at the C position of each CpG site).

[0366] cfDNA isolated and extracted from plasma

[0367] Plasma samples were obtained from human blood. Cell-free DNA (ctDNA) was extracted using the QIAamp Cyclic Nucleic Acid Kit (Qiagen, Hilden, Germany). Figure 2 As shown, and in an environment for evaluating titration and detection sensitivity, the workflow below is used to perform targeted ctDNA methylation sequencing.

[0368] Titration and detection sensitivity

[0369] As mentioned above, NA12878 genomic DNA was purchased from the Coriell Institute (Coriell Institute, Camden, NJ), and LS1034 genomic DNA was purchased from ATCC (ATCC, Manassas, VA). Genomic DNA fragmentation was performed using a Covaris M200 (Covaris, Woburn, MA), and its size was selected to 130-250 bp using a BlackPippin (Sage Science, Beverly, MA) to mimic the size distribution of cfDNA. Quant-iT was used... TM DNA quantification was performed using a dsDNA assay kit (ThermoFisher Scientific, Grand Island, NY). 10%, 1%, or 0.1% LS1034 DNA was added to the NA12878 DNA background to prepare DNA mixtures. 30 ng of each mixture, 100% NA12878, or 100% LS1034 DNA was used in library preparation. Three replicate libraries were generated for each titration level. A set of six replicates of NA12878 was used as a baseline reference genome.

[0370] Extracted cfDNA or cleaved and size-selected genomic DNA was treated with bisulfite and purified using the EZ DNA Methylation-Lightning kit (Zymo Research, Irvine, CA).

[0371] use Bisulfite sequencing libraries were prepared using the Methylation Sequencing DNA Library Kit (Swift Biosciences, Ann Arbor, MI).

[0372] Targeted capture of bisulfite sequencing libraries was performed using probes complementary to fragments having CpG sites listed in Table I or Table II. The capture probes were synthesized and biotinylated at Illumina. Tllumina TruSight was used according to the manufacturer's instructions. TM The rapid capture kit performs target capture, but with the difference of using a custom capture probe and performing hybridization and washing steps at 48°C.

[0373] According to the manufacturer's instructions, the products of the capture step were sequenced on an Illumina HiSeq 2500 sequencer using 2 x 100 cycles, with four samples in fast run mode.

[0374] Bioinformatics Analysis

[0375] The FASTQ sequences were demultiplexed and then computer-demethylated, whereby all Cs in read 1 were converted to Ts and all Gs in read 2 were converted to As. These “demethylated” FASTQ sequences were then aligned with an index containing the “demethylated” hg19 genome using BWA (v 0.7.10-r789). BWA alignment is described in Li and Durbin (2010), Fast and accurate long-read alignment with Burrows-Wheeler Transform. Bioinformatics, Epub. [PMID: 20080505], which is incorporated herein by reference. After alignment, the “demethylated” FASTQ sequences were replaced with the original FASTQ sequences. Methylation levels were calculated as the fraction of 'C' bases at target CpG sites in the total 'C'+'T' bases.

[0376] After calculating the methylation level at each CpG site for each sample and replicate, the coverage-weighted methylation fraction was calculated as follows.

[0377] (1) The methylation level at each CpG site was normalized by subtracting the mean methylation level from the baseline and dividing by the standard deviation of the methylation level at the baseline to obtain the z-score for each site. Specifically, the normalized methylation difference at each CpG site was determined according to the following formula:

[0378]

[0379] Z i χ represents the normalized methylation difference at a specific site identified as i. i μ represents the methylation level at site i in the tested genomic DNA. i σ represents the average methylation level at site i in the reference genome, and σ i This represents the standard deviation of the methylation level at site i in the reference genomic DNA.

[0380] (2) By converting z-scores into one-sided p-values ​​(p i ) to calculate the z-score (Z) of each CpG site i i This is converted into the probability of observing this or a larger z-score. The probability of a normal distribution is calculated, although other distributions (e.g., the t-distribution or the binomial distribution) can also be used.

