Accurate detection of origin of cell-free nucleic acids

Abstract

Techniques are provided for accurately detecting a mutation in a cell-free nucleic acid shed from a tissue (e.g., a tumor) into blood. An example technique can distinguish between a mutation in a cell-free nucleic acid molecule derived from blood cells and a mutation in a cell-free nucleic acid molecule derived from a tumor. A cell-free sample (e.g., plasma or serum) and a cellular sample (e.g., white blood cells in a buffy coat) can be sequenced. After applying filtering criteria to respective sets of sequence reads, a particular mutation can be detected in both the cell-free sample and the cellular sample. Confidence intervals for the detection of the mutation can be determined for both samples, with an upper bound for the cellular sample and a lower bound of the cell-free sample being used to determine the origin of the mutation in the cell-free sample.

Description

ACCURATE DETECTION OF ORIGIN OF CELL-FREE NUCLEIC ACIDSCROSS-REFERENCES TO RELATED APPLICATIONS

[0001] The present application claims priority from and is a PCT application of U.S. Provisional Application No. 63 / 548,123, entitled “Accurate Detection Of Origin Of Cell-Free Nucleic Acids” filed November 10, 2023, the entire contents of which are herein incorporated by reference for all purposes.BACKGROUND

[0002] Nucleic acids, including both deoxyribonucleic acids (DNA) and ribonucleic acids (RNAs), can be found in extracellular, or cell-free, forms in bodily fluids. Cell-free nucleic acids (cfNAs) circulating in the blood of healthy individuals are primarily of hematopoietic origin with minimal contribution from healthy tissue. However, tumors shed detectable cfNAs into circulation through apoptosis or other mechanisms. See, e.g., Pos, O., Biro, O., Szemes, T. et al. Circulating cell-free nucleic acids: characteristics and applications. Eur J Hum Genet 26, 937-945 (2018); Stejskal, P., Goodarzi, H., Srovnal, J. et al. Circulating tumor nucleic acids: biology, release mechanisms, and clinical relevance. Mol Cancer 22, 15 (2023).

[0003] Molecular profiling refers to identifying molecular characteristics of a biological sample, including without limitation genetic, proteomic and / or expression characteristics. For example, a molecular profile may be determined using next-generation sequencing technology' to sequence genomic DNA and / or mRNA transcripts from a patient sample. When treating a subject for cancer, it can be beneficial to know the molecular profile of the tumor. Such a molecular profile can help to determine a diagnosis, prognosis, or theranosis, such as whether a treatment is likely or not to be beneficial to the patient, and whether future treatment is expected to continue to be beneficial. A patient may be monitored before and / or after treatment to determine whether the molecular profile has changed, which may allow detection of minimum residual disease (MRD) and can also affect treatment. There is also the desire to use molecular profiling to screen for disease, such as multi cancer early detection (MCED) in asymptomatic individuals.

[0004] However, it is not practical or may be impossible to perform a tissue biopsy in all settings, such as for screening applications or at regular intervals for monitoring. Even when a tumor has been identified, waiting for surgery or biopsy procedure can delay treatment decisions. Liquid biopsies are assays that aim to characterize tumors by analyzing tumor derived cfNAs in bodily fluids, using rapid and non-invasive techniques such as blood draws. Thus, molecular profiling may be performed preferentially using liquid biopsies in certain settings.

[0005] However, blood and other bodily fluids includes cfNAs from various sources that can interfere with tumor characterization based on liquid biopsy. A false positive event based on liquid biopsy, such as misattribution of a tumor as the source of a detected mutation (or wild type), can lead to misdiagnosis, and / or suggest a treatment that is not likely to be beneficial or is potentially harmful. Such misattributions are clinical false positives in the sense that they do not arise from the desired target of the liquid biopsy, such as a tumor. Clonal hematopoiesis (CH) is a phenomenon wherein certain mutations, including tumor associated mutations, accumulate in blood via clonal expansion of hematopoietic cells carry ing the mutations. Clonal hematopoiesis of indeterminate potential (CHIP) refers to cancer-associated clonal mutations. See, e.g., Reed, S., Croessmann, S., and B.H. Park, CHIP Happens: Clonal Hematopoiesis of Indeterminate Potential and Its Relationship to Solid Tumors. Clin Cancer Res 15 April 2023; 29 (8): 1403-1411. CHIP can lead to false positives in liquid biopsy. Therefore, it is desirable to provide techniques to accurately detect tumor- derived cfNAs apart from cfNAs of hematopoietic origin such as CHIP.BRIEF SUMMARY

[0006] Techniques are provided for accurately detecting a cell-free nucleic acid shed from a tissue (e.g., a tumor) into blood. An example technique can distinguish between a mutation arising from a cfNA molecule from blood cells, considered to be a false positive in the disclosure, and a mutation in a cfNA molecule arising from a tumor, considered a true mutation for purposes of this disclosure. In the example, a cell-free sample (e.g., plasma or serum) and a cellular sample (e.g., white blood cells in a buffy coat) can be sequenced. After applying filtering cnteria to respective sets of sequence reads, a particular mutation can be detected in both the cell-free sample and the cellular sample. Confidence intervals for the detection of the mutation can be determined for both samples, with an upper bound for thecellular sample and a lower bound of the cell-free sample being used to identify the origin of the mutation shed into the cell-free sample.

[0007] These and other embodiments of the disclosure are described in detail below. For example, other embodiments are directed to systems, devices, and computer readable media associated with methods described herein.

[0008] A better understanding of the nature and advantages of embodiments of the present disclosure may be gained with reference to the following detailed description and the accompanying drawings.BRIEF DESCRIPTION OF THE DRAWINGS

[0009] FIG. 1 is a flowchart illustrating a method for determining an origin of a mutation using measurements in plasma and buffy coat according to embodiments of the present disclosure.

[0010] FIG. 2 is a flowchart illustrating a method for determining an origin of a mutation using measurements in a cell-free sample and an associated cellular sample (e.g., buffy coat) according to embodiments of the present disclosure.

[0011] FIG. 3 is a block diagram of an example system for training a machine learning model.

[0012] FIG. 4 is a block diagram of a system that generates training data structures for training a machine learning model to predict a classification of a medical condition, e.g., cancer, in a subject.

[0013] FIG. 5 is a block diagram of a system for using a machine learning model that has been trained to predict a classification of a medical condition, e.g., effectiveness of a treatment for a disease or disorder of a subject having a particular molecular profile.

[0014] FIG. 6 illustrates a measurement system according to an embodiment of the present disclosure.

[0015] FIG. 7 shows a block diagram of an example computer system usable with systems and methods according to embodiments of the present disclosure.

[0016] FIG. 8 shows clonal hematopoeisis (CH) variants by patient age.

[0017] FIG. 9 shows percent of patients by cancer type with a CH variant in a gene that would drive therapy selection.

[0018] FIG. 10 shows variant classification by gDNA and plasma variant frequency.TERMS

[0019] Nucleic acids include deoxyribonucleotides or ribonucleotides and polymers thereof in either single- or double-stranded form, or complements thereof. Nucleic acids can contain known nucleotide analogs or modified backbone residues or linkages, which are synthetic, naturally occurring, and non-naturally occurring, which have similar binding properties as the reference nucleic acid, and which are metabolized in a manner similar to the reference nucleotides. Examples of such analogs include, without limitation, phosphorothioates, phosphoramidates, methyl phosphonates, chiral-methyl phosphonates, 2-O-methyl ribonucleotides, peptide-nucleic acids (PNAs). Nucleic acid sequence can encompass conservatively modified variants thereof (e.g., degenerate codon substitutions) and complementary sequences, as well as the sequence explicitly indicated. Specifically, degenerate codon substitutions may be achieved by generating sequences in which the third position of one or more selected (or all) codons is substituted with mixed-base and / or deoxyinosine residues (Batzer et al., Nucleic Acid Res. 19:5081 (1991); Ohtsuka et al., J. Biol. Chem. 260:2605-2608 (1985); Rossohm et al., Mol. Cell Probes 8:91-98 (1994)). The term nucleic acid is sometimes used interchangeably with gene, cDNA, mRNA, oligonucleotide, and polynucleotide.

[0020] A sample as used herein includes any relevant biological sample that can be used for molecular profiling, e.g., sections of tissues such as biopsy or tissue removed during surgical or other procedures, bodily fluids, autopsy samples, and frozen sections taken for histological purposes. Such samples include blood and blood fractions or products (e.g., serum, buffy coat, plasma, platelets, red blood cells, and the like), sputum, malignant effusion, cheek cells tissue, cultured cells (e.g., primary cultures, explants, and transformed cells), stool, urine, other biological or bodily fluids (e.g., prostatic fluid, gastric fluid, intestinal fluid, renal fluid, lung fluid, cerebrospinal fluid, and the like), etc. The sample can comprise biological material that is a fresh frozen & formalin fixed paraffin embedded (FFPE) block, formalin-fixed paraffin embedded, or is within an RNA preservative + formalin fixative. More than one sample of more than one type can be used for each patient. In a preferred embodiment, the sample comprises a fixed tumor sample. Unless otherwisenoted, a "sample" as referred to herein for molecular profiling of a patient may comprise more than one physical specimen. As an example, a “sample” may comprise multiple biopsy specimens, e.g., one or more surgical biopsy specimen, one or more core-needle biopsy specimen, one or more fine-needle aspiration biopsy specimen, or any useful combination thereof. As still another non-limiting example, a molecular profile may be generated for a subject using a “sample” comprising a solid tumor specimen and a bodily fluid specimen. In some embodiments, a sample is a unitary sample, i.e., a single physical specimen.

[0021] A molecular profile as used herein can refer to molecular characteristics of a biological sample, including without limitation genetic, epigenetic, proteomic, and / or expression characteristics of molecules in the biological sample. The biological sample may be a biopsy of a tumor. A summary and / or analysis of a patient / disease state based on a molecular profile may be included in a molecular profiling report, as further described herein.

[0022] A “cell-free sample” as used herein can mean any sample where a majority' of the nucleic acids are cell-free prior to any artificial fragmentation processes, e g., greater than 50%, 60%, 70%, 80%, 90%, 95%, or 99% of the nucleic acids can be cell-free. A cell-free sample can be a cell free portion of a bodily fluid such as peripheral blood, sera / serum, plasma, ascites, urine, cerebrospinal fluid (CSF), sputum, saliva, bone marrow, synovial fluid, aqueous humor, amniotic fluid, cerumen, breast milk, broncheoalveolar lavage fluid, semen, prostatic fluid, cowper’s fluid, pre- ejaculatory fluid, female ejaculate, sweat, fecal matter, tears, cyst fluid, pleural fluid, peritoneal fluid, pericardial fluid, lymph, chyme, chyle, bile, interstitial fluid, menses, pus, sebum, vomit, vaginal secretions, mucosal secretion, stool water, pancreatic juice, lavage fluids from sinus cavities, bronchopulmonary aspirates, or blastocyst cavity fluid. As an example, blood can be fractionated via centrifugation into an upper layer of cell-free clear fluid known as plasma, a lower layer of red fluid comprising granulocytes and red blood cells, and a thin layer in between known as the huffy coat, which comprises mainly white blood cells and platelets.

[0023] A “cellular sample” as used herein can mean any sample where a majority of the nucleic acids are within an intact cell, e.g., greater than 50%, 60%, 70%, 80%, 90%, 95%, or 99% of the DNA can be cellular. Examples of such samples include the buffy coat of a blood sample, any bodily fluid sample subjected to an enrichment process (e.g., centrifuging or filtration) to enrich for cells, a tissue sample, tumor biopsy, a hair follicle, and the like.

[0024] A biological sample (e.g., blood) can include both a cell-free portion and a cellular portion, which can be separated from each other. After a separation process (e.g., centrifuging), the separated portions can be considered a cell-free sample and a cellular sample, e.g., plasma and buffy coat, respectively, both derived from a same biological sample. In other embodiments, the cell-free sample and the cellular sample can be derived from different biological samples.

[0025] In various embodiments, at least 1,000, 5,000, 10,000 or 50,000 or 100,000 or 500,000 or 1,000,000 or 5,000,000 nucleic acid molecules, or more, can be analyzed. At least a same number of sequence reads can be analyzed. Examples sizes of a sample can include 30, 50, 100, 200, 300, 500, 1,000, 5,000, or 10,000 or more nanograms, or 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10 ml.

[0026] The term “phenotype” as used herein can mean any trait or characteristic that can be identified in part or in whole by using the systems and / or methods provided herein. In some embodiments, the systems can include one or more computer programs on one or more computers in one or more locations, e.g., configured for use in a method described herein. Phenotypes may be determined by analyzing a biological sample obtained from a subject. Phenotypes to be characterized can be any phenotype of interest, including without limitation a tissue, anatomical origin, medical condition, ailment, disease, disorder, or useful combinations thereof. A phenotype can be any observable characteristic or trait of, such as a disease or disorder, a stage of a disease or disorder, susceptibility to a disease or disorder, prognosis of a disease stage or disorder, a physiological state, or response I potential response (or lack thereof) to interventions such as therapeutics. A phenotype can result from a subject’s genetic makeup as well as the influence of environmental factors and the interactions between the two, as well as from epigenetic modifications to nucleic acid sequences. In various embodiments, a phenotype in a subject is characterized by obtaining a biological sample from a subject and analyzing the sample using the systems and / or methods provided herein. For example, characterizing a phenotype for a subject or individual can include detecting a disease or disorder (including pre-symptomatic early -stage detection), determining a prognosis, diagnosis, or theranosis of a disease or disorder, or determining the stage or progression of a disease or disorder. Characterizing a phenotype can include identifying appropriate treatments or treatment efficacy for specific diseases, conditions, disease stages and condition stages, predictions and likelihood analysis of disease progression, particularly disease recurrence, metastatic spread or disease relapse. Aphenotype can also be a clinically distinct type or subty pe of a condition or disease, such as a cancer or tumor. Phenotype determination can also be a determination of a physiological condition, or an assessment of organ distress or organ rejection, such as post-transplantation. The compositions and methods described herein allow assessment of a subject on an individual basis, which can provide benefits of more efficient and economical decisions in treatment.

[0027] A medical condition as used herein can refer to a disease or disorder of a subject, as well as a stage or severity of the medical condition.

[0028] A subject (individual, patient, or the like) can be any animal which may benefit from the methods described herein. A subject or individual can be any animal which may benefit from the methods described herein, including, e.g., humans and non-human mammals, such as primates, rodents, horses, dogs and cats. Subjects include without limitation a eukaryotic organisms, most preferably a mammal such as a primate, e g., chimpanzee or human, cow; dog; cat; a rodent, e.g., guinea pig, rat, mouse; rabbit; or a bird; reptile; or fish. Subjects specifically intended for treatment using the methods described herein include humans. A subject may also be referred to herein as an individual or a patient. The subject can have a pre-existing disease or disorder, including without limitation cancer. Alternatively, the subject may not have any known pre-existing condition. The subject may also be non-responsive to an existing or past treatment, such as a treatment for cancer.

[0029] Theranostics as used herein includes therapy -related diagnostic testing that provides the ability to affect therapy or treatment of a medical condition such as a disease or disease state. Theranostics testing provides a theranosis in a similar manner that diagnostics or prognostic testing provides a diagnosis or prognosis, respectively. As used herein, theranostics encompasses any desired form of therapy related testing, including predictive medicine, personalized medicine, precision medicine, integrated medicine, pharmacodiagnostics and Dx / Rx partnering. Therapy related tests can be used to predict and assess drug response in individual subjects, thereby providing personalized medical recommendations. Predicting a likelihood of response can be determining whether a subject is a likely responder or a likely non-responder to a candidate therapeutic agent, e.g., before the subject has been exposed or otherwise treated with the treatment. Assessing a therapeutic response can be monitoring a response to a treatment, e.g., monitoring the subject’s improvement or lack thereof over a time course after initiating the treatment. Therapy relatedtests are useful to select a subject for treatment who is particularly likely to benefit or lack benefit from the treatment or to provide an early and objective indication of treatment efficacy in an individual subject. Characterization using the systems and methods provided herein may indicate that treatment should be altered to select a more promising treatment, thereby avoiding the expense of delaying beneficial treatment and avoiding the financial and morbidity costs of less efficacious or ineffective treatment(s).

[0030] Theranosis can comprise predicting a treatment efficacy or lack thereof, classifying a patient as a responder or non-responder to treatment. A predicted “responder” can refer to a patient likely to receive a benefit from a treatment whereas a predicted “non-responder” can be a patient unlikely to receive a benefit from the treatment. Unless specified otherwise, a benefit can be any clinical benefit of interest, including without limitation cure in whole or in part, remission, or any improvement, reduction or decline in progression of the condition or symptoms. The theranosis can be directed to any appropriate treatment, e.g., the treatment may comprise at least one of chemotherapy, immunotherapy, targeted cancer therapy, a monoclonal antibody, small molecule, surgery, radiation, or any useful combinations thereof.

[0031] A classification of a medical condition can include a diagnosis, prognosis, or theranosis of the subject. The term “ctovz zcafioM” as used herein refers to any number(s) or other characters(s) that are associated with a particular property of a sample, e g., a medicinal condition of a subject from whom the sample was obtained. For example, a “+” symbol (or the word “positive”) could signify that a sample is classified as having deletions or amplifications. The classification can be binary (e.g., positive or negative) or have more levels of classification (e.g., a scale from 1 to 10 or 0 to 1), including probabilities.

[0032] The terms “genetic variant,” “nucleotide variant,” “variant,” and “mutation” may be used herein interchangeably to refer to changes or alterations to the reference human gene sequence at a particular locus, including, but not limited to, nucleotide base point mutations, polymorphisms, deletions, insertions, inversions, substitutions, translocations, rearrangements, fusions, breaks, duplications, amplifications, or repeats in the coding and non-coding regions. Certain mutations may be detectable on a given molecule, e.g., using long read techniques, such as single molecule sequencing such as Single-molecule real-time (SMRT) sequencing and nanopore sequencing. Deletions may be of a single nucleotide base, a portion or a region of the nucleotide sequence of the gene, or of the entire gene sequence. Insertions may be of one or more nucleotide bases. The variants may occur in transcriptionalregulatory regions, untranslated regions of mRNA, exons, introns, exon / intron junctions, etc. Variants can potentially result in stop codons, frame shifts, deletions of amino acids, altered gene transcript splice forms or altered amino acid sequence.

[0033] An allele or gene allele comprises generally a naturally occurring gene having a reference sequence or a gene containing a specific nucleotide variant.

[0034] A haplotype refers to a combination of genetic (nucleotide) variants in a region of an mRNA or a genomic DNA on a chromosome found in an individual. Thus, a haplotype includes a number of genetically linked polymorphic variants which are typically inherited together as a unit.

[0035] The term “locus” refers to a specific position or site in a gene sequence or protein. Thus, there may be one or more contiguous nucleotides in a particular gene locus, or one or more amino acids at a particular locus in a polypeptide. Moreover, a locus may refer to a particular position in a gene where one or more nucleotides have been deleted, inserted, or inverted.

[0036] A “biomarker” or “marker” can comprise a gene and / or gene product depending on the context.

[0037] The terms “label” and “detectable label” can refer to any composition detectable by spectroscopic, photochemical, biochemical, immunochemical, electrical, optical, chemical or similar methods. Such labels include biotin for staining with labeled streptavidin conjugate, magnetic beads (e.g., DYNABEADS™), fluorescent dyes (e.g., fluorescein, Texas red, rhodamine, green fluorescent protein, and the like), radiolabels (e.g., 3H, 1251, 35S, 14C, or 32P), enzymes (e.g., horse radish peroxidase, alkaline phosphatase and others commonly used in an ELISA), and calorimetric labels such as colloidal gold or colored glass or plastic (e.g., polystyrene, polypropylene, latex, etc) beads. Patents teaching the use of such labels include U.S. Pat. Nos. 3,817,837; 3,850,752; 3,939,350; 3,996,345; 4,277,437; 4,275,149; and 4,366,241. Means of detecting such labels are well known to those of skill in the art. Thus, for example, radiolabels may be detected using photographic film or scintillation counters, fluorescent markers may be detected using a photodetector to detect emitted light. Enzymatic labels are typically detected by providing the enzyme with a substrate and detecting the reaction product produced by the action of the enzyme on the substrate, and calorimetric labels are detected by simply visualizing the colored label. Labels can include, e.g., ligands that bind to labeled antibodies, fluorophores, chemiluminescent agents, enzymes,and antibodies which can serve as specific binding pair members for a labeled ligand. An introduction to labels, labeling procedures and detection of labels is found in Polak and Van Noorden Introduction to Immunocytochemistry, 2nd ed., Springer Verlag, NY (1997); and in Haugland Handbook of Fluorescent Probes and Research Chemicals, a combined handbook and catalogue Published by Molecular Probes, Inc. (1996).

[0038] Detectable labels include, but are not limited to, nucleotides (labeled or unlabelled), compomers, sugars, peptides, proteins, antibodies, chemical compounds, conducting polymers, binding moieties such as biotin, mass tags, calorimetric agents, light emitting agents, chemiluminescent agents, light scattering agents, fluorescent tags, radioactive tags, charge tags (electrical or magnetic charge), volatile tags and hydrophobic tags, biomolecules (e.g., members of a binding pair antibody / antigen, antibody / antibody, antibody / antibody fragment, antibody / antibody receptor, antibody / protein A or protein G, hapten / anti -hapten, biotin / avidin, biotin / streptavidin, folic acid / folate binding protein, vitamin B12 / intrinsic factor, chemical reactive group / complementary chemical reactive group (e.g., sulfhydryl / maleimide, sulfhydr l / haloacetyl derivative, amine / isotriocyanate, amine / succinimidyl ester, and amine / sulfonyl halides) and the like.

[0039] The terms "‘primer”, “probe,” and “oligonucleotide” are used herein interchangeably to refer to a relatively short nucleic acid fragment or sequence. They can comprise DNA, RNA, or a hybrid thereof, or chemically modified analog or derivatives thereof. Typically, they are single-stranded. However, they can also be double-stranded having two complementing strands which can be separated by denaturation. Normally, primers, probes and oligonucleotides have a length of from about 8 nucleotides to about 200 nucleotides, preferably from about 12 nucleotides to about 100 nucleotides, and more preferably about 18 to about 50 nucleotides. They can be labeled with detectable markers or modified using conventional manners for various molecular biological applications.

[0040] The term “isolated” when used in reference to nucleic acids (e.g., genomic DNAs, cDNAs, mRNAs, or fragments thereof) is intended to mean that a nucleic acid molecule is present in a form that is substantially separated from other naturally occurring nucleic acids that are normally associated with the molecule. Because a naturally existing chromosome (or a viral equivalent thereof) includes a long nucleic acid sequence, an isolated nucleic acid can be a nucleic acid molecule having only a portion of the nucleic acid sequence in the chromosome but not one or more other portions present on the same chromosome. Morespecifically, an isolated nucleic acid can include naturally occurring nucleic acid sequences that flank the nucleic acid in the naturally existing chromosome (or a viral equivalent thereof). An isolated nucleic acid can be substantially separated from other naturally occurring nucleic acids that are on a different chromosome of the same organism. An isolated nucleic acid can also be a composition in which the specified nucleic acid molecule is significantly enriched so as to constitute at least 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, or at least 99% of the total nucleic acids in the composition.

[0041] An isolated nucleic acid can be a hybrid nucleic acid having the specified nucleic acid molecule covalently linked to one or more nucleic acid molecules that are not the nucleic acids naturally flanking the specified nucleic acid. For example, an isolated nucleic acid can be in a vector. In addition, the specified nucleic acid may have a nucleotide sequence that is identical to a naturally occurring nucleic acid or a modified form or mutein thereof having one or more mutations such as nucleotide substitution, deletion / insertion, inversion, and the like.

[0042] An isolated nucleic acid can be extracted from a tissue sample, prepared from a recombinant host cell (in which the nucleic acids have been recombinantly amplified and / or expressed), or can be a chemically synthesized nucleic acid having a naturally occurring nucleotide sequence or an artificially modified form thereof.

[0043] For the purpose of comparing two different nucleic acid or polypeptide sequences, one sequence (test sequence) may be described to be a specific percentage identical to another sequence (comparison sequence). The percentage identity can be determined by the algorithm of Karlin and Altschul, Proc. Natl. Acad. Sci. USA, 90:5873-5877 (1993), which is incorporated into various BLAST programs. The percentage identity can be determined by the “BLAST 2 Sequences” tool, which is available at the National Center for Biotechnology Information (NCBI) website. See Tatusova and Madden, FEMS Microbiol. Lett., 174(2):247- 250 (1999). For pairwise DNA-DNA comparison, the BLASTN program is used with default parameters (e.g., Match: 1; Mismatch: -2; Open gap: 5 penalties; extension gap: 2 penalties; gap x_dropoff: 50; expect: 10; and word size: 11, with filter). For pairwise protein-protein sequence comparison, the BLASTP program can be employed using default parameters (e.g., Matrix: BLOSUM62; gap open: 11; gap extension: 1; x_dropoff: 15; expect: 10.0; and wordsize: 3, with filter). Percent identity of two sequences is calculated by aligning a test sequence with a comparison sequence using BLAST, determining the number of amino acidsor nucleotides in the aligned test sequence that are identical to amino acids or nucleotides in the same position of the comparison sequence, and dividing the number of identical amino acids or nucleotides by the number of ammo acids or nucleotides in the comparison sequence. When BLAST is used to compare two sequences, it aligns the sequences and yields the percent identity over defined, aligned regions. If the two sequences are aligned across their entire length, the percent identity yielded by the BLAST is the percent identity of the two sequences. If BLAST does not align the two sequences over their entire length, then the number of identical amino acids or nucleotides in the unaligned regions of the test sequence and comparison sequence is considered to be zero and the percent identity is calculated by adding the number of identical amino acids or nucleotides in the aligned regions and dividing that number by the length of the comparison sequence. Various versions of the BLAST programs can be used to compare sequences, e.g., BLAST 2.1.2 or BLAST+ 2.2.22.

[0044] A "'machine learning moder (ML model) can refer to a software module configured to be run on one or more processors to provide a classification or numerical value of a property of one or more samples. An ML model can include various parameters (e.g., for coefficients, weights, thresholds, functional properties of function, such as activation functions). As examples, an ML model can include at least 10, 100, 1,000, 5,000, 10,000, 50,000, 100,000, or one million parameters. An ML model can be generated using sample data (e.g., training samples) to make predictions on test data. Various number of training samples can be used, e.g., at least 10, 100, 1,000, 5,000, 10,000, 50,000, 100,000, or at least 200,000 training samples. One example is an unsupervised learning model such as hidden Markov model (HMM), clustering (e.g., hierarchical clustering, k-means, mixture models, model-based clustering, density-based spatial clustering of applications with noise (DBSCAN), and OPTICS algorithm), approaches for learning latent variable models such as Expectation-maximization algorithm (EM), method of moments, and blind signal separation techniques (e.g., principal component analysis, independent component analysis, nonnegative matrix factorization, singular value decomposition), and anomaly detection (e.g., local outlier factor and isolation forest). Another example type of model is supervised learning that can be used with embodiments of the present disclosure. Example supervised learning models may include different approaches and algorithms including analytical learning, statistical models, artificial neural network (e.g. including convolutional and / or transformer layers) that may have 1-10 layers as examples, recurrent neural network (e.g., long short term memory, LSTM), boosting (meta-algorithm), bootstrap aggregating (bagging)such as random forests, support vector machine (SVM), support vector (SVR), Bayesian statistics, case-based reasoning, decision tree learning, inductive logic programming, linear regression, logistic regression, Gaussian process regression, genetic programming, group method of data handling, kernel estimators, learning automata, learning classifier systems, minimum message length (decision trees, decision graphs, etc.), multilinear subspace learning, naive Bayes classifier, maximum entropy classifier, conditional random field, nearest neighbor algorithm, probably approximately correct learning (PAC) learning, ripple down rules, a knowledge acquisition methodology, symbolic machine learning algorithms, subsymbolic machine learning algorithms, minimum complexity machines (MCM), ordinal classification, data pre-processing, handling imbalanced datasets, statistical relational learning, or Proaftn (a multicriteria classification algorithm), or an ensemble of any of these types. Supervised learning models can be trained in various ways using various cost / loss functions that define the error from the known label (e.g., least squares and absolute difference from known classification) and various optimization techniques, e.g., using backpropagation, steepest descent, conjugate gradient, and Newton and quasi-Newton techniques.

[0045] The term “about” or “approximately” can mean within an acceptable error range for the particular value as determined by one of ordinary skill in the art, which will depend in part on how the value is measured or determined, i.e., the limitations of the measurement system. For purposes of the disclosure, “about” refers to ±10% unless otherwise stated.

[0046] Where a range of values is provided, it is understood that each intervening value, to the tenth of the unit of the lower limit unless the context clearly dictates otherwise, between the upper and lower limits of that range is also specifically disclosed. Each smaller range between any stated value or intervening value in a stated range and any other stated or intervening value in that stated range is encompassed within embodiments of the present disclosure. The upper and lower limits of these smaller ranges may independently be included or excluded in the range (e.g., range can be greater than or less than specified number), and each range where either, neither, or both limits are included in the smaller ranges is also encompassed within the present disclosure, subject to any specifically excluded limit in the stated range. Where the stated range includes one or both of the limits, ranges excluding either or both of those included limits are also included in the present disclosure.DETAILED DESCRIPTION

[0047] The molecular profile of tumors can be helpful for determining a diagnosis, prognosis and / or theranosis. In certain settings, liquid biopsies can be used to obtain molecular profiles. For example, cell-free nucleic acids from tumors are shed into blood. Profiling of such tumor derived cfNAs can be used for detection, monitoring, and characterization of cancers. But mutated nucleic acids from blood cells are also found in plasma, e.g., through clonal hematopoiesis (CH) or other natural processes. When nucleic acids in plasma are sequenced, the detection of such mutations from blood cells might mislead as to the molecular profile (e.g., genetic and / or epigenetic composition) of a tumor. Such a false positive mutation can lead to incorrect diagnosis, prognosis, or theranosis, which may lead to ineffective or even detrimental treatment of the patient.

[0048] Techniques are provided herein for accurately detecting a mutation in cfNAs shed from a tissue (e.g., a tumor) into blood. An example technique can distinguish between a mutation in a cell-free nucleic acid molecule from blood cells and a mutation in a cell-free nucleic acid molecule from a tumor. For purposes of tumor characterization such as described herein, a “false positive” mutation generally refers to a mutation in cfNAs that is not derived from the tumor, whereas a “true” mutation is truly present within the tumor. Mutations in one or more genomic positions can be analyzed to determine whether the mutation is from blood cells or from other tissue from which a cell-free nucleic acid molecule has been shed into the cell-free sample.

[0049] A cell-free sample (e.g., plasma or serum) and a cellular sample (e.g., white blood cells in a buffy coat) can be sequenced. For example, nucleic acids (e.g., DNA or RNA) in plasma and the buffy coat can be separately sequenced using the same or different blood samples and in the same or different sequencing runs.

[0050] A filtering analysis can be performed for each set of nucleic acids (e.g., by analyzing the corresponding sequence reads) to determine whether the mutation is present. The filtering analysis can use a number of sequence reads with a mutation at a given genomic position, along with a total number of sequence reads at the given position for a given sample. A first filtering criterion based on the number of mutated sequence reads can perform an initial filtering out of false positive mutations having minimal support, e.g., less than a minimum threshold, as may occur due to sequencing errors. For example, a log likelihood ratio can be used for such a purpose. A second filtering criterion can use the positions in thesequence reads where the mutation was identified. Occurrence of the mutation, often at the ends of the sequence reads, can indicate a false positive. A median position of the mutation in a read can be used.

[0051] After applying filtering criteria to respective sets of sequence reads, a particular mutation can be detected in both the cell-free sample and the cellular sample. Confidence intervals (e.g., about 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, or 99% confidence) for a frequency of the detection of the mutation can be determined for both samples, with an upper bound for the cellular sample and a lower bound of the cell-free sample being used to identify whether the mutation has been shed into the cell-free sample. For example, the confidence interval can be for how many of the first set of cell-free nucleic acid molecules have a first mutation at the first genomic position. In some embodiments, a 95% confidence interval is chosen.

[0052] Accordingly, if sequencing of both samples indicates presence of the mutation after filtering criteria (e.g., using number of mutation reads and position within the read) are applied, then a confidence interval analysis can be performed based on the measurements of the mutation in the respective samples. If a lower bound for the cell-free sample is greater than the upper bound for the huffy coat, then the mutation in the cell-free sample can be confidently identified as not derived from the white blood cells.

[0053] Knowledge of one or more mutations at one or more genomic positions can be used in various ways. For example, various embodiments can determine a molecular profile, classification (e.g., diagnosis, prognosis, and / or theranosis) for a medical condition based on one or more mutations at one or more genomic positions being identified as shed into the cell -free sample from non-hematopoietic tissue of the subject. Treatment can be determined and administered based an in identified mutation as further described herein. Such determinations can be made using one or more machine learning models.I. USE OF FRACTIONATION OF BLOOD SAMPLE TO DETERMINE ORIGIN OF MUTATION

[0054] Some embodiments can use sequencing information obtained from the buffy coat to inform whether a mutation in the plasma / serum is a true mutation in the tumor or a false positive mutation from the buffy coat.A. Confounding signals in different fractions of a blood sample

[0055] A blood sample can be fractionated into different parts, including plasma, which is cell-free, and the huffy coat, which is primarily comprised of white blood cells. When peripheral blood is separated by centrifugation, a dark red / purple layer forms at the bottom, which comprises platelets and red blood cells. The uppermost layer is plasma, which is yellow. The buffy coat is a cloudy later that forms between the cells and the plasma. An analysis of both the plasma and the buffy coat can help to determine a true origin of a mutation.

[0056] Liquid biopsies typically examine the cell free layer such as plasma. Liquid biopsy assays aim to detect mutations in a bodily fluid that are derived from the tumor. For example, if a KRAS mutation is detected in the plasma, it may be assumed that the mutation is from the tumor. Accordingly, a certain treatment may be prescribed based on the presence of that KRAS mutation in cfNAs in the plasma. For example, a doctor may stop administering EGFR inhibitors to a patient for whom a KRAS mutation was detected in the plasma, as the KRAS mutation may indicate that the tumor is developing resistance.

[0057] However, the mutation (such as KRAS in the example above) could have originated from blood cells. The immune cells (white blood cells) in the buffy coat can have mutations as well. Immune cells divide frequently. Mutations can develop in populations of clones via CH, and may result in CHIP. Such mutations may be benign. For example, an immune cell can have a mutation in a gene that it is not using, e.g., not transcribed. Those cells will continue to replicate with the mutation present. That mutation may be shed into the plasma just as a mutation could be shed into the plasma by a tumor. Such a mutation of hematopoietic origin can be confounding, as its detection in plasma might cause one to believe it is from a tumor. For example, if the KRAS mutation is from the buffy coat, a treatment decision of stopping the administration of EGFR inhibitors may be detrimental to the patent.

[0058] Such a confounding signal may be even more problematic in older subjects, for which cancer is more prevalent. Most advanced cancer patients are older. The older a subject is, the more likely the subject has accumulated mutations in the white blood cells, including but not limited to CHIP. See, e.g., Example 2 herein. Children generally do not have CH mutations. Over the years, a person can develop such clones in the bone marrow that have a mutation, and the clones persist. The blood cells may not be harmed, as the mutation may notalter their fitness. However, the same mutation within a tumor may lead to a certain diagnosis, prognosis and / or theranosis.

[0059] Example 2 herein describes a study that illustrates the prevalence of such a problem. In the Example, 39.5% of 11,914 patients presented at least one pathogenic or likely pathogenic CH variant among reportable clinical genes. For example, we found 79.9% of all BRCA2 variants to be of CH origin, as well as 79.4% of CHEK2, 68.5% of BRCA1, 41.9% of ATM, 6.3% NRAS, 6.2% BRAF, 2.4% KIT, 2.3% KRAS, and 1.8% EGFR. For patients aged 65-69, the median rate of CH variant classification was 17%, whereas it was 29% for patients aged 70-74, 33% for ages 75-79, and 50% for ages 80+. We found high rates of CH detected in what would be otherwise druggable targets in many cancer types typically treated with PARP inhibitors, including breast, female genital tract, ovarian, pancreatic, prostate, and endometrial cancers. These results indicate that the presence of a confounding signal in plasma as a result of nucleic acid molecules from the blood cells is common and must be accounted for to optimize patient care.B. Possible origins of mutation detected in plasma / serum

[0060] A mutation detected in cell-free samples such as plasma / serum or other bodily fluids can be from various origins, e.g., germline, a mosaicism, CH (including CHIP), or from a non-hematopoietic tissue (e.g., a tumor).

[0061] Germline mutations are changes to one’s DNA that are inherited from the egg and sperm cells during conception. Such mutations can predispose one to hereditary cancer. About 15% of all cancers are hereditary. BRCA mutations are such an example, leading to highly penetrant breast cancer. Germline mutations can be found in either half a person’s cells or all the cells depending on if it is inherited from one or both parents, respectively.

[0062] A mosaicism occurs when a person has two or more genetically different sets of cells in his or her body. Mosaicism can occur on a small, unnoticeable level. When cells divide and multiply, mutations may arise (e.g., due to error in the copying process) resulting in a different genetic sequence. The cells carrying the mutations may simply die; or survival may result in mosaicism. The percentage of cells with the mosaicism depends on how early or late in the development of the person it occurs. The impact of mosaicism can be determined by the severity of the mutation in the DNA, as well as the degree of mosaicism.

[0063] Clonal hematopoiesis (CH) is the presence of a clonally expanded hematopoietic stem cells. Hematopoietic stem cells (HSCs) are multipotent cells that give rise to differentiated blood cells including red blood cells, white blood cells, and platelets. HSCs are found in umbilical cord blood, peripheral blood, and adult bone marrow. HSCs with certain mutations (e.g., in ASXL1, JAK2, DNMT3A, TET2) have a competitive advantage, such that they are able to self-renew more often than HSCs without such mutations. This leads to clonal hematopoiesis, in which hematopoietic stem cells (HSCs) clonally expand into subpopulations. CH is an age-related condition, with DTA mutations begin accumulating in HSCs around middle age. CHIP can lead to cancer related mutations, referred to as clonal hematopoiesis of indeterminate potential (CHIP).

[0064] Some embodiments of the system and methods provided herein can determine the origin of a mutation. If the germline or blood origins (e.g., CHIP) can be ruled out, then the origin of the mutation can be identified as from a non-hematopoietic tissue (e.g., a tumor). Accordingly, some embodiments can determine whether a mutation detected in plasma / serum is derived from the white blood cells. If hematopoietic cells can be ruled out as the origin of the mutation (i.e., mutation is not from the buffy coat), then the mutation can be called with high confidence as being shed from non-hematopoietic tissue in the body (e.g., a tumor). In such case, clinical decisions (e.g., treatment) can account for the mutation. In a non-limiting example, an alternate treatment regimen may be prescribed if the mutation indicates that the tumor is now resistant to current treatment. However, if the mutation can be identified as derived from the white blood cells, then current treatment may be maintained.IL DETECTING MUTATIONS IN BUFFY COAT AND PLASMA

[0065] As an initial process, embodiments provided herein can determine whether a mutation is present in a given sample, e.g., the cell-free sample (such as plasma / serum) or the cellular sample, such as huffy coat. This process of calling a mutation (also referred to as a variant) can use nucleic acid sequencing results, namely sequence reads to determine whether a mutation is present or a sequencing error has occurred.

[0066] Nucleic acid sequences and variants can be determined using any number of laboratory techniques, including without limitation polymerase chain reaction (PCR), in situ hybridization, amplification, hybridization, microarray, nucleic acid sequencing, dye termination sequencing, pyrosequencing, nanopore sequencing, next generation sequencing (NGS; also referred to as high-throughput sequencing (HTS) or massively parallel signaturesequencing (MPSS)), targeted sequencing (e.g., using capture probes or primers), whole exome sequencing (WES), whole genome sequencing (WGS), and whole transcriptome sequencing (WTS). In some embodiments, NGS is used for molecular profiling as the technique can perform thousands of sequencing reads in a single assay. Additionally, variants can be identified by aligning (mapping) sequence reads obtained from sequencing to a reference genome. Various levels of sequencing (read) depth may be used for various parts of the genome, e.g., 3x, 5, lOx, 20x, 50x, lOOx, 200x, 400x, 500x, l,000x, l,500x, or higher.

[0067] For the cell-free sample, the nucleic acid molecules may be of a size conducive to sequencing. In such cases, no fragmentation processing is necessary after the cell-free sample is obtained. For the cellular sample, when sequencing DNA, a chromosome can be fragmented into usable nucleic acid molecules, e.g., less than 1,000, 900, 800, 700, 600, 500, 400, 300, or less than 200 bases. Even when long read technology is used, a chromosome can be fragmented in some manner, e.g., mechanically (such as sonication) or enzymatic fragmentation (e.g., using random primers in an amplification process, such as PCR). When sequencing RNA, a fragmentation process can be performed to ensure nucleic acid molecules of a proper size are obtained. Additionally, single molecule sequencing techniques can be used without fragmentation, e.g., a mRNA has an average size of around 1,500 bases.

[0068] The sequencing assay provides one or more sequence reads for a given nucleic acid molecule. Depending on the sequencing technology employed, a sequence read can be for only one end of the given nucleic acid molecule. For example, a sequence read can correspond to N bases at the end of the nucleic acid molecule (e.g., where N is 200, 150, 100, 50, or less bases). Paired-end sequencing of both ends may be performed. Alternatively, a single read may be obtained for the entire nucleic acid molecule, e.g., via single molecule sequencing techniques, such as single-molecule real-time (SMRT) sequencing or nanopore techniques. The single read would cover both ends of the nucleic acid molecule. The sequence reads can be determined using base calling techniques, as will be appreciated by the skilled person. Illumina next-generation sequencing (NGS) systems (Illumina, Inc., San Diego, CA), such as used in Example 1 and 2 herein, can be used for either paired-end and single-read sequencing as desired.

[0069] The sequence read(s) can be aligned (mapped) to each other or to a reference genome, so as to determine a genomic position for which a mutation can be identified. A variation from the reference genome can be identified as a mutation. The reference genomecan be of another person, a population, or of the subject being tested (e.g., as determined from the buffy coat or a predominance of reads in the cell-free sample). The skilled person will appreciate the different alignment tools that may be used, as numerous software tools are available, such as Bowtie and Blast.

[0070] In addition to a genomic position of the mutation, a read position can be determined for each read that exhibits a mutation. For example, a mutation can occur on a sequence read at a position that is four bases from an end of the nucleic acid molecule. Such a position can account for sequence adapters and barcodes that are added to the nucleic acid fragments in order to facilitate sequencing reactions. Thus, the read position refers to the position of the sequence read that corresponds to the target nucleic acid molecule. This read position can be an indicator of whether the mutation is a true positive or is due to errors in the sequencing process, as such errors are more prevalent near the end of the nucleic acid molecule.