[0381] (3) By using the p value (p) at each CpG site i i Multiply by the coverage observed at the CpG site (w) iThe p-values ​​at each CpG site are weighted, and the coverage-weighted methylation fraction (MS) is determined by combining the weighted p-values ​​according to the following formula:

[0382]

[0383] Where p i Let represent the one-sided p-value at site i, k represent the total number of sites, and w i This indicates the coverage at site i.

[0384] result

[0385] Titration experiments were performed using the colorectal cancer cell line LS1034 and the normal cell line NA12878 to demonstrate analytical sensitivity. Specifically, LS1034 was titrated into NA12878 at 0.1%, 1%, and 10% using a combined pancancer and CRC group, and targeted ctDNA methylation sequencing was performed on triplicate copies of both pure LS1034 and pure NA12878. For each of the 15 sample replicates, a coverage-weighted methylation fraction was calculated using a normal NA12878 sample as a baseline. Figure 4 The results indicate that the variation within the sample is much smaller than the variation between titration levels. Specifically, the significant difference between the LS1034 titrated at 0.1% in NA12878 and the NA12878 sample indicates that the determination and the accompanying coverage-weighted methylation fraction can achieve a detection limit of 0.1% (see [reference]). Figure 4 (Illustration).

[0386] Similar to the results in Example I, the results obtained using the method of this example provide a highly sensitive assessment of the cumulative impact of multiple affected CpG sites across the genome. By providing an alternative method for detecting methylation patterns, the method of this example can provide a more sensitive cancer diagnosis than methods that rely on somatic mutation detection.

[0387] Example III

[0388] Clinical performance of ctDNA methylation-based cancer detection using normalized coverage-weighted methylation fraction differences

[0389] This embodiment uses normalized coverage-weighted methylation fraction differences to assess the clinical sensitivity and specificity of methylation-based cancer detection in circulating tumor DNA (ctDNA). As noted above, the assays described in this embodiment can be used for cancer screening, monitoring disease progression, or assessing patient response to therapeutic treatments.

[0390] Targeted capture probe design

[0391] For this study, the two targeted methylation groups described in Example I will be combined. The pan-cancer group targets 9,921 CpG sites affected in 20 major cancer types, as selected from the Cancer Genome Atlas database. The CpG sites included in the pan-cancer group are listed in Table I. The CRC group targets 1,162 CpG sites affected in colorectal cancer. The CpG sites included in the CRC group are listed in Table II. The combined CpG sites listed in Tables I and II refer to Genome Construction 37.

[0392] The probe sequences for CpG sites were selected from an Infinium HM450 array (Illumina, San Diego, CA). The probe design principle is based on... Figure 1 As shown in the diagram. Two probes are used for targets with more than 4 CpG sites, including a fully methylated probe (with a G nucleotide complementary to the C position of each CpG site) and a fully unmethylated probe (with an A nucleotide complementary to the U nucleotide expected to be generated by the bisulfite conversion of each C position of the CpG site), as shown in the diagram. Figure 1 As shown. In contrast, only one probe is used for targets with 4 or fewer CpG sites (the probe includes a degenerate nucleotide R complementary to U or C at the C position of each CpG site).

[0393] Blood sample collection and processing

[0394] Cancer patients were recruited at MD Anderson Cancer Center (Houston, Texas). A total of 70 blood samples were collected from 63 patients with advanced-stage cancer across three cancer types (n=30 for colorectal cancer (CRC), n=14 for breast cancer (BRCA), and n=19 for lung cancer) in this study. Four CRC patients had blood samples collected at multiple time points. Three breast cancer samples and one colorectal cancer sample did not meet sample quality control and were therefore excluded from the analysis, resulting in a final set of 66 cancer samples (36 CRC, 11 BRCA, and 19 lung), representing 59 distinct patients (29 CRC, 11 BRCA, and 19 lung). A total of 65 normal blood samples were collected from healthy subjects to serve as baseline methylation controls (20), training controls (20), and test controls (25), as described herein.