[0071] A filtering process can be applied to the sequencing results to determine whether a mutation detected at a given genomic originated from the biological sample or is due to technical error (e.g., base misincorporation during amplification). Various filtering criteria can be used. For example, one filtering criterion can determine whether a sufficient number (e.g., greater than a threshold) of sequence reads have the mutation. The threshold can depend on the sequencing depth at that position, i.e., the number of reads that aligned to that genomic position. One such filtering criterion can use an odds ratio, which can use a Baysian analysis. Another filtering criterion can use a read position of the mutation, e.g., a median, mean (average), or other statistical value of the mutation in the set of reads that include the mutation at the genomic position. If the statistical value is below a threshold, then the mutation can be filtered out as likely due to sequencing errors.A. Count of mutated sequence reads (e.g., odds ratio)

[0072] Some embodiments can use a filtering criterion (parameter) of an odds ratio, e.g., a log odds ratio. As examples, a tumor (variant) log odds ratio or a normal log odds ratio can be used. The odds ratio can be calculated independently of any matched normal. Variants with a odds ratio exceeding a threshold pass filtering and can be identified as a true mutation, and variants with an odds ratio below the threshold can be filtered out as due to sequencing error. Different thresholds can be used for the different samples. For example, a threshold of around two (e.g., + / - 10% or 20%) can be used for the cellular sample, indicating it is twice as likely to be a true positive than not. For a cell-free sample, a threshold of around six (e.g., + / - 10%or 20%) can be used. Other example parameters for such a filtering criterion include a p- value.

[0073] Such a count parameter can relate to how many mutations are found at a given genomic position relative to how many are expected. The expected mutational rate can be measured per mega base. Thus, a different expected rate can be used, depending on which region the genomic position is in.

[0074] Such a count parameter can also use sequencing results for a set of normal samples on a given sequencing platform. A rate of sequencing errors (variants called but assumed errors as normal samples are used) can be determined for a platform’s chemistry. The amount of variation in the particular position can be used to determine an expected mutation rate for a given position. The amount of variation in normal samples can be viewed as a noise distribution. Additionally or alternatively, a set of samples (cell-free or cellular) from subjects with a tumor can be analyzed to determine a non-noise distribution. The normal samples could be from the same subject, e.g., a buffy coat sample.

[0075] In some embodiments, a filtenng criterion can use the number of nucleic acid molecules that have the mutation by determining a probability that the genomic position includes the respective mutation based on the number of observed sequence reads having the mutation and an expected number of sequence reads having the mutation as a result of one or more errors (e.g., sequencing errors). For example, a count parameter (e.g., an odds ratio to be used as a filtering criterion) can be determined using Baysian analysis. A posterior probability (e.g., as an odds ratio) can be determined as using the number of sequence reads that included the mutation at the genomic position and a prior probability. A prior probability (expected rate) can be determined for a normal sample at the given sequencing depth (i.e., number of total reads at that position). The prior probability distribution can be determined based on normal samples that are analyzed or based on an assumed probability distribution, e.g., a normal distribution or a binomial distribution. In some embodiments, the prior probability can be the same at all genomic positions. The probability can be compared to a threshold, which can be determined using normal samples or ones with known mutations or other known properties. The threshold may be adjusted to reach certain statistical criteria, e.g., a false positive rate, false negative rate, or accuracy, such as positive predictive value.

[0076] In some embodiments, a reference model can assume there is no variant at the site and any observed non-reference bases are due to random sequencing errors; and a variantmodel can assume the site contains a true variant allele at an allele fraction F in addition to sequencing errors. The allele fraction F can be estimated as the fraction of tumor reads that support the mutation. A mutation at a given position can be initially identified. A mutation can be called if the log-likelihood ratio of the data under the variant and reference models (e.g., log odds ratio) exceeds a threshold that depends on the expected mutation frequency and the desired error rate. The choice of threshold can be used to control the tradeoff between specificity and sensitivity. Further details can be found in Cibulskis et al., “Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples.” Nat Biotechnol. 2013 Mar;31(3):213-9. doi: 10.1038 / nbt.2514. Epub 2013 Feb 10. PMID: 23396013; PMCID: PMC3833702.

[0077] Some embodiments can independently determine whether a variant exists in the cell-free sample (e.g., plasma) and the cellular sample (e.g., huffy coat).B. Read position

[0078] Sequencing errors occur more frequently at the ends of a nucleic acid molecule. If the read positions of a mutation detected at a genomic position tend to occur near the end of the nucleic acid molecule, then that mutation can be filtered out as resulting from error. The end of a nucleic acid molecule can be specified as a threshold position or distance, e.g., 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, or 20 bases from the end of the nucleic acid molecule. If there is equal probability of a DNA molecule being fragmented at any location in a reference genome (or at least similar probabilities), then a true mutation should be toward the center of a nucleic acid molecule. In a sequencing system using baits to capture regions of interest, greater bait coverage over a genomic region can increase the probability of observing sequences away from the ends of the sequenced molecules. The read position is only within the given sequence read and does not account for the genomic position within a genome.

[0079] A parameter for the read position can be determined from the set of sequence reads that have the mutation at the given genomic position. For example, a median or mean read position can be determined and compared to a threshold (e.g., 10, 9, 8, 7, 6, 5, 4, or 3). If the read position parameter is below the threshold, then the mutation can be filtered out. The read position for a given nucleic acid molecule can be taken from the closest end.

[0080] The sequencing protocol can include adding adapters and / or barcodes to the ends of the nucleic acid molecules to be sequenced. Such additions may be required to performsequencing reactions but can add noise that causes sequencing errors to occur even further from the end of the nucleic acid molecule. In such situations, a larger threshold can be used.

[0081] The threshold can be determined based on a distribution of the parameter for sequence errors from reference samples compared to a distribution of the parameter for confirmed mutations (e.g., obtained from a subject known to have a tumor). As an example, a reference sample can include a library of synthetic nucleic acid molecules that have mutations at known read positions, at a known rate, or a known distribution of read position. Different samples can have mutations at different positions, which can allow a threshold to be selected precisely. For example, a threshold can be selected, and an accuracy determined using the reference samples. The threshold (as with other thresholds described herein) can be selected based on a tradeoff of sensitivity and specificity. For example, a threshold can be selected based on a positive predictive value (PPV) for the reference samples while maintaining sensitivity. Accuracy can also be checked by sequencing a tumor and a blood sample (e.g., just plasma / serum or also the buffy coat) to confirm the answers match. Accordingly, the parameter for the read position can be compared to a threshold, which can be determined using normal samples or ones with known mutations at known read positions or other known properties. The threshold may be adjusted to reach certain statistical criteria, e.g., a false positive rate, false negative rate, or accuracy, such as positive predictive value.

[0082] Accordingly, a filtering criterion can be applied using the first set of read positions by determining a parameter of a distribution of the first set of read positions. As examples, the parameter can be a median, mean, mode, or other statistical value of the distribution. The parameter can then be compared to a threshold. Such additional filtering criteria can be performed for any set of read positions and distributions described herein.III. ASCRIBING ORIGIN TO MUTATION (BLOOD OR TUMOR)

[0083] If the mutation is detected in only one of the cell-free sample or the cellular sample, using methodology such as described above, then there is little difficulty in determining whether the mutation is likely to be from a tumor. If the mutation is only found in the cell- free sample, then the mutation can be identified as being from a non-hematopoietic tissue (e.g., a tumor). If the mutation is only found in the buffy coat, then the mutation can be identified having hematopoietic origin.

[0084] However, if a mutation occurs in both the plasma and the huffy coat, then further analysis can be performed to determine the true origin of the mutation. The origin analysis can be whether the origin is non-hematopoietic tissue (e.g., a tumor) or from blood tissue (e.g., white blood cells). If the origin is determined to be from blood cells (not non- hematopoietic tissue), then a more detailed origin analysis can also be performed, e.g., whether the mutation is germline, a mosaicism, or CHIP, which occur at different rates. A CHIP mutation is a low-level mutation that exhibits a lower proportion of the reads having a mutation, then mosaicism, which has a lower proportion (rate) than a germline mutation.

[0085] FIG. 1 is a flowchart illustrating a method 100 for determining an origin of a mutation using measurements in plasma and buffy coat according to embodiments of the present disclosure. Method 100 may be performed by a computer system, as described herein. The measurement in plasma and buffy coat can be sequencing results as described above, including mutations detected as described above.

[0086] Method 100 shows various determinations for the origin of the mutation depending on the measurements in the samples. Confidence intervals for the number of mutated sequence reads (e.g., a normalized amount, such as a proportion also called a variant fraction or variant frequency) in both the plasma and the buffy coat can be used to determine whether the mutation is from non-hematopoietic tissue, e.g., a somatic mutation from a tumor, as is shown. Other determinations can be made based on the specific location of the mutation (e.g., CHIP for certain known mutation in hematopoietic stem cells (HSCs). Other mutations are known mosaicisms or germline mutations. And the variant fraction (number of sequence reads having the mutation relative to all of the reads at the genomic position) can be used to differentiate among CHIP, mosaicisms, and germline mutations.

[0087] Method 100 can be performed for each mutation detected at a genomic position. It is possible that two or more mutations could be detected at a given genomic position. In such a case, method 100 can be performed for each of the mutations at the given genomic position. And method 100 can be performed for each genomic position at which a mutation was detected in both the cell-free sample and the cellular sample.A. Overlapping confidence intervals of plasma and buffy coat

[0088] At block 141, a first confidence interval for the mutation presence in plasma (or other cell-free sample) is compared to the second confidence interval for the mutationpresence in the buffy coat (BC). The confidence intervals can be for a specific confidence, such as 99%, 98%, 97%. 96%, 95%, 94%, 93%, 92%, 91%, 90%, 85%, 80%, 75%, or 70% confidence. In some embodiments, a 95% confidence interval is used (95CI in block 141). The interval is the range for the amount of mutated sequence reads actually in the sample based on the measured number of mutated sequence reads and total sequence reads at the genomic position. The amount can be a normalized amount, such as a frequency or proportion (e.g., a percentage) of the reads that have the mutation, an amount per sample size, or a ratio compared to an amount of reads at one or more other genomic positions.

[0089] The confidence interval depends on how many nucleic acid molecules have the mutation at the genomic position as well as the total number of sequence reads at the genomic position. For example, if there were 20,000 reads and 10,000 of them had the variant, then the confidence interval would probably be around 49-51 for the variant frequency. However, if there were five reads and two of them were alternate, the confidence is probably from 10-90. Some embodiments can use bootstrapping, which randomly samples a certain percentage of the reads and determines the variant frequency for that sampled subset. Such random sampling is performed many times (e.g., 10, 20, 50, 100, 200, 500, or more). Such results are then combined to obtain the confidence interval.

[0090] The comparison of the confidence intervals can include comparisons of the bounds of the confidence intervals. For example, a lower bound of a first confidence interval for how many of a first set of cell-free nucleic acid molecules in the cell-free sample (e.g., plasma) have a mutation at a genomic position can be determined, e.g., using a first number of the first set of sequence reads that includes the respective mutation and a first total number of a first set of sequence reads from the cell-free sample. And an upper bound of a second confidence interval for how many of the second set of nucleic acid fragments have the mutation at the genomic position can be determined, e.g., using a second number of the second set of sequence reads that includes the respective mutation and a second total number of the second set of sequence reads.

[0091] If the confidence intervals overlap (e.g., based on a comparison of the lower bound for the cell-free sample and the upper bound for the cellular sample), then the mutation can be determined as coming from a blood source and not from non-hematopoietic tissue. Whereas, if the confidence interval for the cell-free sample is higher (e.g., lower bound of cell-free sample is greater than upper bound of the cellular sample), then the mutation can beidentified as from non-hematopoietic tissue, such as a tumor as depicted in block 142. The confidence intervals and whether overlapping help to adjudicate whether the mutation is derived from the same source or not.

[0092] For example, assume the variant frequency is 10% in plasma and a 95% confidence interval is between 6-14%. If a mutation has a variant frequency of 50% in the huffy coat, with a 95% confidence interval of 49-51%, then those two confidence intervals are very far away from each other. In such a case, the mutation would be identified as coming from a blood cell origin and not non-hematopoietic tissue. However, if the plasma confidence interval indicates a range of 10-16% and the buffy coat confidence interval is 2-6%, then the mutation can be considered to at least partially be from non-hematopoietic tissue.

[0093] Such a use of confidence intervals works as the plasma has a far lower concentration of nucleic acids. If nucleic acid molecule(s) from a tumor are present in the plasma, such molecules will have a higher concentration (e.g., higher parts per million (ppm)) relative to the other plasma nucleic acids. However, if such a tumor fragment finds its way into the buffy coat, it would be diluted massively by the nucleic acids (e.g., DNA) of the buffy coat. Accordingly, it is unlikely to find nucleic acids from the tumor at the same variant frequencies in the plasma and buffy coat because the buffy coat should greatly dilute the variant frequency.

[0094] Block 142 can provide a distinction between the mutation being from non- hematopoietic tissue (e.g., a tumor) and from other sources.B. Distinction between germline, CHIP, and unknown

[0095] Assuming the lower bound of the confidence interval for the cell-free sample is equal to or greater than the upper bound of the confidence interval for the cellular sample, then method 100 proceeds to block 143. The rest of method 100 can determine CHIP, mosaicism, germline, or indeterminant origin.1. Common CHIP genes

[0096] Block 143 determines whether the mutation is at a common CHIP mutation site. Block 143 lists four genes (ASXL1, JAK2, DNMT3A, TET2) but others can be used. Those are the four of the most common CHIP genes, and so block 144 can indicate the mutation is a CHIP mutation. Such mutations are not clinically-significant for purposes of the cancer andthus can be ignored for treatment purposes. Remember to reach block 143, there had to be overlapping confidence intervals or the buffy coat is higher.2. Variant frequency

[0097] Block 145 uses the variant frequency (VF) in the buffy coat to determine whether the mutation might be germline or a potential mosaicism. If the variant frequency is greater than a first variant threshold (e.g., greater than 25%, 30%, 35%, 40%, or 45%), then it can be determined the mutation is not CHIP. a) Potential mosaicism

[0098] If the variant frequency is greater than the first variant threshold, the method 100 moves to block 146, which can detect a potential mosaicism. Potential mosaicism is determined if 30-40% of that patient’s blood cells harbor that mutation, which is different than CHIP, which is typically observed at a small percentage, e.g., 1-2%.

[0099] Block 146 determines whether the mutation (variant) is on the TP53 gene. High variant frequency often occurs on the TP53 gene in the buffy coat. Thus, a potential mosaicism is called in such a situation at block 147. Confirmation could be performed using a hair follicle test, a fibroblast test, or other secondary test to determine whether the mutation is actually in other cells in their body. It generally would not be a germline mutation because it is not uncommon to observe variant frequency in TP53 that is found in blood but is not germline. The TP53 can have various mutations, e.g., around 15. Any one of these mutations can lead to block 147.

[0100] A potential mosaicism can also be called if block 151 determines the variant frequency is below the first variant threshold but is greater than a second variant threshold, e.g., 20% but could be higher than or lower, depending on value for the first variant threshold, e.g. 15% or 25%. b) Germline reportable gene

[0101] Results obtained from testing patient samples are provided to ordering physicians in the form of a molecular profiling report. Back to block 146, if the mutation is not on the TP53 gene, then method 100 moves to block 148, which can determine whether the variant is a germline reportable gene. Mutations in certain genes are required for reporting due to their clinical significance. The American College of Medical Genetics provides such a list, whichis around 70+ genes and is regularly updated. See, e.g., Green, RC et al., ACMG Recommendations for Reporting of Incidental Findings in Clinical Exome and Genome Sequencing, Genet Med. 2013 Jul; 15(7): 565-574; Miller, DT et al., ACMG STATEMENT ACMG SF v3.1 list for reporting of secondary findings in clinical exome and genome sequencing: A policy statement of the American College of Medical Genetics and Genomics (ACMG), Genet in Med (2022) 24, 1407-1414. In some embodiments, the germline reportable genes include, but are not limited to, one or more of APC, RET, BRCA1, BRCA2, PALB2, SDHD, SDHAF2, SDHC, SDHB, MAX, TMEM127, BMPR1A, SMAD4, TP53, MLH1, MSH2, MSH6, PMS2, MEN1, MUTYH, NF2, STK11, PTEN, RBI, TSC1, TSC2, VHL, WT1, BAP1 ,BRIP1 ,CDH1, CHEK2 ,CDKN2A, FLCN, RAD51C, and RAD51D. Only some of the mutations satisfying the criteria to block 148 might be reportable in the molecular profiling report, as other mutations do not have a known clinical significance. If the mutation is on the list, then block 149 would identify the mutation as a germline mutation, e.g., as the mutation occurs sufficiently often in a population. If the mutation is not on the list, then block 150 would identify the mutation as not reportable. Thus, the mutation would not be reported in the molecular profiling report but may still be identified as a germline mutation. c) CHIP

[0102] At block 151, if the variant frequency is less than the second variant threshold, then the mutation can be identified as a CHIP mutation 153. The mutation is at a sufficiently low frequency in the buffy coat to be identified as a CHIP mutation. Otherwise, the mutation is identified as a potential mosaicism 152, e.g., if the variant frequency is between the first variant threshold and the second variant threshold.IV. DETERMINING SUBJECT PHENOTYPE USING MUTATION(S)

[0103] After the mutational analysis ascribing an origin is performed, the identified mutation(s) can be used for determining a phenotype of the subject, such as classifying a medical condition. As examples, such a classification can include a diagnosis of the presence / absence of a disease or disorder, a stage of a disease or disorder, susceptibility to a disease or disorder, prognosis of a disease stage or disorder, theranosis of a disease or disorder, a physiological state, or response / potential response (or lack thereof) to interventions such as therapeutics. A phenotype can result from a subject’s genetic makeup aswell as the influence of environmental factors and the interactions between the two, as well as from epigenetic modifications to nucleic acid sequences.

[0104] The determination of the phenotype can use mutations that were only identified in the cell-free sample, e.g., plasma or serum, in addition to any mutations for which an origin analysis was required due to detection in the cell-free sample and the cellular sample (e.g., a huffy coat), both of which were present in the original sample (e.g., a blood sample). Mutation(s) identified as being from a non-hematopoietic tissue can be used, e.g., to treat a tumor that is known or assumed to be in the subject as a result of a diagnosis.

[0105] The determination of the phenotype can use any information in a molecular profile as well as any patient specific and / or clinical information. The phenotype can be delivered to a treating physician in a molecular profiling report. A molecular profiling report can include various genetic and epigenetic information (e.g., as determined from laboratory measurements of samples, such as those described herein) as well other characteristics of a subject, such as height, weight, age, blood pressure, medical history, etc.

[0106] Embodiments can determine a phenotype (e.g., a classification of a medical condition) multiple times over a time period (time course), e.g., intervals of 1, 2, 3, 4 weeks;1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, or 12 months; or 1, 2, 3, 4, 5 years. For instance, a liquid biopsy may be performed on a time course following treatment. Such monitoring over a time course can track a medical condition, such as remission, recurrence or disease progression.A. Diagnosis

[0107] Any one or more of the identified mutation(s) may be a biomarker for a particular disease or disorder. The diagnosis may be performed based on the detection of one or more biomarkers. In another embodiment, a diagnosis may include the determination of a quantitative parameter (e.g., an expression level of RNA or a copy number of DNA) that is compared to one or more cutoff values that are selected to differentiate between different phenotype classifications, e.g., disease is present or not.

[0108] As examples, the phenotype can comprise detecting the presence of or likelihood of developing a tumor, neoplasm, or cancer, or characterizing the tumor, neoplasm, or cancer (e.g., stage, grade, aggressiveness, likelihood of metastasis or recurrence, etc). In some embodiments, the cancer comprises an acute myeloid leukemia (AML), breast carcinoma, cholangiocarcinoma, colorectal adenocarcinoma, extrahepatic bile duct adenocarcinoma,female genital tract malignancy, gastric adenocarcinoma, gastroesophageal adenocarcinoma, gastrointestinal stromal tumors (GIST), glioblastoma, head and neck squamous carcinoma, leukemia, liver hepatocellular carcinoma, low grade glioma, lung bronchioloalveolar carcinoma (BAC), lung non-small cell lung cancer (NSCLC), lung small cell cancer (SCLC), lymphoma, male genital tract malignancy, malignant solitary fibrous tumor of the pleura (MSFT), melanoma, multiple myeloma, neuroendocrine tumor, nodal diffuse large B-cell lymphoma, non epithelial ovarian cancer (non-EOC), ovarian surface epithelial carcinoma, pancreatic adenocarcinoma, pituitary carcinomas, oligodendroglioma, prostatic adenocarcinoma, retroperitoneal or peritoneal carcinoma, retroperitoneal or peritoneal sarcoma, small intestinal malignancy, soft tissue tumor, thymic carcinoma, thyroid carcinoma, or uveal melanoma. The systems and methods herein can be used to characterize these and other cancers. Characterizing a phenotype can be providing a diagnosis, prognosis or theranosis of the cancer.B. Treatment selection

[0109] The phenotype classification can include determining one or more treatments based on one or more of the identified mutations, including those identified only in the cell-free sample, those identified as being from non-hematopoietic tissue via the origin analysis, and / or those identified as being from hematopoietic origin via the origin analysis. When the phenotype classification is whether the subject is a likely responder to a particular treatment, that particular treatment may be selected for the patient. In other cases, the determination that a mutation is of blood origin (e.g., from the buffy coat), a treatment can be selected using information in the molecular profile other than the mutation of blood origin.

[0110] For example, a patient with lung cancer might receive immunotherapy (IO therapy), such as a checkpoint inhibitor therapy. In a non-limiting example, consider that a mutation in STK11 is detected in plasma. In this case, one might assume the cancer will resist IO therapy and not provide such treatment. However, if the mutation is determined to have a blood origin (e.g., CHIP), then the patient may benefit from IO treatment.[oni] As another example, a low-level BRCA mutation may be identified as CHIP. In such a case, the molecular profile may suggest likely lack of benefit of PARP inhibitors, which might be indicated if the BRCA mutation is in the tumor (e.g., as identified using the origin analysis). Performing the methods provided herein can avoid treatments based on CHIP-derived false positives.

[0112] The treatment selection can include determining individualized medical intervention for a particular disease state that uses molecular profiling of a patient’s biological specimen. Various findings within the molecular profile along with indicated treatments of likely benefit, likely lack of benefit, or indeterminate benefit may be compiled into a patientspecific molecular profiling report. Such a patient report may include test results for targets and proposed drug therapies based on the test results. Such reports may be compiled using information in one or more databases, which may include, but are not limited to, patient biological sample / specimen information and tracking, clinical data, patient data, patient tracking, file management, study protocols, patient test results from molecular profiling, and billing information and tracking. Other useful databases may include, but are not limited to, drug libraries, gene libraries, disease libraries, and public and private databases such as UniGene, OMIM, GO, TIGR, GenBank, KEGG and Biocarta.

[0113] For a given patient, molecular profiling test results can be compiled to determine the individual characteristics of a cancer. After determining the characteristics of the cancer, a therapeutic regimen may be identified. A patient profile report may be provided which includes the patient’s test results for various targets and any proposed therapies based on those results. The report can comprise a list describing the expected benefit of treatment options based on identified mutations and other measurements (e.g., assessed copy number), thereby identifying candidate treatment options for the subject.

[0114] Accordingly, molecular profiling can be used to identify one or more candidate therapeutic agents for an individual suffering from a condition in which one or more of the biomarkers are targets for treatment. For example, the method can identify one or more chemotherapy treatments for a particular cancer, e.g., having a mutation identified as tumor derived by the origin analysis. A particular mutation need only have some direct or indirect association with a treatment to be useful in guiding treatment decisions. Misattributing a mutation to a tumor when the mutation is actually derived from blood cells may lead to erroneous guidance. Thus, the methods provided herein can be crucial in precision medicine decisions that are based on liquid biopsy.

[0115] In some embodiments, biomarker association rules are used to provide the suggested treatments based on the molecular profiling results. Simple rules can be constructed in the format of “if biomarker positive then treatment option one, else treatment option two.” Other rules may suggest no treatment with a specific drug indicated to havelikely lack of benefit, or treatment with a specific regimen (e.g., immunotherapy and / or chemotherapy) of likely benefit to the patient. In some embodiments, more complex rules are constructed that involve the interaction of multiple biomarkers. The molecular profiling report may describe the association of the predicted benefit of a treatment and the biomarker and optionally a summary statement of the best evidence supporting the treatment guidance. Ultimately, the treating physician will decide on the best course of treatment for a patient.

[0116] When multiple treatment options are revealed by molecular profiling, decision rules can be put in place to prioritize the selection of a treatment regimen. For example, treatments may be prioritized based on direct results of molecular profiling, anticipated efficacy of therapeutic agent, prior history with the same or other treatments, expected side effects, availability of therapeutic agent, cost of therapeutic agent, drug-drug interactions, and other factors. Based on the recommended and prioritized therapeutic agent targets, a treating physician can decide on the course of treatment for a particular individual.

[0117] The methods described herein are used to provide personalized treatment options for cancer patients. In some embodiments, the subject has been previously treated with one or more therapeutic agents to treat the cancer. The cancer may be refractory to one of these agents, e.g., by acquiring drug resistance mutations. Such acquired mutations may be identified over a time course using liquid biopsy and the methods provided herein. In some embodiments, the cancer is metastatic. In some embodiments, the subject has not previously been treated with one or more therapeutic agents identified by the method. Using molecular profiling, candidate treatments can be selected regardless of the stage, anatomical location, or anatomical origin of the cancer cells. Accordingly, molecular profiling methods and systems as described herein can indentify treatments based on individual characteristics of diseased cells, e.g., tumor cells, and other personalized factors in a subject in need of treatment, as opposed to relying on a traditional one-size fits all approach that is conventionally used to treat individuals suffering from a disease, especially cancer. In some cases, the recommended treatments are those not typically used to treat the disease or disorder inflicting the subject. In some cases, the recommended treatments are used after standard-of-care therapies are no longer providing adequate efficacy.

[0118] Additional details are provided in the molecular profiling section below.V. METHOD

[0119] FIG. 2 is a flowchart illustrating a method 200 for determining an origin of a mutation using measurements in a cell-free sample and an associated cellular sample (e.g., buffy coat) according to embodiments of the present disclosure. Some or all of various embodiments of method 200 may be performed or controlled by a computer system.

[0120] At block 210, a first set of cell-free nucleic acid molecules is sequenced from a cell- free sample of a subject to obtain a first set of sequence reads. In various embodiments, the cell-free sample can be plasma or serum.

[0121] At block 215, at each genomic position of one or more genomic positions, it is determined that a respective mutation exists in the first set of sequence reads. As examples, the number of genomic positions can be greater than or less than 1, 2, 3, 5, 10, 20, 50, 100, 200, 500, 1,000, 2,000, 5,000, 10,000, 20,000, 50,000, 100,000, 200,000, or 500,000, or any number therebetween. In some embodiments, the determination includes measuring a first number of the first set of sequence reads that includes the respective mutation and applying filtering criteria using the first number to filter out false positive mutations from the one or more genomic positions. Additionally, first read positions of the respective mutation in a first subset of the first set of sequence reads that include the respective mutation at the genomic position can be determined, thereby determining a first set of read positions. Additional filtering criteria can use the first set of read positions to filter out false positive mutations from the one or more genomic positions. For example, an average position near the end of a cell-free nucleic acid molecule can be compared to a threshold.

[0122] Applying the filtering criteria using the first number can include determining a probability that the genomic position includes the respective mutation based on the first number and an expected number of sequence reads having the respective mutation as a result of one or more errors and comparing the probability to a threshold.

[0123] Applying the filtering criteria using the first set of read positions can include determining a parameter of a distribution of the first set of read positions, wherein the parameter is a median or mean and comparing the parameter to a threshold.

[0124] At block 235, a second set of nucleic acid molecules generated from a buffy coat sample of the subject is sequenced to obtain a second set of sequence reads. A fragmentation process (e.g., mechanical shearing such as sonication or enzymatic processes) can beperformed on the buffy coat sample to generate the second set of nucleic acid molecules. As examples, the first set of cell-free nucleic acid molecules and / or the second set of nucleic acid molecules can comprise deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or a combination thereof, optionally wherein the RNA comprises messenger RNA (mRNA).

[0125] In various embodiments, the cell-free sample and the buffy coat sample are derived from a same blood sample from the subject, or the cell-free sample and the buffy coat sample are derived from different blood samples from the subject. The different blood samples can be taken from the subject at different time points. In various embodiments, the cell-free sample and the buffy coat sample are sequenced in a same sequencing reaction, or the cell- free sample and the buffy coat sample are sequenced in different sequencing reactions.

[0126] As examples, sequencing the first set of cell-free nucleic acid molecules and / or the second set of nucleic acid molecules can comprise amplification, hybridization, microarray, dye termination sequencing, pyrosequencing, next generation sequencing (NGS), whole exome sequencing (WES), whole transcriptome sequencing (WTS), whole genome sequencing (WGS), or any combination thereof. Sequencing the first set of cell-free nucleic acid molecules and / or the second set of nucleic acid molecules can provide sequence reads that include information indicating a mutation, polymorphism, deletion, insertion, substitution, translocation, fusion, break, duplication, amplification, repeat, copy number (copy number variation; CNV; copy number alteration; CNA), transcript level (expression level), or any combination thereof. The sequencing can also be used to determine genomic signatures, including, without limitation, tumor mutational burden (TMB), microsatellite instability (MSI), human leukocyte antigen (HLA) genotype, mismatch repair deficiency, homologous recombination deficiency (HRD), homologous recombination repair (HRR) deficiency, loss of heterozygosity (LOH), or any useful combination thereof.

[0127] At block 245, at each genomic position of the one or more genomic positions, it is determined that the respective mutation exists in the second set of sequence reads. In some embodiments, the determination includes measuring a second number of the second set of sequence reads that includes the respective mutation and applying the filtering criteria using the second number to filter out the false positive mutations from the one or more genomic positions. Additionally, second read positions of the respective mutation in a second subset of the second set of sequence reads that include the respective mutation at the genomic position can be determined, thereby determining a second set of read positions. Additional filteringcriteria using the second set of read positions can be applied to filter out the false positive mutations from the one or more genomic positions.

[0128] At block 255, a lower bound of a first confidence interval is determined for how many of the first set of cell-free nucleic acid molecules have a first mutation at a first genomic position of the one or more genomic positions. The determination can use, among other values, the first number and a first total number of the first set of sequence reads.

[0129] At block 265, an upper bound of a second confidence interval is determined for how many of the second set of nucleic acid molecules have the first mutation at the first genomic position. The determination can use, among other values, the second number and a second total number of the second set of sequence reads.

[0130] As examples, the first confidence interval and the second confidence interval have a confidence in a range of 85 %-98%. For instance, the confidence can be 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, or 99%. In some embodiments, the confidence is set at 95%.

[0131] At block 275, when the lower bound is greater than the upper bound, it is identified that the first mutation at the first genomic position is shed into the cell-free sample from a non-hematopoietic tissue of the subject. As an example, the non-hematopoietic tissue is a tumor.

[0132] The identification of the origin being from non-hematopoietic tissue can be used in various ways. For example, method 200 can further comprise determining a molecular profile, and / or a diagnosis, prognosis, and / or theranosis for a medical condition, based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non-hematopoietic tissue of the subject. As another example, method 200 can determine whether a treatment for a medical condition is of likely benefit, lack of benefit, or indeterminate benefit in treating the subject, based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non-hematopoietic tissue of the subject. The treatment to the subject can be administered based on the determining. Examples for such a determination are described in later sections and elsewhere in this disclosure.

[0133] In some embodiments, e.g., as described in FIG. 1, method 200 can further comprise identifying whether the first mutation is from clonal hematopoiesis of indeterminatepotential (CHIP), germline, or a potential mosaicism, when the first mutation is detected in the huffy coat sample and the lower bound is not greater than the upper bound. The first mutation can be identified as CHIP when the first mutation is in one of a set of specified genes. As an example, the set of specified genes can include ASXL1, JAK2, DNMT3A, and TET2.

[0134] Identifying whether the first mutation is from CHIP, germline, or the potential mosaicism can comprise determining a variant frequency of the first mutation in the buffy coat sample using the second number of the second set of sequence reads that includes the respective mutation and comparing the variant frequency to one or more variant thresholds. Blocks 145 and 151 of method 100 show such examples.

[0135] The first mutation can be determined to be from the potential mosaicism or germline when the variant frequency (VF) is equal to or greater than a first variant threshold of the one or more variant thresholds. As examples, the first variant threshold can be between 25-35%, e.g., 25%, 26%, 27%, 28%, 29%, 30%, 31%, 32%, 33%, 34%, or 35%. In some embodiments, the first variant threshold is 30%. After determining the VF satisfies a threshold criterion, the first mutation can be determined to be from the potential mosaicism when the first mutation is on a first set of more specified genes. The first set of more specified genes can include TP53. The first mutation can be determined to be from germline when the first mutation is not on the first set of more specified genes. The first mutation can be reported to the subject when the first mutation in on a second set of specified genes that are identified as reportable.

[0136] The first mutation can be determined to be from the potential mosaicism when the variant frequency is equal to or greater than a second variant threshold of the one or more variant thresholds and less than the first variant threshold. As examples, the second variant threshold can be between 15-25%, e.g., 15%, 16%, 17%, 18%, 19%, 20%, 21%, 22%, 23%, 24%, or 25%. In some embodiments, the second variant threshold is 20%.

[0137] Other mutations (e.g., a second mutation at a second genomic position) not detected in the buffy coat sample can be identified as shed into the cell-free sample from the non- hematopoietic tissue of the subject.

[0138] The origin of any of one or more mutations can be used in further classifications.For example, a classification of a medical condition of the subject can be determined based on the first mutation at the first genomic position being identified as shed into the cell-freesample from the non-hematopoietic tissue of the subject. An example of the medical condition is a cancer. Various example types of cancer are provided herein. The classification can be performed using a machine learning model, which can be trained using training samples of subjects with a cancer (e.g., a particular type of cancer or cancer generally) and training samples of subjects without cancer (e.g., the particular type or not having any cancer).

[0139] Determining the classification of the medical condition of the subject can include: generating input data that includes a set of features that include a plurality of mutations identified as being shed into the cell-free sample from the non-hematopoietic tissue, the plurality of mutations including the first mutation; and providing the input data into a predictive model. The predictive model can comprise at least one machine learning model. Each machine learning model of the at least one machine learning models can be trained to generate output data that indicates the classification of the medical condition of the subject.

[0140] As examples, the classification can include a diagnosis, prognosis, and / or theranosis for the medical condition based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non-hematopoietic tissue of the subject. As another example, the classification can include whether a treatment for the medical condition is of likely benefit, lack of benefit, or indeterminate benefit in treating the subject, based on the first mutation at the first genomic position being identified as shed into the cell- free sample from the non-hematopoietic tissue of the subject. As a further example, determining the classification of the medical condition can include screening for the medical condition in the subject who is not currently diagnosed as having the medical condition. The classification of the medical condition can be performed over a time course to track a progression of the medical condition.

[0141] Any or all of the mutation(s) and information derived from the mutation(s) (e.g., classification of a medical condition) can be provided in a report. Accordingly, method 200 can provide a report comprising mutations identified according to any one of the preceding claims and / or information derived from the mutations. As examples, the information can comprise a presence or absence, diagnosis, prognosis, and / or theranosis of a cancer in the subject.VI. MOLECULAR PROFILING

[0142] As further described above, the methods of determining the origin of a mutation may play an important role in molecular profiling of a cancer in a subject. As a non-limiting example, consider that a mutation detected via molecular profiling of a biological sample from a cancer patient may indicate that a certain therapy is likely to benefit or not benefit the patient. If the biological sample is a blood sample, it is important to know whether the mutation is derived from the cancer in order to make accurate treatment decisions.

[0143] The molecular profiling approach provides a method for selecting treatments for an individual that could favorably change the clinical course of a medical condition, including without limitation cancer. Molecular profiling can provide a personalized approach to selecting treatments that are more likely to benefit a cancer. The molecular profiling methods described herein can be used to guide treatment in any desired setting, including without limitation the front-line I standard of care setting, or for patients with poor prognosis, such as those with metastatic disease or those whose cancer has progressed on standard front line therapies, or whose cancer has progressed on previous chemotherapeutic or hormonal regimens. The cancer can be a metastatic cancer or other recunent cancer. The treatments can be on-compendium or off-compendium treatments. The treatments may be standard of care for the type of cancer in the individual, or the treatments may be typically used for other types of cancer. Thus, profiling may expand the choice of treatments for the individual.

[0144] The systems and methods provided herein may be used to classify patients as more or less likely to benefit or respond to various treatments. Unless otherwise noted, the terms “response” or “non-response,” as used herein, refer to any appropriate indication that a treatment provides a benefit to a patient (a “responder” or “benefiter”) or has a lack of benefit to the patient (a “non-responder” or “non-benefiter”). Such an indication may be determined using accepted clinical response criteria such as the standard Response Evaluation Criteria in Solid Tumors (RECIST) criteria, or other useful patient response criteria such as progression free survival (PFS), time to progression (TTP), disease free survival (DFS), time-to-next treatment (TNT, TTNT), tumor shrinkage or disappearance, or the like. RECIST is a set of rules published by an international consortium that define when tumors improve (“respond”), stay the same (“stabilize”), or worsen (“progress”) during treatment of a cancer patient. As used herein and unless otherwise noted, a patient “benefit” from a treatment may refer to any appropriate measure of improvement, including without limitation a RECIST response orlonger PFS / TTP / DFS / TNT / TTNT. Beneficial or desired clinical results include, but are not limited to, alleviation or amelioration of one or more symptoms, diminishment of extent of disease, stabilized (i.e., not worsening) state of disease, preventing spread of disease, delay or slowing of disease progression, amelioration or palliation of the disease state, and remission (whether partial or total), whether detectable or undetectable. Benefit also includes prolonging survival as compared to expected survival if not receiving a treatment or if receiving a different treatment. Likewise, “lack of benefit” from a treatment may refer to any appropriate measure of worsening disease during treatment. Generally disease stabilization is considered a benefit, although in certain circumstances, if so noted herein, stabilization may be considered a lack of benefit. A predicted or indicated benefit may be described as “indeterminate” if there is not an acceptable level of prediction of benefit or lack of benefit. In some cases, benefit is considered indeterminate if it cannot be calculated, e.g., due to lack of necessary data.

[0145] Molecular profiling can be performed by any known means for detecting molecules in a biological sample. Useful biological samples include those described above. In preferred embodiments, blood is used as the biological sample for liquid biopsy. Molecular profiling assays can include without limitation, protein and nucleic acid analysis techniques. Protein analysis techniques include, by way of non-limiting examples, immunoassays, immunohistochemistry (IHC), and mass spectrometry. Nucleic acid analysis techniques include, by way of non-limiting examples, amplification such as polymerase chain amplification (PCR) amplification (e.g., qPCR or RT-PCR), hybridization, microarrays, in situ hybridization, and DNA sequencing or RNA sequencing (e.g., dye termination sequencing, Sanger, high throughput or next generation sequencing (NGS), pyrosequencing, and restriction fragment analysis). Other examples of useful assays include in situ hybridization (ISH); fluorescent in situ hybridization (FISH); chromogenic in situ hybridization (CISH); various types of microarray (mRNA expression arrays, low density array s, protein arrays, etc); comparative genomic hybridization (CGH); Northern blot; Southern blot; and any other appropriate technique to assay the presence or quantity of a biological molecule of interest. In various embodiments, any one or more of these methods can be used concurrently or subsequent to each other for assessing target genes disclosed herein.

[0146] As part of determining a molecular profile, sequencing nucleic acid molecules can provide sequence reads that include molecular information indicating point mutations,polymorphisms, deletions, insertions, substitutions, translocations, fusions, breaks, duplications, amplification, repeats, copy numbers (including determining copy number alterations (CNA); also referred to as copy number variation or CNV), transcript levels (expression levels), or any combination thereof Information in sequence reads can also be used to determine genomic signature, including without limitation tumor mutational burden (TMB), microsatellite instability (MSI), human leukocyte antigen (HLA) genotype, mismatch repair deficiency, homologous recombination deficiency (HRD), homologous recombination repair (HRR) deficiency, loss of heterozygosity (LOH), or any combination thereof. Such information can be determined by analyzing the sequence reads. In preferred embodiments, NGS is used to sequence nucleic acid molecules. The high-throughput nature of NGS allows for analysis of the whole exome or whole genome and / or whole transcriptome of >22,000 genes / gene products in a single assay, thereby providing a comprehensive overview of such molecular information.

[0147] The molecular profiling provided by the disclosure is not limited to identifying candidate treatments for patients in need thereof. Indeed, the data derived from molecular profiling can be used to characterize various phenotypes of interest. For example, molecular profiling may be used to screen for disease, monitor disease before and / or after treatment, or characterize a disease aggressiveness. For example, molecular profiling may be used to predict the risk that a primary tumor will metastasize. Thus, molecular profiling of a primary tumor may provide both personalized treatment options for the patient and in addition provide a metastatic potential for the tumor. The treating physician may consider the predicted metastatic potential when deciding a course of treatment for the patient.

[0148] As noted above, a molecular profiling report can be delivered to the treating physician or other caregiver of the subject whose cancer has been profiled. The report can comprise multiple sections of relevant information, including but not limited to: 1) description of the patient and sample; 2) a complete or partial listing of the biomarkers (nucleic acids, proteins, or other biological matter of interest) in the molecular profile; 3) a description of the state of one or more of the biomarkers in the molecular profile as determined for the subject; 4) a description of one or more biological signatures as determined for the molecular profile, such as microsatellite stability, tumor mutational load / burden, tissue-of-origin, recurrence predictors, treatment response predictors; and / or metastasis predictors; 5) one or more treatment associated with one or more of the biomarkers, groups of biomarkers, and / or biological signatures determined for the molecularprofile; 6) an indication whether one or more treatment is likely to benefit the patient, not benefit the patient, or has indeterminate benefit; 7) one or more clinical trials for which the patient may be eligible based on the molecular profile; 7) an indication whether the cancer is predicted to recur and / or metastasize; and / or 8) evidence relevant to the foregoing, such as literature reports and / or clinical trial results.