[0395] Plasma was separated by centrifugation at 1600G for 10 minutes. The supernatant was transferred to 15 mL centrifuge tubes and centrifuged at 3000G for 10 minutes at room temperature. The supernatant was then transferred to fresh 15 mL centrifuge tubes and stored in a freezer (-80°C) and transported on dry ice. Plasma samples from healthy donors were obtained from Bioreclamation IVT (Westbury, New York). All samples were deidentified.

[0396] cfDNA isolated and extracted from plasma

[0397] Cell-free DNA (cfDNA) was extracted using the QIAamp Cyclic Nucleic Acid Kit (Qiagen, Hilden, Germany). Figure 2 As shown, and in an environment for evaluating titration and detection sensitivity, the workflow below is used to perform targeted ctDNA methylation sequencing.

[0398] Preparation and sequencing of targeted bisulfite sequencing libraries

[0399] The cfDNA was treated with bisulfite and purified using the EZ DNA Methylation-Lightning kit (Zymo Research, Irvine, California).

[0400] use The Methylation Sequencing DNA Library Kit (Swift Biosciences, Ann Arbor, Michigan) performs whole-genome amplification of bisulfite-converted DNA.

[0401] Targeted capture of bisulfite sequencing libraries was performed using probes complementary to fragments having CpG sites listed in Table I or Table II. Capture probes were synthesized and biotinylated at Illumina (San Diego, Califomia). Illumina TruSight was used according to the manufacturer's instructions. TM The rapid capture kit performs target capture. Hybridization and washing conditions are modified to produce optimal capture efficiency.

[0402] According to the manufacturer's instructions, the products of the capture step were sequenced on an Illumina Hiseq 2500 sequencer using 2 x 100 cycles, with four samples in fast run mode.

[0403] Bioinformatics Analysis

[0404] The FASTQ sequences were demultiplexed and then computer-demethylated, whereby all Cs in read 1 were converted to Ts and all Gs in read 2 were converted to As. These “demethylated” FASTQ sequences were then aligned with an index containing the “demethylated” hg19 genome using BWA (v 0.7.10-r789). BWA alignment is described in Li and Durbin (2010), Fast and accurate long-read alignment with Burrows-Wheeler Transform. Bioinformatics, Epub. [PMiD: 20080505], which is incorporated herein by reference. After alignment, the “demethylated” FASTQ sequences were replaced with the original FASTQ sequences. Methylation levels were calculated as the fraction of 'C' bases at target CpG sites in the total 'C'+'T' bases.

[0405] After calculating the methylation level at each CpG site for each sample and replicate, the coverage-weighted methylation fraction was calculated as follows.

[0406] (1) Initially determine methylation fractions for a training set of 20 normal genomic DNA samples. First, determine the normalized methylation difference (z-score) at a specific site i (e.g., CpG site) according to the following formula:

[0407]

[0408] Where Zi represents the normalized methylation difference at a specific site identified as i, and χ i μ represents the methylation level at site i in the training set of members of the normal genomic DNA. i σ represents the average methylation level at site i in the baseline sample, and σ i This represents the standard deviation of the methylation level at site i in the baseline sample.

[0409] (2) By converting z-scores into one-sided p-values ​​(p i ) to calculate the z-score (Z) of each CpG site i i This is converted into the probability of observing this or a larger z-score. The probability of a normal distribution is calculated, although other distributions (e.g., the t-distribution or the binomial distribution) can also be used.

[0410] (3) By using the p value (p) at each CpG site i i Multiply by the coverage observed at the CpG site (w) iThe p-values ​​at each CpG site are weighted, and the coverage-weighted methylation fraction (MS) is determined by combining the weighted p-values ​​according to the following formula:

[0411]

[0412] Where p i Let represent the one-sided p-value at site i, k represent the total number of sites, and w i Indicates the significance of site i, such as coverage.

[0413] (4) Then perform statistical analysis of the methylation fraction of the training set. Calculate the average methylation fraction (μ) of the training set of normal genomic DNA. MS ) and the standard deviation of methylation fraction (σ) MS This allows us to characterize the distribution of methylation fractions in the normal population.

[0414] (5) Next, methylation scores were determined for 66 cancer genomic DNA samples and 25 test controls. First, the normalized methylation difference (z-score) at each CpG site was determined using the following formula:

[0415]

[0416] Z i χ represents the normalized methylation difference at a specific site identified as i. i μ represents the methylation level at site i in the tested genomic DNA. i σ represents the average methylation level at site i in the reference genome, and σ i This represents the standard deviation of the methylation level at site i in the reference genomic DNA.

[0417] (6) By converting z-scores to one-sided p-values ​​(p i ) to calculate the z-score (Z) of each CpG site i i This is converted into the probability of observing this or a larger z-score. The probability of a normal distribution is calculated, although other distributions (e.g., the t-distribution or the binomial distribution) can also be used.

[0418] (7) By using the p value (p) at each CpG site i i Multiply by the coverage observed at the CpG site (w) i The p-values ​​at each CpG site are weighted, and the coverage-weighted methylation fraction (MS) is determined by combining the weighted p-values ​​according to the following formula:

[0419]

[0420] Where pi Let represent the one-sided p-value at site i, k represent the total number of sites, and w i This indicates the coverage at site i.

[0421] (8) Finally, the average methylation fraction (μ) of the training set composed of normal genomic DNA was calculated. MS ) and the standard deviation of methylation fraction (σ) MS The distribution of methylation fractions determined for the training set is represented by ), which evaluates the methylation fractions of the test genomic DNA sample. The methylation fraction of the test genomic DNA is determined by the following formula, comparing it to the average methylation fraction of the training set of normal genomic DNA (μ). MS The number of standard deviations between )

[0422]

[0423] Z MS The difference in normalized methylation fraction is represented by MS, which represents the methylation fraction of the test sample, in μ. MS σ represents the average methylation fraction of the training set of normal genomic DNA, and σ MS ZM represents the standard deviation of the methylation fraction in the training set of normal genomic DNA. A ZM value greater than 3 standard deviations indicates a methylation fraction. S The value is used as a threshold to identify cancer samples.

[0424] result

[0425] As noted above, the purpose of this experiment was to evaluate the clinical performance of a normalized coverage-weighted methylation score differential algorithm, including its clinical sensitivity and specificity. Sixty-six cancer samples and 25 normal samples underwent methylation score analysis as described herein, including determining the z-score for each CpG site listed in Tables I and II, converting the z-scores to one-sided p-values ​​based on a normal distribution, weighting the p-values ​​by coverage, and summing the individually weighted p-values ​​using the Fisher formula to obtain a single methylation score. The resulting methylation scores were used to distinguish cancer samples from normal samples. Figure 5 and Figure 6 The normalized coverage-weighted methylation fraction difference algorithm demonstrated its ability to detect 34 out of 36 CRC samples (94.4% sensitivity), 8 out of 11 BRCA samples (72.7% sensitivity), and 10 out of 19 lung cancer samples (52.6% sensitivity). The algorithm exhibited 100% specificity, correctly identifying all 25 test control samples as normal.

[0426] The results obtained using the method of this embodiment provide a highly sensitive and specific assessment of the cumulative effects of multiple affected CpG sites across the genome. By providing an alternative method for detecting methylation patterns, the method of this embodiment can provide a more sensitive and specific cancer diagnosis than methods that rely on somatic mutation detection.

[0427] Example IV

[0428] Clinical performance of a cancer type classification method based on the average methylation level of a preselected subset of transmethylation sites

[0429] This embodiment evaluates the clinical sensitivity of a method for cancer type classification based on the average methylation level across a preselected subset of CpG methylation sites, referred to herein as “hypermethylation sites.” The assay described in this embodiment can be used to identify the origin of tumors in circulating cell-free DNA.