[0149] The description of the molecular profile within the report can include such information as the laboratory technique used to assess each biomarker, optionally including the result and any criteria used to score each technique. By way of non-limiting example, the criteria for scoring a copy number alteration (or variation, CNA or CNV) may be a presence (i.e., a copy number that is greater or lower than the “normal” copy number present in a subject who does not have cancer, or statistically identified as present in the general population, typically diploid) or absence (i.e., a copy number that is considered the same as the “normal” copy number present in a subject who does not have cancer, or statistically identified as present in the general population, typically diploid). Treatments associated with one or more of the biomarkers or biosignatures may be determined using treatment association such as in any of International Patent Publications WO / 2007 / 137187 (Int’l Appl. No. PCT / US2007 / 069286), published November 29, 2007; WO / 2010 / 045318 (Int’l Appl. No. PCT / US2009 / 060630), published April 22, 2010; WO / 2010 / 093465 (Int’l Appl. No. PCT / US2010 / 000407), published August 19, 2010; WO / 2012 / 170715 (Int’l Appl. No. PCT / US2012 / 041393), published December 13, 2012; WO / 2014 / 089241 (Int’l Appl. No. PCT / US2013 / 073184), published June 12, 2014; WO / 2011 / 056688 (Int’l Appl. No. PCT / US2010 / 054366), published May 12, 2011; WO / 2012 / 092336 (Int’l Appl. No.PCT / US2011 / 067527), published July 5, 2012; WO / 2015 / 116868 (Int’l Appl. No. PCT / US2015 / 013618), published August 6, 2015; WO / 2017 / 053915 (Int’l Appl. No. PCT / US2016 / 053614), published March 30, 2017; WO / 2016 / 141169 (Int’l Appl. No. PCT / US2016 / 020657), published September 9, 2016; and WO2018175501 (Int’l Appl. No. PCT / US2018 / 023438), published September 27, 2018; WO / 2020 / 113237 (based on Int’l Patent Appl. No. PCT / US2019 / 064078, filed December 2, 2019); WO / 2020 / 146554 (based on Int’l Patent Appl. No. PCT / US2020 / 012815, filed January 8, 2020; WO / 2021 / 112918 (based on Int’l Patent Appl. No. PCT / US2020 / 035990, filed June 3, 2020); WO / 2021 / 163706 (based on Int’l Patent Appl. No. PCT / US2021 / 018263, filed February 16, 2021); WO / 2021 / 222867 (based on Int’l Patent Appl. No. PCT / US2021 / 030351, filed April 30, 2021); WO / 2022 / 056328 (based on Int’l Patent Appl. No. PCT / US2021 / 049966, filedSeptember 10, 2021); WO / 2022 / 103809 (based on Int’l Patent Appl. No.PCT / US2021 / 058741, filed November 10, 2021); and WO / 2022 / 132964 (based on Infl Patent Appl. No. PCT / US2021 / 063603, file December 15, 2021); each of which publications is incorporated by reference herein in its entirety.

[0150] Further exemplary associations between genomic and transcriptomic findings and therapy options are shown in Table 1. In the table, the column ‘‘Biomarker” uses gene identifiers commonly accepted in the scientific community at the time of filing and that can be used to look up the genes at various well-known databases such as the HUGO Gene Nomenclature Committee (HNGC; genenames.org), NCBI’s Gene database (ncbi.nlm.nih.gov / gene), GeneCards (genecards.org), Ensembl (ensembl.org), UniProt (uniprot.org), and others. Such identifiers may be used elsewhere in the disclosure. However, the following “Biomarkers” are genomic signatures and / or groups of genes as follows: gLOH: genomic loss of heterozygosity; HLA Genotype: human leukocyte antigen genotype; HRD: homologous recombination deficiency; HRR: homologous recombination repair genes (comprises ATM, BARD1, BRCA1, BRCA2, BRIP1, CDK12, CHEK1, CHEK2, FANCL, PALB2, RAD51B, RAD51C, RAD51D, RAD54L, MLH1, MRE11, FANCA, NBN); MMR Deficiency: mismatch repair deficiency (comprises MLH1, MSH2, MSH6, PMS2); MSI: microsatellite instability; MMR Proficiency: mismatch repair proficiency (comprises MLH1, MSH2, MSH6, PMS2); MSS: microsatellite stable; TMB: tumor mutational burden (also referred to as tumor mutational load (TML)). The column “Alteration” notes the class of biological features that are associated with the various therapies, each of which may be determined using the LP Assay. CNA refers to copy number alteration. The column “Therapy” notes therapies that are linked to the corresponding alteration / s in the biomarker / s. In some cases, certain therapies are preferential for certain cancer types as noted in parenthesis. In anon-limiting example, an RNA fusion detected involving the ALK gene may indicate brigatinib or lorlatinib as of likely benefit for NSCLC. Abbreviations in the Therapy column include: NSCLC: non-small cell lung cancer; CRC: colorectal cancer; DFSP: dermatofibrosarcoma protuberans; CUP: cancer of unknown primary; GIST: gastrointestinal stromal tumor; CNS: central nervous system.Table 1. Biomarkers and therapy associations

[0151] In some embodiments, the genomic and transcriptomic findings are associated with an ongoing clinical trial. As a non-limiting example, sequencing may reveal a mutation that is a requirement for enrollment in a certain trial. The therapy associations may include on-label options, off-label options, clinical trial options, or any desired combination thereof. One of skill will appreciate that on-label refers to use of a therapy in a regulatory approved setting, whereas off-label refers to use of a regulatory approved therapy in an unapproved setting, including without limitation use of a therapy to treat a certain cancer lineage when the therapy is only approved to treat other types of cancer, or use of unapproved dosage forms or regimens.

[0152] In some embodiments, therapy selection is based upon multiple marker signatures, including those that use machine learning and artificial intelligence. See, e.g., International Patent publications WO / 2020 / 113237 (based on Inf 1 Patent Appl. No. PCT / US2019 / 064078, filed December 2, 2019); WO / 2020 / 146554 (based on Int’l Patent Appl. No.PCT / US2020 / 012815, filed January 8, 2020; WO / 2021 / 112918 (based on Infl Patent Appl. No. PCT / US2020 / 035990, filed June 3, 2020); WO / 2021 / 163706 (based on In l Patent Appl. No. PCT / US2021 / 018263, filed Februaiy 16, 2021); WO / 2021 / 222867 (based on Infl Patent Appl. No. PCT / US2021 / 030351, filed April 30, 2021); WO / 2022 / 056328 (based on Infl Patent Appl. No. PCT / US2021 / 049966, filed September 10, 2021); WO / 2022 / 103809 (based on Int’l Patent Appl. No. PCT / US2021 / 058741, filed November 10, 2021); and WO / 2022 / 132964 (based on Infl Patent Appl. No. PCT / US2021 / 063603, file December 15, 2021); each of which publications is incorporated by reference herein in its entirety.

[0153] Treatment associations can be updated as new information becomes available regarding various biomarkers, biosignatures, treatments, and the relationships thereof. The indication whether each treatment is likely to benefit the patient, not benefit the patient, or has indeterminate benefit may be weighted. For example, a likely or potential benefit may be a strong potential benefit or a lesser potential benefit. Such weighting can be based on any appropriate criteria, e.g., the strength of the evidence of the biomarker-treatment association, or the results of the profiling, e.g., a degree or level of over- or underexpression, mutation, or any other relevant state (e.g., wild type or altered). As the treating physician is ultimately responsible for treating their patient, such physician may use the report to assist in guiding their treatment recommendations.

[0154] The molecular profiling report can be delivered to the caregiver for the subject, e.g., the oncologist or other treating physician. The caregiver can use the results of the report to guide a treatment regimen for the subject. For example, the caregiver may administer one or more treatments indicated as likely benefit in the report. Similarly, the caregiver may avoid treating the patient with one or more treatments indicated as likely lack of benefit in the report. In some embodiments, such as when the report includes a biosignature indicating a likely recurrence or metastasis, the treating physician may choose, for example, a more aggressive treatment regimen, more frequent monitoring, or both. Such decisions are made by the caregiver with guidance from the report.

[0155] The report can be computer generated, and can be a printed report, a computer file or both. The report can be made accessible via a secure web portal. The report may be displayed using any desired medium. In some embodiments, the display is a printout, a computer file, including without limitation a pdf file, or may be displayed via an applicationon a computer display such as a computer monitor, laptop display, tablet, smartphone, or other mobile device.

[0156] In an aspect, the disclosure provides a system for generating a molecular profiling report such as described above, comprising: (a) at least one host server; (b) at least one user interface for accessing the at least one host server to access and input data; (c) at least one processor for processing the inputted data; (d) at least one memory coupled to the processor for storing the processed data and instructions for: i) accessing a biomarker status (e.g., a tumor derived mutation) determined by molecular profiling methodology as described herein; and ii) identifying biomarkers, biosignatures and related data and any information derived using such data (treatments, clinical trials, phenotypes, predictions, etc., as described herein); and (e) at least one display for displaying results and outcomes of the molecular profiling. In some embodiments, the system further comprises at least one memory coupled to the processor for storing the processed data and instructions for identifying, based on the generated molecular profile according to the methods above, at least one therapy with potential benefit for treatment of the cancer; and at least one display for display thereof. The system may further comprise at least one database comprising references for various biomarker states, data for drug / biomarker associations, or both. The at least one display can be a report provided by the present disclosure.

[0157] The methods provided herein are useful for identifying mutations in any type of cancer. As examples, the cancer can compnse an acute lymphoblastic leukemia; acute myeloid leukemia; adrenocortical carcinoma; AIDS-related cancer; AIDS-related lymphoma; anal cancer; appendix cancer; astrocytomas; atypical teratoid / rhabdoid tumor; basal cell carcinoma; bladder cancer; brain stem glioma; brain tumor, brain stem glioma, central nervous method atypical teratoid / rhabdoid tumor, central nervous method embryonal tumors, astrocytomas, craniopharyngioma, ependymoblastoma, ependymoma, medulloblastoma, medulloepithelioma, pineal parenchymal tumors of intermediate differentiation, supratentorial primitive neuroectodermal tumors and pineoblastoma; breast cancer; bronchial tumors; Burkitt lymphoma; cancer of unknown primary site (CUP); carcinoid tumor; carcinoma of unknown primary site; central nervous method atypical teratoid / rhabdoid tumor; central nervous method embryonal tumors; cervical cancer; childhood cancers; chordoma; chronic lymphocytic leukemia; chronic myelogenous leukemia; chronic myeloproliferative disorders; colon cancer; colorectal cancer; craniopharyngioma; cutaneous T-cell lymphoma; endocrine pancreas islet cell tumors; endometrial cancer;ependymoblastoma; ependymoma; esophageal cancer; esthesioneuroblastoma; Ewing sarcoma; extracranial germ cell tumor; extragonadal germ cell tumor; extrahepatic bile duct cancer; gallbladder cancer; gastric (stomach) cancer; gastrointestinal carcinoid tumor; gastrointestinal stromal cell tumor; gastrointestinal stromal tumor (GIST); gestational trophoblastic tumor; glioma; hairy cell leukemia; head and neck cancer; heart cancer; Hodgkin lymphoma; hypopharyngeal cancer; intraocular melanoma; islet cell tumors; Kaposi sarcoma; kidney cancer; Langerhans cell histiocytosis; laryngeal cancer; lip cancer; liver cancer; malignant fibrous histiocytoma bone cancer; medulloblastoma; medulloepithelioma; melanoma; Merkel cell carcinoma; Merkel cell skin carcinoma; mesothelioma; metastatic squamous neck cancer with occult primary; mouth cancer; multiple endocrine neoplasia syndromes; multiple myeloma; multiple myeloma / plasma cell neoplasm; mycosis fungoides; myelodysplastic syndromes; myeloproliferative neoplasms; nasal cavity cancer; nasopharyngeal cancer; neuroblastoma; Non-Hodgkin lymphoma; nonmelanoma skin cancer; non-small cell lung cancer; oral cancer; oral cavity cancer; oropharyngeal cancer; osteosarcoma; other bram and spinal cord tumors; ovarian cancer; ovarian epithelial cancer; ovarian germ cell tumor; ovarian low malignant potential tumor; pancreatic cancer; papillomatosis; paranasal sinus cancer; parathyroid cancer; pelvic cancer; penile cancer; pharyngeal cancer; pineal parenchymal tumors of intermediate differentiation; pineoblastoma; pituitary tumor; plasma cell neoplasm / multiple myeloma; pleuropulmonary blastoma; primary central nervous method (CNS) lymphoma; primary hepatocellular liver cancer; prostate cancer; rectal cancer; renal cancer; renal cell (kidney) cancer; renal cell cancer; respiratory tract cancer; retinoblastoma; rhabdomyosarcoma; salivary gland cancer; Sezary syndrome; small cell lung cancer; small intestine cancer; soft tissue sarcoma; squamous cell carcinoma; squamous neck cancer; stomach (gastric) cancer; supratentorial primitive neuroectodermal tumors; T-cell lymphoma; testicular cancer; throat cancer; thymic carcinoma; thymoma; thyroid cancer; transitional cell cancer; transitional cell cancer of the renal pelvis and ureter; trophoblastic tumor; ureter cancer; urethral cancer; uterine cancer; uterine sarcoma; vaginal cancer; vulvar cancer; Waldenstrom macroglobulinemia; or Wilm’s tumor

[0158] As further examples, the cancer can comprise an acute myeloid leukemia (AML), breast carcinoma, cholangiocarcinoma, colorectal adenocarcinoma, extrahepatic bile duct adenocarcinoma, female genital tract malignancy, gastric adenocarcinoma, gastroesophageal adenocarcinoma, gastrointestinal stromal tumor (GIST), glioblastoma, head and necksquamous carcinoma, leukemia, liver hepatocellular carcinoma, low grade glioma, lung bronchioloalveolar carcinoma (BAC), non-small cell lung cancer (NSCLC), lung small cell cancer (SCLC), lymphoma, male genital tract malignancy, malignant solitary fibrous tumor of the pleura (MSFT), melanoma, multiple myeloma, neuroendocrine tumor, nodal diffuse large B-cell lymphoma, non epithelial ovarian cancer (non-EOC), ovarian surface epithelial carcinoma, pancreatic adenocarcinoma, pituitary carcinomas, oligodendroglioma, prostatic adenocarcinoma, retroperitoneal or peritoneal carcinoma, retroperitoneal or peritoneal sarcoma, small intestinal malignancy, soft tissue tumor, thymic carcinoma, thyroid carcinoma, or uveal melanoma.VII. USE OF MACHINE LEARNING

[0159] The thresholds described above and other parameters, such as lower and upper bounds for detecting a variant, and variant thresholds (e.g., at blocks 145 and 151) can be determined using machine learning techniques. Additionally, as described above, classifications of a medical condition can be determined using machine learning models. Accordingly, aspects of the present disclosure can be directed towards a system that generates a set of one or more training data structures that can be used to train a machine learning model to provide various classifications, such as characterizing a phenotype of a biological sample. As described above, characterizing a phenotype can include providing a diagnosis, prognosis, theranosis or other relevant classification. For example, the classification can be predicting a disease state, such as a state or other characteristic of a cancer (e.g., tissue-of- origin, metastatic potential, etc.), or likely benefit or lack of benefit (also referred to as effectiveness) of a treatment for a disease or disorder of a subject based on a molecular profile. The likely benefit or lack of benefit can be a probability that is output by a ML model, where the probability can be a prediction that applies to a population of subjects with similar attributes as the patient being tested.

[0160] Once trained, the trained machine learning model can be used to process input data provided by the system and make predictions based on the processed input data. The input data may include a set of features related to a subject such as data representing one or more subject biomarkers and data representing a disease or disorder or related characteristic. In some embodiments, the input data may include features representing an observable characteristic of a biological sample and make a prediction about the sample, such as the tissue-of-origin of the sample, or, in the case of a cancer sample, whether the subject is likelyto respond to a particular treatment. Additional inputs can include other patient specific or clinical info (e.g., age, sex, medical history, type / location of tumor, etc.).

[0161] The prediction may include data that is output by the machine learning model based on the machine learning model’s processing of a specific set of features provided as an input to the machine learning model. The data may include data representing one or more biomarkers, such as tumor derived mutations as determined according to the methods herein, data representing a disease or disorder, data representing characteristics of a disease or disorder within a particular subject, or data representing a proposed treatment type as desired.

[0162] Aspects of the present disclosure include the extraction of specific data from incoming data streams for use in generating training data structures. The selection of a specific set of one or more biomarkers can have importance for inclusion in the training data structure. This is because the presence, absence or state of particular biomarkers may be indicative of the desired classification. For example, certain biomarkers may be selected to determine whether a treatment for a disease or disorder will be effective or not effective, certain biomarkers may be selected to predict the tissue origin of a biological sample, and / or certain biomarkers may be selected to predict the progression of a disease, including without limitation metastatic potential.

[0163] The system is configured to obtain output data generated by the trained machine learning model based on the machine learning model’s processing of the data. In various embodiments, the data comprises biological data representing one or more biomarkers, data representing a disease or disorder, data representing characteristics of a disease or disorder within a particular subject, and / or data representing a treatment type. The system may make a prediction for a subject having a particular molecular profile, including but not limited to effectiveness of a treatment or other classification of a medical condition.

[0164] Consider an example wherein the system is making a prediction about efficacy of a treatment. The treatment for the subject may include one or more therapeutic agents, e.g., small molecule drugs, biologies, and various combinations thereof. In this setting, output of the trained machine learning model that is generated based on trained machine learning model processing of the input data that includes the molecular profile, the disease or disorder and the treatment type includes data representing the level of responsiveness that the subject will be have to the treatment for the disease or disorder.

[0165] In some embodiments, the output data generated by the trained machine learning model may include a probability of the desired classification. By way of illustration, such probability may be a probability that the subject will favorably respond to the treatment for the disease or disorder. In other embodiments, the output data may include any output data generated by the trained machine learning model based on the trained machine learning model's processing of the input data.

[0166] In some embodiments, the training data structures generated by the present disclosure may include a plurality of training data structures that each include fields representing feature vector corresponding to a particular training sample. The feature vector includes a set of features derived from, and representative of, a training sample. The training sample may include, for example, characteristics of one or more biomarkers in a biological sample from a subject, a disease or disorder of the subject, various characteristics of the disease or disorder of the subject, and / or a proposed treatment for the disease or disorder. The training data structures are flexible because each respective training data structure may be assigned a weight representing each respective feature of the feature vector. Thus, each training data structure of the plurality of training data structures can be particularly configured to cause certain inferences to be made by a machine learning model during training.

[0167] FIG. 3 is a block diagram of an example system 300 for training a machine learning model 110. The machine learning may employ any useful predictive modelling approach. The machine learning model may be, for example, a decision tree, such as a random forest model or gradient boosted tree. In some embodiments, the machine learning model may include a support vector machine, neural network model, a linear regression model, a logistic regression model, a naive Bayes model, a quadratic discriminant analysis model, a K-nearest neighbor model, or the like. The machine learning model training system 300 may be implemented as computer programs on one or more computers in one or more locations, in which the systems, components, and techniques described below can be implemented. The machine learning model training system 300 trains the machine learning model 110 using training data items from a database (or data set) 120 of training data items. The training data items may include a plurality of feature vectors. Each training vector may include a plurality of values that each correspond to a particular feature of a training sample that the training vector represents. The training features may be referred to as independent variables. Inaddition, the system 300 maintains a respective weight for each feature that is included in the feature vectors.

[0168] The machine learning model 110 is configured to receive an input training data item 122 and to process the input training data item 122 to generate an output 118. The input training data item may include a plurality of features (or independent variables “X”) and a training label (or dependent variable ‘ Y”). The machine learning model may be trained using the training items, and once trained, is capable of predicting X = f(Y).

[0169] To enable machine learning model 110 to generate accurate outputs for received data items, the machine learning model training system 300 may train the machine learning model 110 to adjust the values of the parameters of the machine learning model 110, e.g., to determine trained values of the parameters from initial values. These parameters derived from the training steps may include weights that can be used during the prediction stage using the fully trained machine learning model 110.

[0170] In training, the machine learning model 110, the machine learning model training system 300 uses training data items stored in the database (data set) 120 of labeled training data items. The database 120 stores a set of multiple training data items, with each training data item in the set of multiple training items being associated with a respective label. Generally, the label for the training data item identifies a correct classification (or prediction) for the training data item, i.e., the classification that should be identified as the classification of the training data item by the output values generated by the machine learning model 110.A training data item 122 may be associated with a training label 122a.

[0171] The machine learning model training system 300 trains the machine learning model 110 to optimize an objective function. Optimizing an objective function may include, for example, minimizing a loss funebon 130. Generally, the loss function 130 is a function that depends on the (i) output 118 generated by the machine learning model 110 by processing a given training data item 122 and (ii) the label 122a for the training data item 122, i.e., the target output that the machine learning model 110 should have generated by processing the training data item 122.

[0172] Machine learning model training system 300 can train the machine learning model 110 to minimize the (cumulative) loss function 130 by performing multiple iterations of conventional machine learning model training techniques on training data items from the database 120, e.g., hinge loss, stochastic gradient methods, stochastic gradient descent withbackpropagation, or the like, to iteratively adjust the values of the parameters of the machine learning model 110. A fully trained machine learning model 110 may then be deployed as a predicting model that can be used to make predictions based on input data that is not labeled.

[0173] FIG. 4 is a block diagram of a system 400 that generates training data structures for training a machine learning model to predict a classification of a medical condition, e.g., cancer, in a subject. For example, the classification can be the effectiveness of treatment for a disease or disorder of a subject having a particular molecular profile.

[0174] The system 400 includes two or more distributed computers 410, 310, a network 230, and an application server 240. The application server 240 includes an extraction unit 242, a memory unit 244, a vector generation unit 250, and a machine learning model 270. The machine learning model 270 may include one or more of a vector support machine, a neural network model, a linear regression model, a random forest model, a logistic regression model, a naive Bayes model, a quadratic discriminant analysis, model, a K-nearest neighbor model, a support vector machine, or the like. Each distributed computer 410, 310 may include a smartphone, a tablet computer, laptop computer, or a desktop computer, or the like. Alternatively, the distributed computers 410, 310 may include server computers that receive data input by one or more terminals 205, 305, respectively. The terminal computers 205, 305 may include any user device including a smartphone, a tablet computer, a laptop computer, a desktop computer or the like. The network 230 may include one or more networks 230 such as a LAN, a WAN, a wired ethemet network, a wireless network, a cellular network, the internet, or any combination thereof.

[0175] The application server 240 is configured to obtain, or otherwise receive, data records 220, 222, 224, 320 provided by one or more distributed computers such as the first distributed computer 410 and the second distributed computer 310 using the network 230. In some embodiments, each respective distributed computer 410, 310 may provide different types of data records 220, 222, 224, 320. For example, the first distributed computer 410 may provide biomarker data records 220, 222, 224 representing biomarkers for a subject and the second distributed computer 310 may provide patient attributes 320 (e.g., an outcome) representing attributes for a subject obtained from the attributes database 312.