[0430] Correlation of methylation profiles between plasma and tissue DNA samples

[0431] As an initial query, the correlation between the methylation profiles of circulating tumor DNA (ctDNA) isolated from plasma samples and those methylation profiles of DNA isolated from tumor tissue was determined. A high correlation would increase the credibility of the view that the methylation profiles of cfDNA can be used to classify tumors of origin. To this end, the methylation profiles of colorectal cancer, breast cancer, and lung cancer samples examined in Example III were compared with the average methylation profiles of each of the 32 cancer types from TCGA (Cancer Genome Atlas) with a minimum of 30 cancer samples in the database. Methylation profiles were determined substantially as described in Examples I-III and consisted of methylation levels at 9,242 CpG sites (with poor-performing methylation sites from the original CpG group filtered out to improve accuracy).

[0432] Comparisons were performed in pairs between each cancer-positive plasma sample from Example III and each of the 32 cancer types from TCGA, resulting in correlation coefficients ranging from 0 to 1. The correlations were plotted as a two-dimensional correlation graph. Figure 7 The graph is shown in the figure. Darker areas correspond to higher correlations, while brighter areas represent lower correlations. The observed correlations between methylation profiles are generally highest for the matched tumor type. For example, in breast cancer samples from plasma, the correlation is highest for breast cancer tissue (invasive breast cancer) and lower for all other tumor tissue types. Similarly, for CRC plasma samples, the correlation is highest for colon and rectal tissue (e.g., colon adenocarcinoma, esophageal cancer, colon adenocarcinoma, and gastric adenocarcinoma). The correlations are less pronounced in lung cancer samples.

[0433] Development and testing of cancer type classification

[0434] A significant correlation has been identified between ctDNA and the methylation profiles of DNA from tumor tissue. We will continue to develop and test computer-based cancer type classification methods.

[0435] First, 24 cancer types with more than 100 samples were identified in the TCGA database. For each of these types, a list of “hypermethylated” sites was created, defined as sites with an average methylation level (across samples) that is in the top 3% across the entire group and greater than 6% in absolute terms.

[0436] Given a test sample, its cancer type is determined in a three-step process. First, for each of the 24 cancer types, the methylation level of each “hypermethylation” site on the list is determined as described in Examples I-III. Next, for each of the 24 cancer types, the average methylation level across the “hypermethylation” sites is calculated. Finally, each of the 24 cancer types is ranked by its average methylation level across the “hypermethylation” sites, and the test sample is classified by the cancer type with the highest average methylation level.

[0437] The method was then retested for each of the TCGA tissue samples used to generate the list of "hypermethylated" sites. The accuracy of the method was defined as the ratio of the number of correctly identified samples of a specific type of cancer to the total number of samples of that cancer type. The results of this analysis were... Figure 8 As shown in the attached figure, 22 out of 24 cancer types were identified with an accuracy of over 75%. In fact, many of the cancer types were correctly identified in approximately 90% or higher cases. Only two types—esophageal cancer and testicular germ cell tumors—failed to exceed the 75% threshold.

[0438] Cancer type classification of plasma samples

[0439] Fifty-two plasma samples (34 CRC, 8 BRCA, and 10 lung) that were correctly identified as cancer samples in Example III underwent cancer type classification analysis as described above. The results of this analysis were... Figure 9 As shown in the figure, the cancer classification algorithm correctly identified 28 out of 34 CRC samples (82%), 7 out of 8 BRCA samples (88%), and 7 out of 10 lung cancer samples (70%). These results demonstrate that the cancer type classification method described in this paper can be used with high clinical sensitivity to identify the tissue of origin in ctDNA from plasma samples.

[0440] Throughout this application, various publications, patents, or patent applications have been referenced. All of these disclosures are hereby incorporated by reference in order to provide a more complete description of the state of the art to which this invention relates.

[0441] The term "includes" is intended to be open-ended in this document, encompassing not only the listed elements but also any additional elements.

[0442] Although the present invention has been described with reference to the embodiments provided above, it should be understood that various modifications can be made without departing from the invention. Therefore, the invention is limited only by the claims.