[0176] The biomarker data records 220, 222, 224 may include any type of biomarker data that describes relevant biometric attributes of a subject. By way of example, the biomarker data records can include data records representing DNA 220, protein 222, and RNA data 224.These biomarker data records may each include data structures having fields that include information 220a, 222a, 224a describing biomarkers of a subject such as DNA 220a, protein 222a, or RNA 224a. As a non-limiting example, the DNA data records 220 may include next generation sequencing data such as genomic variants, including without limitation single variants I point mutations, insertions and deletions, substitutions, translocations, fusions, breaks, duplications, amplification, loss, copy numbers, repeats, tumor mutational burden, microsatellite instability, or the like. Next generation sequencing data can be for specific sets of genes or any other desired genomic loci, e.g., whole exome or whole genome data. Alternatively, or in addition, the RNA data records 224 may include data such as gene expression levels or the presence of gene fusions, and may include without limitation data derived from whole transcriptome sequencing. Alternatively, or in addition, the protein data records 222 may include protein expression data (e g., amount and / or localization) such as obtained using immunohistochemistry (IHC).

[0177] The patient attributes data records 320 may describe outcomes of a treatment for a subject. For example, the patient attributes data records 320 obtained from the attributes database 312 may include one or more data structures having fields that structure data attributes of a subject such as a disease or disorder 320a, a treatment 320a the subject received for the disease or disorder, a treatment result 320a, or a combination thereof. In addition, the patient attributes data records 320 may also include fields that structure data attributes describing details of a treatment and a subject’s response to the treatment. An example of a disease or disorder may include, for example, a type of cancer. A type of treatment may include, for example, a type of drug, biologic, or other treatment that the subject has received for the disease or disorder included in the patient attributes data records 320. A treatment result may include data representing a subject’s outcome of a treatment regimen such as beneficial, moderately beneficial, not beneficial, or the like. In some embodiments, the treatment result may include descriptions of a tumor at the end of treatment such as an amount that the tumor was reduced, an overall size of the tumor after treatment, or the like. Alternatively, or in addition, the treatment result may include a number or ratio of white blood cells, red blood cells, or the like. Details of the treatment may include dosage amounts such as an amount of drug taken, a drug regimen, number of missed doses, or the like. Accordingly, though the example of FIG. 4 shows that outcome data may include a disease or disorder, a treatment, and a treatment result, the outcome data may include other types of information, such as described herein. Moreover, there is no requirements that theoutcome data be limited to human “patients.” Instead, the outcome data records 220, 222, 224 and biometric data records 320 (other example of patient attributes) may be associated with any desired subject including any non-human organism.

[0178] In some embodiments, each of the data records 220, 222, 224, 320 may include keyed data that enables the data records from each respective distributed computer to be correlated by application server 240. The keyed data may include, for example, data representing a subject identifier. The subject identifier may include any form of data that identifies a subject and that can associate biomarker for the subject with outcome data for the subject.

[0179] The first distributed computer 410 may provide 208 the biomarker data records 220, 222, 224 to the application server 240. The second distributed computer 310 may provide the patient attributes data records 320 to the application server 240. Different distributed computers may be the same computer. The application server 240 can provide the biomarker data records 220 and the patient attributes data records 220, 222, 224 to the extraction unit 242.

[0180] The extraction unit 242 can process the received biomarker data 220, 222, 224 and patient attributes data records 320 in order to extract data 220a- 1, 222a- 1, 224a- 1, 320a- 1, 320a-2, 320a-3 that can be used to train the machine learning model. For example, the extraction unit 242 can obtain data structured by fields of the data structures of the biometric data records 220, 222, 224, obtain data structured by fields of the data structures of the patient attributes data records 320, or a combination thereof. The extraction unit 242 may perform one or more information extraction algorithms such as keyed data extraction, pattern matching, natural language processing, or the like to identify and obtain data 220a- 1, 222a- 1, 224a-l, 320a-l, 320a-2, 320a-3 from the biometnc data records 220, 222, 224 and patient attributes data records 320, respectively. The extraction unit 242 may provide the extracted data to the memory unit 244. The extracted data unit may be stored in the memory unit 244 such as flash memory (as opposed to a hard disk) to improve data access times and reduce latency in accessing the extracted data to improve system performance. In some embodiments, the extracted data may be stored in the memory unit 244 as an in-memory data grid.

[0181] In more detail, the extraction unit 242 may be configured to filter a portion of the biomarker data records 220, 222, 224 and the patient attributes data records 320 that will beused to generate an input data structure 260 for processing by the machine learning model 270 from the portion of the patient attributes data records 320 that will be used as a label for the generated input data structure 260. Such fdtering includes the extraction unit 242 separating the biomarker data and a first portion of the outcome data that includes a disease or disorder, treatment, treatment details, or a combination thereof, from the treatment result. The application server 240 can then use the biomarker data 220a- 1, 222a- 1, 224a- 1, 320a- 1, 320a-2 and the first portion of the outcome data that includes the disease or disorder 320a- 1, treatment 320a-2, treatment details (not shown in FIG. 4), or a combination thereof, to generate the input data structure 260. In addition, the application server 240 can use the second portion of the outcome data describing the treatment result 320a-3 as the label for the generated data structure.

[0182] The application server 240 may process the extracted data stored in the memory unit 244 correlate the biomarker data 220a- 1, 222a- 1, 224a- 1 extracted from biomarker data records 220, 222, 224 with the first portion of the patient attributes data 320a- 1, 320a-2. The purpose of this correlation is to cluster biomarker data with outcome data so that the outcome data for the subject is clustered with the biomarker data for the subject. In some embodiments, the correlation of the biomarker data and the first portion of the outcome data may be based on keyed data associated with each of the biomarker data records 220, 222, 224 and the patient attributes data records 320. For example, the keyed data may include a subject identifier.

[0183] The application server 240 provides the extracted biomarker data 220a-l, 222a-l, 224a- 1 and the extracted first portion of the patient attributes data 320a- 1, 320a-2 as an input to a vector generation unit 250. The vector generation unit 250 is used to generate a data structure based on the extracted biomarker data 220a- 1, 222a- 1, 224a- 1 and the extracted first portion of the patient attributes data 320a-l, 320a-2. The generated data structure is a feature vector 260 that includes a plurality of values that numerical represents the extracted biomarker data 220a- 1, 222a- 1, 224a-l and the extracted first portion of the patient attributes data 320a-l, 320a-2. The feature vector 260 may include a field for each type of biomarker and each type of outcome data. For example, the feature vector 260 may include one or more fields corresponding to (i) one or more types of DNA sequencing data such as single variants I point mutations, insertions and deletions, substitutions, translocations, fusions, breaks, duplications, amplification, loss, copy numbers, repeats, tumor mutational burden, microsatellite instability, (ii) one or more types of RNA data such as gene expression or genefusion, (lii) one or more types of protein data such as obtained using immunohistochemistry, and (iv) one or more types of patient attributes data such as disease or disorder, treatment type, each type of treatment details, or the like.

[0184] The vector generation unit 250 is configured to assign a weight to each field of the feature vector 260 that indicates an extent to which the extracted biomarker data 220a- 1, 222a- 1, 224a- 1 and the extracted first portion of the patient attributes data 320a- 1, 320a-2 includes the data represented by each field. In some embodiments, for example, the vector generation unit 250 may assign a11’ to each field of the feature vector that corresponds to a feature found in the extracted biomarker data 220a- 1, 222a- 1, 224a- 1 and the extracted first portion of the patient attributes data 320a-l, 320a-2. In such embodiments, the vector generation unit 250 may, for example, also assign a ‘0’ to each field of the feature vector that corresponds to a feature not found in the extracted biomarker data 220a-l, 222a-l, 224a-l and the extracted first portion of the patient attributes data 320a- 1, 320a-2. The output of the vector generation unit 250 may include a data structure such as a feature vector 260 that can be used to train the machine learning model 270.

[0185] The application server 240 can label the training feature vector 260. Specifically, the application server can use the extracted second portion of the patient attributes data 320a- 3 to label the generated feature vector 260 with a treatment result 320a-3. The label of the training feature vector 260 generated based on the treatment result 320a-3 can provide an indication of an effectiveness of the treatment 320a-2 for a disease or disorder 320a- 1 of a subject defined by the specific sets of biomarkers 220a- 1, 222a- 1, 224a- 1, each of which is described by described in the training data structure 260.

[0186] The application server 240 can train the machine learning model 270 by providing the feature vector 260 as an input to the machine learning model 270. The machine learning model 270 may process the generated feature vector 260 and generate an output 272. The application server 240 can use a loss function 280 to determine the amount of error between the output 272 of the machine learning model 270 and the value specified by the training label, which is generated based on the second portion of the extracted patient outcome data describing the treatment result 320a-3. The output 282 of the loss function 280 can be used to adjust the parameters of the machine learning model 270.

[0187] In some embodiments, adjusting the parameters of the machine learning model 270 may include manually tuning of the machine learning model parameters model parameters.Alternatively, in some embodiments, the parameters of the machine learning model 270 may be automatically tuned by one or more algorithms of executed by the application server 240.

[0188] The application server 240 may perform multiple iterations of the process described above for each patient attributes data record 320 stored in the attributes database that correspond to a set of biomarker data for a subject. This may include hundreds of iterations, thousands of iterations, tens of thousands of iterations, hundreds of thousands of iterations, millions of iterations, or more, until each of the patient attributes data records 320 stored in the attributes database 312 and having a corresponding set of biomarker data for a subject are exhausted, until the machine learning model 270 is trained to within a particular margin of error, or a combination thereof. A machine learning model 270 is trained within a particular margin of error when, for example, the machine learning model 270 is able to predict, based upon a set of unlabeled biomarker data, disease or disorder data, and treatment data, an effectiveness of the treatment for the subject having the biomarker data. The effectiveness may include, for example, a probability, an odds ratio, a general indication of the treatment being successful or unsuccessful, or the like.

[0189] FIG. 5 is a block diagram of a system for using a machine learning model that has been trained to predict a classification of a medical condition, e.g., effectiveness of a treatment for a disease or disorder of a subject having a particular molecular profile.

[0190] The machine learning model 370 includes a machine learning model that has been trained using the process described with reference to the system of FIG. 4 above. The trained machine learning model 370 is capable of predicting, based on an input feature vector representative of a set of one or more biomarkers, a disease or disorder, and a treatment, a level of effectiveness for the treatment in treating the disease or disorder for the subject having the biomarkers. In some embodiments, the “treatment’" may include a drug or biologic, treatment details (e.g., dosage, regimen, missed doses, etc.), or any combination thereof.

[0191] The application server 240 hosting the machine learning model 370 is configured to receive unlabeled biomarker data records 320, 322, 324. The biomarker data records 320, 322, 324 include one or more data structures that have fields structuring data that represents one or more biomarkers such as DNA 320a, protein 322a, RNA 324a, or any combination thereof. The received biomarker data records may include characteristics of the biomarkers such as (i) one or more types of DNA data such as single variants, insertions and deletions,substitutions, translocations, fusions, breaks, duplications, amplification, loss, copy numbers, repeats, tumor mutational burden, microsatellite instability, (ii) one or more types of RNA data such as gene expression or gene fusion, (ii) one or more types of protein data such as obtained using immunohistochemistry (IHC), for example, protein expression and / or localization.

[0192] The application server 240 hosting the machine learning model 370 is also configured to receive data representing a proposed treatment data 422a for a disease or disorder described by medical condition data 420a of the subject having biomarkers represented by the received biomarker data records 320, 322, 324. The proposed treatment data 422a for the disease or disorder 422a are also unlabeled and merely a suggestion for treating a subject having biomarkers representing by biomarker data records 320, 322, 324.

[0193] In some embodiments, the medical condition data 420a and the proposed treatment 422a is provided by a terminal 405 over the network 230 and the biomarker data is obtained from a second distributed computer 310. The biomarker data may be derived from laboratory machinery used to perform various assays. In other embodiments, the medical condition data 420a, the proposed treatment 422a, and the biomarker data 320, 322, 324 may each be received from the terminal 405. For example, the terminal 405 may be user device of a treating physician, an employee or agent of the treating physician, a laboratory testing outfit, or other entity that inputs data representing a disease or disorder, data representing a proposed treatment, and a data representing one or more biomarkers for a subject having the disease or disorder. In some embodiments, the treatment data 422 may include data structures structuring fields of data representing a proposed treatment described by a name of a drug, biologic, or combination therapy. In other embodiments, the treatment data 422 may include data structures structuring fields of data representing more complex treatment data such as dosage amounts, a drug regimen, number of allowed missed doses, or the like.

[0194] The application server 240 receives the biomarker data records 320, 322, 324, the medical condition data 420, and the treatment data 422. The application server 240 provides the biomarker data records 320, 322, 324, the medical condition data 420, and the treatment data 422 to an extraction unit 242 that is configured to extract (i) particular biomarker data such as DNA data 320a-l, protein data 322a-l, 324a-l, (ii) medical condition (e g., disease or disorder) data 420a- 1, and (iii) proposed treatment data 422a- 1 from the fields of the biomarker data records 320, 322, 324 and the patient attributes data records 420, 422. Insome embodiments, the extracted data is stored in the memory unit 244 as a buffer, cache or the like, and then provided as an input to the vector generation unit 250 when the vector generation unit 250 has bandwidth to receive an input for processing. In other embodiments, the extracted data is provided directly to a vector generation unit 250 for processing. For example, in some embodiments, multiple vector generation units 250 may be employed to enable parallel processing of inputs to reduce latency.

[0195] The vector generation unit 250 can generate a data structure such as a feature vector 360 that includes a plurality of fields and includes one or more fields for each type of biomarker data and one or more fields for each type of outcome data. For example, each field of the feature vector 360 may correspond to (i) each type of extracted biomarker data that can be extracted from the biomarker data records 320, 322, 324 such as each type of DNA data, each type of RNA data, and each type of protein data, and (ii) each type of outcome data that can be extracted from the patient attributes data records 420, 422 such as each type of disease or disorder, each type of treatment, and each type of treatment details.

[0196] The vector generation unit 250 is configured to assign a weight to each field of the feature vector 360 that indicates an extent to which the extracted biomarker data 320a-l, 322a-l, 324a-l, the extracted condition 420a-l, and the extracted treatment 422a-l includes the data represented by each field. In some embodiments, for example, the vector generation unit 250 may assign a T to each field of the feature vector 360 that corresponds to a feature found in the extracted biomarker data 320a-l, 322a-l, 324a-l, the extracted condition 420a-l, and the extracted treatment 422a-l. In such embodiments, the vector generation unit 250 may, for example, also assign a ‘0’ to each field of the feature vector that corresponds to a feature not found in the extracted biomarker data 320a- 1, 322a- 1, 324a- 1, the extracted condition 420a-l, and the extracted treatment 422a-l. The output of the vector generation unit 250 may include a data structure such as a feature vector 360 that can be provided as an input to the trained machine learning model 370.

[0197] The trained machine learning model 370 processes the generated feature vector 360 based on the adjusted parameters that were determining during the training stage and described with reference to FIG. 4. The output 272 of the trained machine learning model provides an indication of the effectiveness of the treatment 422a- 1 of the condition 420a- 1 for the subject having biomarkers 320a-l, 322a-l, 324a-l. In some embodiments, the output 272 may include a probability that is indicative of the effectiveness of the treatment 422a- 1 of thecondition 420a-l for the subject having biomarkers 320a-l, 322a-l, 324a-l. In such embodiments, the output 272 may be provided 311 to the terminal 405 using the network 230. The terminal 405 may then generate output on a user interface 430 that indicates a predicted level of effectiveness of a treatment of the disease or disorder for a person having the biomarkers represented by the feature vector 360.

[0198] In some embodiments, the output 272 may be provided to a prediction unit 380 that is configured to decipher the meaning of the output 272. For example, the prediction unit 380 can be configured to map the output 272 to one or more categories of effectiveness. Then, the output 382 of the prediction unit 380 can be used as part of message 390 that is provided 311 to the terminal 305 using the network 230 for review by the laboratory' testing outfit, treating physician, or other appropriate party.VIII. COMPUTER SYSTEM

[0199] FIG. 6 illustrates a measurement system 600 according to an embodiment of the present disclosure. The system as shown includes a sample 605, such as cell-free nucleic acid molecules (e.g., DNA and / or RNA) within an assay device 610, where an assay 608 can be performed on sample 605. For example, sample 605 can be contacted with reagents of assay 608 to provide a signal of a physical characteristic 615 (e.g., sequence information of a cell-free nucleic acid molecule). An example of an assay device can be a flow cell that includes probes and / or primers of an assay or a tube through which a droplet moves (with the droplet including the assay). Physical characteristic 615 (e.g., a fluorescence intensity, a voltage, or a current), from the sample is detected by detector 620. Detector 620 can take a measurement at intervals (e.g., periodic intervals) to obtain data points that make up a data signal. In one embodiment, an analog-to-digital converter converts an analog signal from the detector into digital form at a plurality of times.

[0200] Assay device 610 and detector 620 can form an assay system, e.g., a sequencing system that performs sequencing according to embodiments described herein, such as NGSs. A data signal 625 is sent from detector 620 to logic system 630. As an example, data signal 625 can be used to determine sequences and / or locations in a reference genome of nucleic acid molecules (e.g., DNA and / or RNA). Data signal 625 can include various measurements made at a same time, e.g., different colors of fluorescent dyes or different electrical signals for different molecule of sample 605, and thus data signal 625 can correspond to multiple signals. Data signal 625 may be stored in a local memory 635, an external memory 640, or astorage device 645. The assay system can be comprised of multiple assay devices and detectors.

[0201] Logic system 630 may be, or may include, a computer system, ASIC, microprocessor, graphics processing unit (GPU), etc. It may also include or be coupled with a display (e.g., monitor, LED display, etc.) and a user input device (e.g., mouse, keyboard, buttons, etc.). Logic system 630 and the other components may be part of a stand-alone or network connected computer system, or they may be directly attached to or incorporated in a device (e.g., a sequencing device) that includes detector 620 and / or assay device 610. Logic system 630 may also include software that executes in a processor 650. Logic system 630 may include a computer readable medium storing instructions for controlling measurement system 600 to perform any of the methods described herein. For example, logic system 630 can provide commands to a sy stem that includes assay device 610 such that sequencing or other physical operations are performed. Such physical operations can be performed in a particular order, e.g., with reagents being added and removed in a particular order. Such physical operations may be performed by a robotics system, e.g., including a robotic arm, as may be used to obtain a sample and perform an assay. Logic system 630 can perform any steps of methods described herein that perform computer processing.

[0202] Measurement system 600 may also include a treatment device 660, which can provide a treatment to the subject. Treatment device 660 can determine a treatment and / or be used to perform a treatment. Examples of such treatment can include surgery, radiation therapy, chemotherapy, immunotherapy, targeted therapy, hormone therapy, and stem cell transplant. Logic system 630 may be connected to treatment device 660, e.g., to provide results of a method described herein. The treatment device may receive inputs from other devices, such as an imaging device and user inputs (e.g., to control the treatment, such as controls over a robotic system).

[0203] Measurement system 600 may also include a reporting device 655, which can present results of any of the methods describe herein, e.g., as determined using the measurement system. Reporting device 655 can be in communication with a reporting module within logic system 630 that can aggregate, format, and send a report to reporting device 655. The reporting module can present information determined using any of the method described herein. The information can be presented by reporting device 655 in any format that can be recognized and interpreted by a user of the measurement system 600. Forexample, the information can be presented by reporting device 655 in a displayed, printed, or transmitted format, or any combination thereof.

[0204] Any of the computer systems mentioned herein may utilize any suitable number of subsystems. Examples of such subsystems are shown in FIG. 7 in computer system 10. In some embodiments, a computer system includes a single computer apparatus, where the subsystems can be the components of the computer apparatus. In other embodiments, a computer system can include multiple computer apparatuses, each being a subsystem, with internal components. A computer system can include desktop and laptop computers, tablets, mobile phones and other mobile devices.

[0205] The subsystems shown in FIG. 7 are interconnected via a system bus 75. Additional subsystems such as a printer 74, keyboard 78, storage device(s) 79, monitor 76 (e.g., a display screen, such as an LED), which is coupled to display adapter 82, and others are shown. Peripherals and input / output (I / O) devices, which couple to I / O controller 71, can be connected to the computer system by any number of means known in the art such as input / output (I / O) port 77 (e.g., USB, FireWire®). For example, I / O port 77 or external interface 81 (e.g., Ethernet, Wi-Fi, etc.) can be used to connect computer system 10 to a wide area network such as the Internet, a mouse input device, or a scanner. The interconnection via system bus 75 allows the central processor 73 to communicate with each subsystem and to control the execution of a plurality of instructions from system memory 72 or the storage device(s) 79 (e.g., a fixed disk, such as a hard drive, or optical disk), as well as the exchange of information between subsystems. The system memory 72 and / or the storage device(s) 79 may embody a computer readable medium. Another subsystem is a data collection device 85, such as a camera, microphone, accelerometer, and the like. Any of the data mentioned herein can be output from one component to another component and can be output to the user.

[0206] A computer system can include a plurality of the same components or subsystems, e.g., connected together by external interface 81, by an internal interface, or via removable storage devices that can be connected and removed from one component to another component. In some embodiments, computer systems, subsystem, or apparatuses can communicate over a network. In such instances, one computer can be considered a client and another computer a server, where each can be part of a same computer system. A client and a server can each include multiple systems, subsystems, or components. In various embodiments, methods may involve various numbers of clients and / or servers, including atleast 10, 20, 50, 100, 200, 500, 1,000, or 10,000 devices. Methods can include vanous numbers of communication messages between devices, including at least 100, 200, 500, 1,000, 10,000, 50,000, 100,000, 500,00, or one million communication messages. Such communications can involve at least 1 MB, 10 MB, 100 MB, 1 GB, 10 GB, or 100 GB of data.

[0207] Aspects of embodiments can be implemented in the form of control logic using hardware circuitry (e.g., an application specific integrated circuit or field programmable gate array) and / or using computer software stored in a memory with a generally programmable processor in a modular or integrated manner, and thus a processor can include memory storing software instructions that configure hardware circuitry, as well as an FPGA with configuration instructions or an ASIC. As used herein, a processor can include a single-core processor, multi-core processor on a same integrated chip, or multiple processing units on a single circuit board or networked, as well as dedicated hardware. Based on the disclosure and teachings provided herein, a person of ordinary skill in the art will know and appreciate other ways and / or methods to implement embodiments of the present disclosure using hardware and a combination of hardware and software.

[0208] Any of the software components or functions described in this application may be implemented as software code to be executed by a processor using any suitable computer language such as, for example, Java, C, C++, C#, Objective-C, Swift, or scripting language such as R, Perl or Python using, for example, conventional or object-oriented techniques. The software code may be stored as a series of instructions or commands on a computer readable medium for storage and / or transmission. A suitable non-transitory computer readable medium can include random access memory (RAM), a read only memory (ROM), a magnetic medium such as a hard-drive or a floppy disk, or an optical medium such as a compact disk (CD) or DVD (digital versatile disk) or Blu-ray disk, flash memory, and the like. The computer readable medium may be any combination of such devices. In addition, the order of operations may be re-arranged. A process can be terminated when its operations are completed but could have additional steps not included in a figure. A process may correspond to a method, a function, a procedure, a subroutine, a subprogram, etc. When a process corresponds to a function, its termination may correspond to a return of the function to the calling function or the main function.

[0209] Such programs may also be encoded and transmited using carrier signals adapted for transmission via wired, optical, and / or wireless networks conforming to a variety of protocols, including the Internet. As such, a computer readable medium may be created using a data signal encoded with such programs. Computer readable media encoded with the program code may be packaged with a compatible device (e.g., as firmware) or provided separately from other devices (e.g., via Internet download). Any such computer readable medium may reside on or within a single computer product (e.g., a hard drive, a CD, or an entire computer system), and may be present on or within different computer products within a system or network. A computer system may include a monitor, printer, or other suitable display for providing any of the results mentioned herein to a user.

[0210] Any of the methods described herein may be totally or partially performed with a computer system including one or more processors, which can be configured to perform the steps. Any operations performed with a processor (e.g., aligning, determining, comparing, computing, calculating) may be performed in real-time. The term “real-time” may refer to computing operations or processes that are completed within a certain time constraint. The time constraint may be 1 minute, 1 hour, 1 day, or 7 days. Thus, embodiments can be directed to computer systems configured to perform the steps of any of the methods described herein, potentially with different components performing a respective step or a respective group of steps. Although presented as numbered steps, steps of methods herein can be performed at a same time or at different times or in a different order. Additionally, portions of these steps may be used with portions of other steps from other methods. Also, all or portions of a step may be optional. Additionally, any of the steps of any of the methods can be performed with modules, units, circuits, or other means of a system for performing these steps.

[0211] The specific details of particular embodiments may be combined in any suitable manner without departing from the spirit and scope of embodiments of the disclosure. However, other embodiments of the disclosure may be directed to specific embodiments relating to each individual aspect, or specific combinations of these individual aspects.

[0212] The above description of example embodiments of the present disclosure has been presented for the purposes of illustration and description. It is not intended to be exhaustive or to limit the disclosure to the precise form described, and many modifications and variations are possible in light of the teaching above.

[0213] A recitation of "a", "an" or "the" is intended to mean "one or more" unless specifically indicated to the contrary. The use of “or” is intended to mean an “inclusive or,” and not an “exclusive or” unless specifically indicated to the contrary. Reference to a “first” component does not necessarily require that a second component be provided. Moreover, reference to a “first” or a “second” component does not limit the referenced component to a particular location unless expressly stated. The term “based on” is intended to mean “based at least in part on.”

[0214] The claims may be drafted to exclude any element which may be optional. As such, this statement is intended to serve as antecedent basis for use of such exclusive terminology as “solely”, “only”, and the like in connection with the recitation of claim elements, or the use of a “negative” limitation.

[0215] All patents, patent applications, publications, and descriptions mentioned herein are incorporated by reference in their entirety for all purposes. None is admitted to be prior art. Where a conflict exists between the instant application and a reference provided herein, the instant application shall dominate.EXAMPLES

[0216] The invention is further described in the following examples, which do not limit the scope as described herein described in the claims.A. Example 1: Al enabled whole exome / transcriptome liquid biopsy addressing MCEL). MRD, and therapy selection on a single platform

[0217] Effective clinical management of cancer patients requires an accurate and early diagnosis, highly sensitive monitoring of minimal residual disease (MRD), and precise therapy selection. There are multiple tests available that attempt to address each of these needs independently with varying degrees of clinical utility. Caris Assure is a circulating nucleic acid sequencing platform that interrogates the whole transcriptome and whole exome of both plasma and white blood cells to address these needs coupled with the latest advancements in machine learning. The platform is capable of detecting single nucleotide variants (SNVs), insertion / deletions (INDELs), structural variants, copy number, gene expression, tumor mutational burden (TMB), microsatellite instability (MSI), cfDNA fragment length, mutational motifs, and aneuploidy of both somatic (tumor and clonal hematopoiesis) and germline origin. Our extensive database of over 350,000 tissue wholeexomes and trans criptomes from solid malignancies was used to train deep learning neural networks to identify the molecular underpinnings of cancer. These neural networks were then deployed on whole exome\transcriptome data from plasma and huffy coat in pursuit of signals that can inform diagnosis, MRD and therapy selection.

[0218] Validation studies were performed to characterize the analytic and clinical performance of Assure on more than 3000 patient blood samples. These samples include -1000 non-cancer patients (controls), -1700 newly-diagnosed patients across stages where blood was collected at surgery (early detection), -500 early-stage patients during adjuvant therapy at multiple time points (MRD), and -200 locally advanced / metastatic patients where matched tissue testing was also performed (therapy selection).

[0219] For early detection, stratification of stage III\IV cancer samples versus samples with no reported cancer resulted in an AUC > 0.99 highlighting the ability of the Al to diagnose cancer in the blood. In the MRD setting, the Al models were able to differentiate patients whose cancers recurred versus those that did not with a significantly inferior median. This performance was observed across multiple lineages of cancer including breast, colon, lung, bladder, and others. Lastly, for therapy selection, detection of driver mutations where blood was collected within 30 days of matched tissue demonstrated a high concordance with a PPA of 93.8% and PPV 96.8%.

[0220] These data were obtained using the methods provided herein to differentiate tumor mutations from those of hematopoietic origin. CHIP correction proved to be essential as, -35% percent of patients had CHIP mutations, including KRAS, BRAF, ATM, BRCA1 / 2, which could lead to improper therapy selection if such mutations were misattributed to the cancer.B. Example 2: Characterization of plasma cell-free DNA variants as of tumor- or clonal hematopoiesis-origin in 11,914 advanced cancer patients

[0221] This Example illustrates use of the methods provided herein to assign cell-free DNA variants as of tumor- or clonal hematopoiesis-origin in a data set of 11,914 advanced cancer patients.

[0222] Direct sequencing of tumor tissue has been the gold standard when conducting molecular profiling for cancer patients. However, this approach may be limited in some cases, e.g., by the availability of sufficient tissue or the ability of patients to undergoadditional invasive procedures. Plasma-based liquid biopsy tests enable detection of tumorspecific genetic alterations without a tissue biopsy. Plasma-based detection of pathogenic cell-free (cf)DNA variants facilitates access to therapeutic options for patients that have not undergone tissue testing (for whatever reason), unlocking access to therapies for which they would otherwise be ineligible. Studies have demonstrated that the availability of blood-based profiling has increased the overall rate of molecular profiling and thus provided benefits to a greater number of patients [1-3], However, challenges such as non-shedding tumors, tumor heterogeneity, and non-tumor derived mutations can lead to false positive and / or false negative results, leading to a reduced ability for liquid biopsy assays to detect alterations present in the tumor and overall reduced assay sensitivity compared to direct tissue testing [2, 4-8],

[0223] DNA purified from plasma may originate from tumor cells, but also may come from other somatic tissues including the hematopoietic system. Thus, it is important to classify mutations identified using blood-based molecular profiling assays according to the cellular source of origin of each variant. Non-germline variants that are known to be pathogenic and / or targetable by various treatments in tumor tissue are often not contributory to malignancy when present in otherwise healthy cells.

[0224] A common source of non-tumor derived variants is clonal hematopoiesis (CH). CH alterations represent the age-related accumulation of somatic mutations within non-malignant hematopoietic cells and are a known confounder of cfDNA testing [9, 10], Since these variants are derived from hematopoietic cells, they are biologically real, present, and detectable in the plasma, and in typical laboratory validation studies they would be considered true positives when confirmed with an orthogonal plasma-only test. However, CH variants are not part of the tumor or its pathogenic mechanisms and are thus considered false positives with regard to therapeutic decision making. We refer to these mutations as clinical false positives, as they can be erroneously assigned a tumor origin when they in fact originated in non-tumor cells.

[0225] Numerous studies have documented the problem of CH variants leading to false positive mutation detection in liquid biopsies. In a study of 122 advanced non-small cell lung cancer (NSCLC) patients tested with a commercially available plasma-based molecular profiling assay, Hu et al. found up to 53% of variants detected in plasma w'ere likely derived from CH, not tumor tissue

[0011] , In a more recent study, Jensen et al. reviewed cfDNA testingresults from 69 patients with advanced prostate cancer using a large panel NGS assay

[0012] , Plasma cfDNA and a paired whole-blood control sample were simultaneously tested in patients to assess the prevalence of clinically relevant CH interference in prostate cancer cfDNA testing. CH variants were detected in 13 of 69 patients (19%), with seven patients (10% of the total cohort) harboring CH variants in DNA repair genes that are commonly used for poly (ADP-ribose) polymerase inhibitor (PARPi) selection (ATM N = 5, BRCA2 N = 1, CHEK2 N = 1). The remaining six patients had CH variants in genes that are frequently impacted by clonal hematopoiesis, but do not inform therapy selection. This study illustrated that without the abi 1 i ty to identify clinical false positive results traceable to age-related CH variants, 10% of patients would have been misdiagnosed and incorrectly classified as eligible for PARPi based on cfDNA testing results

[0012] ,

[0226] To better understand the prevalence and clinical relevance of CH mutation classification in plasma-based molecular profiling across a large cohort of diverse solid tumors, we analyzed results of plasma-derived cfDNA and buffy coat whole exome sequencing (WES) using a cohort of 11,914 advanced cancer patients sent to our laboratories for tumor profiling. In this Example, we present the abundance and type of CH mutations identified in this cohort. These findings demonstrate the rate of misleading CH mutations across cancer types and can be used to aid in accurate therapy selection while avoiding therapy regimens that are destined to be unsuccessful.C. Methods1. Patient Cohort and Molecular Profiling Assay Workflow

[0227] The patient cohort used in this Example comprised de-identified WES data from cancer patients clinically profiled from 11 / 9 / 2023 date to 8 / 30 / 2024 using our proprietary liquid biopsy assay. The assay prepares nucleic acid libraries and sequences both cfDNA and cfRNA simultaneously in a single sequencing run using a custom hybridization / capture methodology. The cfTNA was extracted from plasma using a proprietary high-throughput automated method, customized from the DSP Virus / Pathogen Midi kit (custom, Qiagen, Germantown, MD) and Hamilton Star liquid handler system. Paired buffy coat samples had genomic DNA (gDNA) extracted using either the automated DSP DNA Midi kit (Cat# 937255, Qiagen) or the Mag-Bind Blood & Tissue DNA HDQ 96 Kit (Cat# M6399-01, Omega Biotek, Norcross, GA). Sequencing libraries were prepared using HyperPrep kits, HyperPure Beads, custom primer mixes and baits, (Roche Sequencing Solutions, Pleasanton,CA) and custom cDNA primers (custom, Integrated DNA Technologies, Inc., Coralville, IA). Sequencing was performed on a NovaSeq System with NovaSeq 6000 SI and S2 Reagent Kits (Illumina, San Diego, CA). Separate sequencing was performed on the nucleic acids extracted from the matched huffy coat to identify potential germline variants and variants arising from clonal hematopoiesis. Buffy coat samples were sequenced at the same batch size as plasma samples so that comparable depth was achieved with comparable limit of detection for both sample types.

[0228] Total FASTQ fdes were divided into DNA FASTQ and RNA FASTQ using the tag on the modified primer in addition to aligning the cf-TNA FASTQ file to a Custom Genomic and Transcriptomic Reference. Reads in the BAM file that were aligned to the Transcript Reference and those with the tag were identified as RNA reads. These RNA reads and their mates were then written into RNA FASTQ files and the remaining reads were written into DNA FASTQ files. These DNA and RNA FASTQ files were then aligned to the hg38 reference sequence using BWA-Mem. See Li H. (2013) Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv: 1303.3997v2 [q-bio.GN],

[0229] The BAM file for the sample was analyzed for the presence of SNV / Indel mutations using Mutect2. See gatk.broadinstitute.org / hc / en-us / articles / 360037593851-Mutect2. We removed variants that were in an exclusion list of frequently detected false-positive variants created during the initial validation phase of the pipeline. Only the pathogenic and likely pathogenic variants were reported. False positives were removed using the following thresholds:

[0230] Minimum Base Quality (MBQ) of alternate (ALT) sequence >= 30

[0231] Minimum Mapping Quality (MMQ) of ALT >= 20

[0232] MBQ of ALT / MBQ of REF >= 0.5

[0233] Allele Frequency (AF) > 0.001

[0234] Median Position (MPOS) of variant in alternate reads >= 9

[0235] Tumor LOD (TLOD) > 4, for lineage relevant variants, 6.3 otherwise.2. Variant Source

[0236] The variants detected in the plasma were characterized by comparing the plasma and buffy coat allele frequency (AF; also referred to as variant frequency or VF herein) and sequencing results as described herein. See, e.g., FIG. 1 and related discussion. As an overview, a variant was characterized as being of tumor origin (block 142) if the lower 95% confidence interval (CI) of the plasma AF is greater than the upper bound of the 95% CI of the buffy coat AF (block 141). The variant was characterized as being of clonal hematopoiesis (CH) origin if the buffy coat AF is less than 20% and the 95% CI overlaps with the plasma AF or if the buffy coat AF was less than 20% and this AF was greater than the plasma AF (blocks 145, 151). It was characterized as germline if the buffy coat AF was 30% or higher and the plasma AF was less than the buffy coat AF (block 145). The variant source was considered unknown if the buffy coat AF was between 20% and 30% to account for potential postzygotic mosaicism resulting in the presence of certain variants only in a subset of cells

[0013] , Variants detected in genes that are known to commonly harbor CH mutations (e.g., ASXL1, JAK2, DNMT3A, TET2, etc.) were characterized as CH if the buffy coat AF was greater than the plasma AF (blocks 143, 144).D. Results

[0237] A total of 11,914 patients were included in the study (median age 68), with 48.0% female (median age 66) and the remaining male (median age 66). A total of 48 different cancer types were included and a breakout of specimens by cancer lineage is included in Table 2.Table 2. Specimens by Cancer Type

[0238] Across all patients, a total of 28,004 reportable tumor or CH-derived mutations were detected in the plasma. Of these, 7,474 (26.7%) variants were classified as CH and not of tumor origin based on their relative buffy coat vs. plasma variant frequencies. 39.5% of patients presented at least one CH mutation among reportable clinical genes. We then determined whether individual gene variants were CH- or tumor-derived. See Table 3, with breakdown on an individual gene basis. For example, we found 79.9% of all BRCA2 variants to be of CH origin, as well as 79.4% of CHEK2, 68.5% of BRCA1, 41.9% of ATM, 6.3% of NRAS, 6.2% of BRAF, 2.4% of KIT, 2.3% of KRAS, and 1.8% of EGFR variants.

[0239] FIG. 10 shows variant classification by gDNA and plasma variant frequency. In the figure, CHIP mutations 1010 are generally clustered along the diagonal between 0.0 and 0.2 along both axes; tumor mutations 1020 are generally clustered along the Plasma Variant Frequency axis, germline mutations 1030 are generally clustered in the center of the plot, and common CH genes 1040 (including ASXL1, IAK2, DNMT3A, and TET2) are sporadically clustered between 0.0 and 0.65 along both axes, overlapping the CHIP 1010 and germline 1030 clusters.Table 3. Variant Origin CH vs Somatic Tumor73 Attorney Docket No. 110588-1473720-0884W01Client Ref No. CMI 884.601

[0240] Next, we analyzed the prevalence of CH- versus tumor-denved variants in cancer types that had greater than 20 patients in our cohort. Prevalence of CH by cancer type ranged from a low of 11.9% in uveal melanoma to a high of 78. 1 % in glioblastoma. Thyroid, soft tissue, ovarian, kidney, salivary gland, prostate and female genital tract cancers all had CH rates greater than 40%.

[0241] Our results are consistent with previous findings that CH prevalence increases with age [14-16], The median rate of CH variant classification was 17% for patients aged 65-69, 29% for patients aged 70-74, 33% for patients aged 75-79, and 50% for patients aged 80+. See FIG. 8.

[0242] Of the pathogenic / likely pathogenic variants detected, 484 were druggable under existing guidelines, meaning the variant indicates a likely benefit to the patient when treated with a recommended monoclonal antibody (MAB), PARP inhibitor, tyrosine kinase inhibitor (TKI), other small molecule or other small molecule inhibitor. However, we found high rates of CH detected in what would be otherwise druggable targets in many cancer types typically treated with PARP inhibitors, including breast, female genital tract, ovarian, pancreatic, prostate, and endometrial cancers. See FIG. 9.75 Attorney Docket No. 110588-1473720-0884W01Client Ref No. CMI 884.601E. Discussion

[0243] Liquid biopsy assays have demonstrated clinical utility for therapy guidance, prognostication, and treatment monitoring for cancer patients [1, 2, 17-21], However, CH is source of false positives in variant detection in cell free samples [11, 22], This Example provides a characterization of CH using the methods provided herein to assign an origin to the variants in a large, real-world cohort of patients encompassing most solid tumor types. We found CH variants in many genes recurrently altered in cancer and determined that a large proportion of these CH variants occurred in genes that are clinically druggable targets. See, e.g., FIG. 9. Several recent studies have reported variable rates of CH mutations in genes that associate with an approved therapy or clinical trial [10, 22-26], Using our methodology to delineate CH, we observed an overall rate of 26.3% CH among pathogenic / likely pathogenic variants, with high rates among clinically important genes including 79.9% of BRCA2, 79.4% of CHEK2, 68.5% of BRCA1, 41.9% of ATM, KRAS / NRAS, BRAF, KIT, and EGFR. See Table 3. We also observed correlation of CH with increasing patient age [FIG. 8; see also references 14-16], Our findings emphasize the confounding nature of CH variants for therapy selection based on liquid biopsy assays, particularly in older patients.

[0244] Prior studies examining the prevalence of CH have mainly focused on small cohorts of individual cancers, precluding any comparative analysis across cancer types [10, 22-24, 26], One pan-tumor analysis of 8,810 patients reported the presence of CH in a high percentage (>25%) of patients with cancers of the thyroid, bladder, pancreas, skin (melanoma), prostate, biliary tract, lung (NSCLC), and endometrium

[0027] , Our real-world analysis corroborates variable rates of CH across solid tumor types, with the highest rates observed in glioblastoma, thyroid cancer, soft tissue tumors, ovarian surface epithelial carcinoma, kidney cancer, salivary gland tumors, prostate adenocarcinoma, female genital tract malignancies, and uterine serous carcinoma.

[0245] While the nature of the relationship between cancer type and CH mutation rate is not well-characterized, it is plausible that differences in therapy regimens or risk factors driving various cancer types could shape the process of CH. For instance, Coombs et al. postulated that the high incidence of CH that they found in thyroid cancer — which we also observed in our cohort — was related to radioactive iodine exposure

[0027] , Furthermore, Bolton et al. demonstrated that smoking drives CH in the ASXL1 gene, whereas exposure to therapywith radiation, platinum, and topoisomerase II inhibitors promotes CH in the DNA damage response genes TP53, PPM1D, and CHEK2. Their study noted that CH was responsive to dose as well as the type of platinum agent, with carboplatin being significantly associated with CH, but not cisplatin or oxaliplatin

[0028] ,

[0246] We found CH to frequently occur in tumor suppressor genes that are used for PARPi selection (e.g., ATM, ATR, BRCA1 / 2, BRIP1, CHEK2, NBN, PALB2), which is of particular concern for patients with breast cancer, genital tract malignancies, ovarian surface epithelial carcinomas, pancreatic adenocarcinoma, prostate cancer, and endometrial cancer. PARPi has approvals by the United States Food and Drug Administration in breast, ovarian, pancreatic, and prostate cancers

[0029] and clinical trials are ongoing to study its use for other tumor types. As reported by Jensen at al. for patients with prostate cancer

[0012] , we found that a large percentage of patients would be inaccurately directed toward PARPi based on liquid biopsy results that are actually variants of CH origin, not tumor. In addition, we noted a small number of CH variants in patients with colorectal cancer, NSCLC, and pancreatic cancer that would be associated with non-PARPi therapies including monoclonal antibodies, small molecule inhibitors, or tyrosine kinase inhibitors. See FIG. 9. Together, these findings point to CH as a key source of false positive variants for biomarker-directed therapy with PARPi, and demonstrate the importance of careful identification of CH across genes and tumor types.

[0247] Liquid biopsy is complementary to tumor-based molecular profiling, ultimately leading to more patients being treated with targeted therapies [1-3], However, treating clinicians should avoid reliance on liquid biopsy results unless the assay filters confounding CH variants.F. References

[0248] 1. Ezeife, D.A., et al., The economic value of liquid biopsy for genomic profiling in advanced non-small cell lung cancer. Ther Adv Med Oncol, 2022. 14: p.17588359221112696.