Claims

1. A computer processing system for performing a method for distinguishing abnormal methylation levels of DNA in a sample containing cell-free DNA from blood, wherein the method includes: (a) Provide the mean methylation level and standard deviation of the methylation level for each CpG site in the baseline genomic DNA of two or more normal individual baseline organisms; (b) Provide a test dataset containing: The methylation status of the plurality of CpG sites in the first test genomic DNA from an individual test organism, wherein the CpG sites are derived from the sample. (c) Determine the methylation difference between the first test genomic DNA and the baseline genomic DNA for each CpG site, thereby providing a normalized methylation difference for each CpG site; and (d) Convert the normalized methylation difference at each CpG site into a one-sided p value; (e) Determine the total methylation fraction of the combination of one-sided p values ​​for each CpG site of the first test genomic DNA; (f) Provide a training dataset, which contains: The methylation status of the plurality of CpG sites in the training genomic DNA of two or more normal individual training organisms, wherein the CpG sites are derived from a variety of different cell types from the normal individual training organisms; (g) For each CpG site, determine the methylation difference between each training genomic DNA from the normal individual training organism and the baseline genomic DNA, thereby providing the normalized methylation difference for each CpG site of each training genomic DNA; (h) The normalized methylation difference at each CpG site is converted into a one-sided p value; as well as (i) Determine the total methylation fraction of the combination of one-sided p values ​​for each CpG site in each training genomic DNA; (j) Using the total methylation fraction of each training genomic DNA, calculate the mean total methylation fraction and the standard deviation of the total methylation fraction of the training genomic DNA to obtain the distribution of the total methylation fraction of the training genomic DNA; and (k) Evaluate the total methylation score of the first test genomic DNA for the distribution of the total methylation score of the training genomic DNA.

2. The computer processing system of claim 1, wherein (a) includes providing the methylation status of the plurality of CpG sites in the baseline genomic DNA from the two or more normal individual organisms, and determining the average methylation level and the standard deviation of the methylation level of the baseline genomic DNA for each CpG site.

3. The computer processing system of claim 1, wherein the method further comprises... A second test dataset is provided, which contains: The methylation status of the plurality of CpG sites in the second test genomic DNA from the individual test organism, wherein the CpG sites are derived from the sample; For each CpG site, the methylation difference between the second test genomic DNA and the baseline genomic DNA is determined, thereby providing the normalized methylation difference for each CpG site of the second test genomic DNA; and The normalized methylation difference at each CpG site of the second test genomic DNA is converted into a one-sided p value; Determine the total methylation fraction of the combination of one-sided p values ​​for each CpG site in the second test genomic DNA; as well as The total methylation fractions of the first test genomic DNA and the second test genomic DNA are compared to determine whether the total methylation fraction between the first and second test genomic DNAs has changed.

4. The computer processing system of any one of claims 1 to 3, wherein the normalized methylation difference at a specific CpG site is determined according to the following formula: Z i χ represents the normalized methylation difference at a specific CpG site identified as i. i The methylation level at CPG site i in the first test genomic DNA or the training genomic DNA is represented by μ. i This represents the average methylation level at CpG site i in the baseline genome, and σ i The standard deviation represents the methylation level at CpG site i in the baseline genomic DNA.

5. The computer processing system of claim 1, wherein the evaluation in step (k) determines the normalized methylation fraction difference according to the following formula: Z MS The difference in normalized methylation fraction is represented by MS, which represents the total methylation fraction of the first tested genomic DNA, in μ. MS σ represents the average methylation fraction of the training set of normal genomic DNA. MS The standard deviation represents the total methylation fraction of the training set of normal genomic DNA.

6. The computer processing system of any one of claims 1 to 3, wherein the total methylation fraction MS is determined according to the following formula: Where p i denoted by , where p represents the one-sided p value at site i, and k represents the total number of CpG sites.

7. The computer processing system of any one of claims 1 to 3, wherein the total methylation fraction MS is determined according to the following formula: Where p i Let represent the one-sided p-value at site i, k represent the total number of CpG sites, and w i This indicates the coverage of site i.

8. The computer processing system of any one of claims 1 to 3, wherein the sample from the individual test organism comprises circulating tumor DNA and circulating non-tumor DNA.

9. The computer processing system of any one of claims 1 to 3, wherein the individual test organism is a pregnant woman, and the first test genomic DNA comprises genomic DNA derived from the woman's somatic cells and genomic DNA derived from the somatic cells of the woman's prenatal offspring.

10. The computer processing system of any one of claims 1 to 3, wherein providing the sample in step (b) comprises targeting a subset of genomic DNA fragments containing a predetermined set of target CpG sites.

11. The computer processing system of claim 10, wherein providing the sample in step (b) further includes treating the subset of genomic DNA fragments with bisulfite.

12. The computer processing system of any one of claims 1 to 3, wherein providing the sample in step (b) includes detecting the methylation state of the CpG site.

13. The computer processing system of claim 12, wherein detecting the methylation state of the CpG site comprises sequencing technology that sequentially identifies nucleotides in the first test genomic DNA.

14. The computer processing system of claim 5, wherein if Z MS If the value is greater than 3, the first tested genomic DNA is classified as having an abnormal methylation level.

15. A computer processing system for performing a method for identifying cancer present in an individual organism, wherein the method includes: (a) Provide a dataset containing: The methylation status of multiple CpG sites i in genomic DNA from clinical samples known to include specific cancers, wherein the clinical samples contain cell-free DNA from blood; (b) Identifying cancer type-specific hypermethylated CpG sites i, including: (i) Determine the average methylation level of each CpG site i in the genomic DNA of the clinical sample known to include the specific cancer. (ii) Determine which CpG sites i satisfy the first threshold, the second threshold, or a combination thereof. Determining the first threshold includes (1) determining the absolute value of the average methylation level for each CpG site i; (2) sorting the average methylation levels of each CpG site i from low to high; and (3) selecting those CpG sites i that have an average methylation level greater than or equal to a first preselected value in the percentage order. Determining the second threshold includes (1) determining the absolute value of the average methylation level for each CpG site i; and (2) selecting those CpG sites i that have an average methylation level greater than the second preselected value. (iii) Compile a list of hypermethylation sites specific to the stated cancer type. (c) Repeat (a) and (b) for each specific cancer to generate multiple lists of additional cancer type-specific hypermethylation sites; (d) Provide a test dataset, which includes: The methylation level of each hypermethylation site from the test genomic DNA of the individual test organism, wherein the hypermethylation site is one of the list of cancer type-specific hypermethylation sites identified in steps (b) and (c); (e) The methylation level of each of the hypermethylation sites is averaged to produce a single average methylation level of the test genomic DNA for the cancer type identified in steps (b) and (c); (f) Repeat step (e) for each cancer type to generate the average methylation level for each cancer type; (g) The average methylation levels of each cancer type are sorted from low to high, wherein the cancer type corresponding to the highest average methylation level is the cancer present in the individual test organism.

16. The computer processing system of claim 15, wherein the first preselected value is a percentile greater than or equal to 97.

17. The computer processing system of any one of claims 15 to 16, wherein the second preselected value is greater than 6%.

18. The computer processing system of any one of claims 15 to 16, wherein the test genomic DNA from the individual test organism comprises circulating tumor DNA and circulating non-tumor DNA.

19. The computer processing system of any one of claims 15 to 16, wherein the tested genomic DNA comprises cell-free DNA from blood.

20. The computer processing system of any one of claims 15 to 16, wherein providing the test dataset in step (d) comprises a subset of genomic DNA fragments containing a pre-determined set of target CpG sites.

21. The computer processing system of claim 20, wherein providing the test dataset in step (d) further comprises the subset of genomic DNA fragments treated with bisulfite.

22. The computer processing system of any one of claims 15 to 16, wherein providing the test dataset in step (d) includes detecting the methylation state of the CpG sites.

23. The computer processing system of claim 22, wherein detecting the methylation state of the CpG site comprises sequencing technology that sequentially identifies nucleotides in the test genomic DNA.

Images