[0249] 2. Schwartzberg, L.S., et al., Complementary Roles for Tissue- and Blood-Based Comprehensive Genomic Profiling for Detection of Actionable Driver Alterations in Advanced NSCLC. JTO Clin Res Rep, 2022. 3(9): p. 100386.

[0250] 3. Hiemenz, M.C , et al., Real-World Comprehensive Genomic Profiling Success Rates in Tissue and Liquid Prostate Carcinoma Specimens. Oncologist, 2022. 27(12): p. e970-e972.

[0251] 4. Zhang, Y., et al., The correlations of tumor mutational burden among singleregion tissue, multi-region tissues and blood in non-small cell lung cancer. J Immunother Cancer, 2019. 7(1): p. 98.

[0252] 5. Prabhash, K., et al., CONCORDANCE: A real-world evidence study to evaluate the concordance of detecting epidermal growth factor receptor (EGFR) mutation by circulating tumor DNA* versus tissue biopsy in patients with metastatic non-small cell lung cancer. Indian J Cancer, 2022. 59(Supplement): p. S11-S18.

[0253] 6. Procaccio, L., et al., A Real-World Application of Liquid Biopsy in Metastatic Colorectal Cancer: The Poseidon Study. Cancers (Basel), 2021. 13(20).

[0254] 7. Esposito Abate, R., et al., Liquid Biopsy Testing Can Improve Selection of Advanced Non-Small-Cell Lung Cancer Patients to Rechallenge with Gefitinib. Cancers (Basel), 2019. 11(10).

[0255] 8. Adhit, K.K., et al., Liquid Biopsy: An Evolving Paradigm for Non-invasive Disease Diagnosis and Monitoring in Medicine. Cureus, 2023. 15(12): p. e50176.

[0256] 9. Chan, H.T., et al., Clonal Hematopoiesis in Liquid Biopsy: From Biological Noise to Valuable Clinical Implications. Cancers (Basel), 2020. 12(8).

[0257] 10. Chan, H.T., et al., Clinical significance of clonal hematopoiesis in the interpretation of blood liquid biopsy. Mol Oncol, 2020. 14(8): p. 1719-1730.

[0258] 11. Hu, Y., et al., False-Positive Plasma Genotyping Due to Clonal Hematopoiesis. Clin Cancer Res, 2018. 24(18): p. 4437-4443.

[0259] 12. Jensen, K., et al., Association of Clonal Hematopoiesis in DNA Repair Genes With Prostate Cancer Plasma Cell-free DNA Testing Interference. JAMA Oncol, 2021. 7(1): p. 107-110.

[0260] 13. Hallermayr, A., et al., The utility of liquid biopsy in clinical genetic diagnosis of cancer and monogenic mosaic disorders. Med Genet, 2023. 35(4): p. 275-284.

[0261] 14. Genovese, G., et al., Clonal hematopoiesis and blood-cancer risk inferred from blood DNA sequence. N Engl J Med, 2014. 371(26): p. 2477-87.

[0262] 15. Jaiswal, S., et al., Age-related clonal hematopoiesis associated with adverse outcomes. N Engl J Med, 2014. 371(26): p. 2488-98.

[0263] 16. Fabre, M.A., et al., The longitudinal dynamics and natural hi story of clonal haematopoiesis. Nature, 2022. 606(7913): p. 335-342.

[0264] 17. Sonpavde, G., et al., Circulating tumor DNA alterations in patients with metastatic castration-resistant prostate cancer. Cancer, 2019. 125(9): p. 1459-1469.

[0265] 18. Xing, P., et al., Co-mutational assessment of circulating tumour DNA (ctDNA) during osimertinib treatment for T790M mutant lung cancer. J Cell Mol Med, 2019. 23(10): p. 6812-6821.

[0266] 19. Cheng, M.L., et al., Plasma ctDNA Response Is an Early Marker of Treatment Effect in Advanced NSCLC. JCO Precis Oncol, 2021. 5.

[0267] 20. Gandara, D.R., et al., Blood-based tumor mutational burden as a predictor of clinical benefit in non-small-cell lung cancer patients treated with atezolizumab. Nat Med, 2018. 24(9): p. 1441-1448.

[0268] 21. Gouda, M.A., et al., Longitudinal Monitoring of Circulating Tumor DNA to Predict Treatment Outcomes in Advanced Cancers. JCO Precis Oncol, 2022. 6: p. e2100512.

[0269] 22. Razavi, P., et al., High-intensity sequencing reveals the sources of plasma circulating cell-free DNA variants. Nat Med, 2019. 25(12): p. 1928-1937.

[0270] 23. Ococks, E., et al., Longitudinal tracking of 97 esophageal adenocarcinomas using liquid biopsy sampling. Annals of Oncology, 2021. 32(4): p. 522-532.

[0271] 24. Huang, F., et al., Chemotherapy -associated clonal hematopoiesis mutations should be taken seriously in plasma cell-free DNA KRAS / NRAS / BRAF genotyping for metastatic colorectal cancer. Clin Biochem, 2021. 92: p. 46-53.

[0272] 25. van 't Erve, I , et al., Metastatic Colorectal Cancer Treatment Response Evaluation by Ultra-Deep Sequencing of Cell-Free DNA and Matched White Blood Cells. Clin Cancer Res, 2023. 29(5): p. 899-909.

[0273] 26. Leal, A., et al., White blood cell and cell-free DNA analyses for detection of residual disease in gastric cancer. Nat Commun, 2020. 11(1): p. 525.

[0274] 27. Coombs, C.C., et al., Therapy-Related Clonal Hematopoiesis in Patients with Non-hematologic Cancers Is Common and Associated with Adverse Clinical Outcomes. Cell Stem Cell, 2017. 21(3): p. 374-382.e4.

[0275] 28. Bolton, K.L., et al., Cancer therapy shapes the fitness landscape of clonal hematopoiesis. Nat Genet, 2020. 52(11): p. 1219-1226.

[0276] 29. Das, P.K., et al., Poly (ADP -ribose) polymerase (PARP) inhibitors as anticancer agents: An outlook on clinical progress, synthetic strategies, biological activity, and structure-activity relationship. Eur J Med Chem, 2024. 274: p. 116535.

Claims

WHAT IS CLAIMED IS:

1. A method comprising: sequencing a first set of cell-free nucleic acid molecules from a cell-free sample of a subject to obtain a first set of sequence reads, wherein the cell-free sample is plasma or serum; at each genomic position of one or more genomic positions, determining that a respective mutation exists in the first set of sequence reads, including: measuring a first number of the first set of sequence reads that includes the respective mutation; and applying filtering criteria using the first number; sequencing a second set of nucleic acid molecules generated from a buffy coat sample of the subject to obtain a second set of sequence reads; at each genomic position of the one or more genomic positions, determining that the respective mutation exists in the second set of sequence reads, including: measuring a second number of the second set of sequence reads that includes the respective mutation; and applying the filtering criteria using the second number; at a first genomic position of the one or more genomic positions: determining, using the first number and a first total number of the first set of sequence reads, a lower bound of a first confidence interval for how many of the first set of cell-free nucleic acid molecules have a first mutation at the first genomic position; and determining, using the second number and a second total number of the second set of sequence reads, an upper bound of a second confidence interval for how many of the second set of nucleic acid molecules have the first mutation at the first genomic position; and identifying that the first mutation at the first genomic position is shed into the cell -free sample from a non-hematopoietic tissue of the subject when the lower bound is greater than the upper bound.

2. The method of claim 1, wherein determining that the respective mutation exists in the first set of sequence reads further includes:determining first read positions of the respective mutation in a first subset of the first set of sequence reads that include the respective mutation at the genomic position, thereby determining a first set of read positions; and applying additional filtering criteria using the first set of read positions to filter out false positive mutations from the one or more genomic positions.

3. The method of claim 1 or 2, wherein determining that the respective mutation exists in the second set of sequence reads further includes: determining second read positions of the respective mutation in a second subset of the second set of sequence reads that include the respective mutation at the genomic position, thereby determining a second set of read positions; and applying the additional filtering criteria using the second set of read positions to filter out the false positive mutations from the one or more genomic positions.

4. The method of claim 2 or 3, wherein applying the additional filtering criteria using the first set of read positions includes: determining a first parameter of a first distribution of the first set of read positions; and comparing the first parameter to a first threshold.

5. The method of any preceding claim, wherein applying the filtering criteria using the first number includes: determining a first probability that the genomic position includes the respective mutation based on the first number and a first expected number of sequence reads having the respective mutation as a result of one or more errors; and comparing the first probability to a second threshold.

6. The method of any one of claims 3-5, wherein applying the filtering criteria using the second set of read positions includes: determining a second parameter of a second distribution of the second set of read positions; and comparing the second parameter to a third threshold.

7. The method of any preceding claim, wherein applying the filtering criteria using the second number includes:determining a second probability that the genomic position includes the respective mutation based on the second number and a second expected number of sequence reads having the respective mutation as a result of one or more errors; and comparing the second probability to a threshold.

8. The method of any preceding claim, wherein measuring the first number of the first set of sequence reads that includes the respective mutation includes aligning the first set of sequence reads to a reference genome.

9. The method of claim 8, wherein measuring the second number of the second set of sequence reads that includes the respective mutation includes aligning the first set of sequence reads to the reference genome.

10. The method of any preceding claim, wherein the non-hematopoietic tissue is a tumor.

11. The method of any preceding claim, further comprising: performing a fragmentation process on the buffy coat sample to generate the second set of nucleic acid molecules.

12. The method of any one of claims 1-11, wherein the cell-free sample and the buffy coat sample are derived from a same blood sample from the subject.

13. The method of any one of claims 1-11, wherein the cell-free sample and the buffy coat sample are derived from different blood samples from the subject, optionally wherein the different blood samples are taken from the subject at different time points.

14. The method of any one of claims 1-13, wherein the cell-free sample and the buffy coat sample are sequenced in a same sequencing reaction.

15. The method of any one of claims 1-13, wherein the cell-free sample and the buffy coat sample are sequenced in different sequencing reactions.

16. The method of any preceding claim, wherein the first confidence interval and the second confidence interval have a confidence in a range of about 85%-98%.

17. The method of claim 16, wherein the confidence is about 95% or is 95%.

18. The method of any one of claims 1-17, further comprising identifying whether the first mutation is from clonal hematopoiesis of indeterminate potential (CHIP), germline, or a potential mosaicism, when the first mutation is detected in the huffy coat sample and the lower bound is not greater than the upper bound.

19. The method of claim 18, wherein identifying whether the first mutation is from CHIP includes: identifying the first mutation as CHIP when the first mutation is on one of a set of specified genes.

20. The method of claim 19, wherein the set of specified genes includes one or more of ASXL1, JAK2, DNMT3A, and TET2.

21. The method of claim 18, wherein identifying whether the first mutation is from CHIP, germline, or the potential mosaicism comprises: determining a variant frequency of the first mutation in the huffy coat sample using the second number of the second set of sequence reads that includes the respective mutation; and comparing the variant frequency to one or more variant thresholds.

22. The method of claim 21, wherein the first mutation is determined to be from the potential mosaicism or germline when the variant frequency is equal to or greater than a first variant threshold of the one or more variant thresholds.

23. The method of claim 22, wherein the first mutation is determined to be from the potential mosaicism when the first mutation is on a first set of more specified genes.

24. The method of claim 23, wherein the first set of more specified genes includes TP53.

25. The method of claim 23, wherein the first mutation is determined to be from germline when the first mutation is not on the first set of more specified genes.

26. The method of claim 25, further comprising:reporting the first mutation to the subject when the first mutation in on a second set of specified genes that are identified as reportable.

27. The method of claim 26, wherein the second set of specified genes that are identified as reportable comprises one or more of APC, RET, BRCA1, BRCA2, PALB2, SDHD, SDHAF2, SDHC, SDHB, MAX, TMEM127, BMPR1A, SMAD4, TP53, MLH1, MSH2, MSH6, PMS2, MEN1, MUTYH, NF2, STK11, PTEN, RBI, TSC1, TSC2, VHL, WT1, BAP1 ,BRIP1 ,CDH1, CHEK2 ,CDKN2A, FLCN, RAD51C, and RAD51D.

28. The method of claim 22, wherein the first variant threshold is between about 25-35%, optionally wherein the first variant threshold is about 30% or is 30%.

29. The method of claim 22, wherein the first mutation is determined to be from the potential mosaicism when the variant frequency is equal to or greater than a second variant threshold of the one or more variant thresholds and less than the first variant threshold.

30. The method of claim 29, wherein the second variant threshold is between about 15-25%, optionally wherein the second variant threshold is about 20% or is 20%.

31. The method any preceding claim, further comprising: identifying that a second mutation at a second genomic position is shed into the cell-free sample from the non-hematopoietic tissue of the subject when the second mutation is not detected in the buffy coat sample.

32. The method of any preceding claim, wherein the first set of cell-free nucleic acid molecules and / or the second set of nucleic acid molecules comprise deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or a combination thereof, optionally wherein the RNA comprises messenger RNA (mRNA).

33. The method of any preceding claim, wherein sequencing the first set of cell-free nucleic acid molecules and / or the second set of nucleic acid molecules comprises amplification, hybridization, microarray, dye termination sequencing, pyrosequencing, next generation sequencing (NGS), whole exome sequencing (WES), whole transcriptome sequencing (WTS), whole genome sequencing (WGS), or any combination thereof.

34. The method of any preceding claim, wherein sequencing the first set of cell-free nucleic acid molecules and / or the second set of nucleic acid molecules provide sequence reads that include information used to indicate or detect a mutation, polymorphism, deletion, insertion, substitution, translocation, fusion, break, duplication, amplification, repeat, copy number, copy number alteration (CNA; copy number variation; CNV), transcript level (expression level), tumor mutational burden (TMB), microsatellite instability (MSI), human leukocyte antigen (HLA) genotype, mismatch repair deficiency, homologous recombination deficiency (HRD), homologous recombination repair (HRR) deficiency, loss of heterozygosity (LOH), or any combination thereof.

35. The method of any preceding claim, further comprising determining a molecular profile, and / or a diagnosis, prognosis, and / or theranosis for a medical condition, based on the first mutation at the first genomic position being identified as shed into the cell- free sample from the non-hematopoietic tissue of the subject.

36. The method of any preceding claim, further comprising determining whether a treatment for a medical condition is of likely benefit, lack of benefit, or indeterminate benefit in treating the subject, based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non-hematopoietic tissue of the subject.

37. The method of claim 36, further comprising administering the treatment to the subject based on the determining.

38. The method of any one of claims 1-35, further comprising: determining a classification of a medical condition of the subject based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non-hematopoietic tissue of the subject.

39. The method of claim 38, wherein determining the classification of the medical condition of the subject includes: generating, by one or more computers, input data that includes a set of features that include a plurality of mutations identified as being shed into the cell-free sample from the non-hematopoietic tissue, the plurality of mutations including the first mutation; and providing, by the one or more computers, the input data into a predictive model, the predictive model comprising at least one machine learning model, wherein eachmachine learning model of the at least one machine learning models is trained to generate output data that indicates the classification of the medical condition of the subject.

40. The method of claim 38, wherein the classification includes a diagnosis, prognosis, and / or theranosis for the medical condition based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non- hematopoietic tissue of the subject.

41. The method of claim 38, wherein the classification includes whether a treatment for the medical condition is of likely benefit, lack of benefit, or indeterminate benefit in treating the subject, based on the first mutation at the first genomic position being identified as shed into the cell-free sample from the non-hematopoietic tissue of the subject.

42. The method of claim 38, wherein determining the classification of the medical condition includes screening for the medical condition in the subject who is not currently diagnosed as having the medical condition.

43. The method of claim 38, wherein determining the classification of the medical condition includes performing the method over a time course to track a progression of the medical condition.

44. The method of any one of claim 38-43, wherein the medical condition comprises a cancer.

45. The method of claim 44, wherein the cancer comprises an acute lymphoblastic leukemia; acute myeloid leukemia; adrenocortical carcinoma; AIDS-related cancer; AIDS-related lymphoma; anal cancer; appendix cancer; astrocytomas; atypical teratoid / rhabdoid tumor; basal cell carcinoma; bladder cancer; brain stem glioma; brain tumor, brain stem glioma, central nervous method at pical teratoid / rhabdoid tumor, central nervous method embryonal tumors, astrocytomas, craniopharyngioma, ependymoblastoma, ependymoma, medulloblastoma, medulloepithelioma, pineal parenchymal tumors of intermediate differentiation, supratentorial primitive neuroectodermal tumors and pineoblastoma; breast cancer; bronchial tumors; Burkitt lymphoma; cancer of unknown primary site (CUP); carcinoid tumor; carcinoma of unknown primary site; central nervous method atypical teratoid / rhabdoid tumor; central nervous method embryonal tumors; cervical cancer; childhood cancers; chordoma; chronic lymphocytic leukemia; chronic myelogenousleukemia; chronic myeloproliferative disorders; colon cancer; colorectal cancer; craniopharyngioma; cutaneous T-cell lymphoma; endocrine pancreas islet cell tumors; endometrial cancer; ependymoblastoma; ependymoma; esophageal cancer; esthesioneuroblastoma; Ewing sarcoma; extracranial germ cell tumor; extragonadal germ cell tumor; extrahepatic bile duct cancer; gallbladder cancer; gastric (stomach) cancer; gastrointestinal carcinoid tumor; gastrointestinal stromal cell tumor; gastrointestinal stromal tumor (GIST); gestational trophoblastic tumor; glioma; hairy cell leukemia; head and neck cancer; heart cancer; Hodgkin lymphoma; hypopharyngeal cancer; intraocular melanoma; islet cell tumors; Kaposi sarcoma; kidney cancer; Langerhans cell histiocytosis; laryngeal cancer; lip cancer; liver cancer; malignant fibrous histiocytoma bone cancer; medulloblastoma; medulloepithehoma; melanoma; Merkel cell carcinoma; Merkel cell skin carcinoma; mesothelioma; metastatic squamous neck cancer with occult primary; mouth cancer; multiple endocrine neoplasia syndromes; multiple myeloma; multiple myeloma / plasma cell neoplasm; mycosis fungoides; myelodysplastic syndromes; myeloproliferative neoplasms; nasal cavity cancer; nasopharyngeal cancer; neuroblastoma; Non-Hodgkin lymphoma; nonmelanoma skin cancer; non-small cell lung cancer; oral cancer; oral cavity cancer; oropharyngeal cancer; osteosarcoma; other brain and spinal cord tumors; ovarian cancer; ovarian epithelial cancer; ovarian germ cell tumor; ovarian low malignant potential tumor; pancreatic cancer; papillomatosis; paranasal sinus cancer; parathyroid cancer; pelvic cancer; penile cancer; pharyngeal cancer; pineal parenchymal tumors of intermediate differentiation; pineoblastoma; pituitary tumor; plasma cell neoplasm / multiple myeloma; pleuropulmonary blastoma; primary central nervous method (CNS) lymphoma; primary hepatocellular liver cancer; prostate cancer; rectal cancer; renal cancer; renal cell (kidney) cancer; renal cell cancer; respiratory tract cancer; retinoblastoma; rhabdomyosarcoma; salivary gland cancer; Sezary syndrome; small cell lung cancer; small intestine cancer; soft tissue sarcoma; squamous cell carcinoma; squamous neck cancer; stomach (gastric) cancer; supratentorial primitive neuroectodermal tumors; T-cell lymphoma; testicular cancer; throat cancer; thymic carcinoma; thymoma; thyroid cancer; transitional cell cancer; transitional cell cancer of the renal pelvis and ureter; trophoblastic tumor; ureter cancer; urethral cancer; uterine cancer; uterine sarcoma; vaginal cancer; vulvar cancer; Waldenstrom macroglobulinemia; or Wilm’s tumor.

46. The method of claim 44, wherein the cancer comprises an acute myeloid leukemia (AML), breast carcinoma, cholangiocarcinoma, colorectaladenocarcinoma, extrahepatic bile duct adenocarcinoma, female genital tract malignancy, gastric adenocarcinoma, gastroesophageal adenocarcinoma, gastrointestinal stromal tumor (GIST), glioblastoma, head and neck squamous carcinoma, leukemia, liver hepatocellular carcinoma, low grade glioma, lung bronchioloalveolar carcinoma (BAC), non-small cell lung cancer (NSCLC), lung small cell cancer (SCLC), lymphoma, male genital tract malignancy, malignant solitary fibrous tumor of the pleura (MSFT), melanoma, multiple myeloma, neuroendocrine tumor, nodal diffuse large B-cell lymphoma, non epithelial ovarian cancer (non-EOC), ovarian surface epithelial carcinoma, pancreatic adenocarcinoma, pituitary carcinomas, oligodendroglioma, prostatic adenocarcinoma, retroperitoneal or peritoneal carcinoma, retroperitoneal or peritoneal sarcoma, small intestinal malignancy, soft tissue tumor, thymic carcinoma, thyroid carcinoma, or uveal melanoma.

47. The method of any preceding claim, further comprising: providing a report comprising mutations identified according to any one of the preceding claims and / or information derived from the mutations, optionally wherein the information comprises a presence or absence, diagnosis, prognosis, and / or theranosis of a cancer in the subject.

48. A computer product comprising a non-transitory computer readable medium storing a plurality of instructions that, when executed, cause a computer system to perform the method of any one of the preceding claims.

49. A system comprising: the computer product of claim 48; and one or more processors configured to execute instructions stored on the computer readable medium.

50. A system comprising means for performing any of the above methods.

51. A system comprising one or more processors configured to perform any of the above methods.

52. A system comprising modules that respectively perform the steps of any of the above methods.