Methods and systems for autoimmune treatment selection using cell-free nucleic acids

Abstract

Methods, compositions, and systems for identifying autoimmune disease subject response to an immunosuppressant drug are described herein.

Description

WSGR Docket No. 55066-711.601METHODS AND SYSTEMS FOR AUTOIMMUNE TREATMENT SELECTION USING CELL-FREE NUCLEIC ACIDSCROSS-REFERENCE

[0001] This application claims the benefit of U.S. Provisional Application Nos. 63 / 719,276, filed on November 12, 2024 and 63 / 719,286 filed on November 12, 2024, each of which are incorporated by reference herein in their entireties.BACKGROUND

[0002] An autoimmune disease is a condition that results from an anomalous response of the immune system, wherein it mistakenly targets and attacks healthy, functioning parts of the body as if they were foreign organisms. Approximately 24 million people, or roughly 7.5% of the population, in the United States are affected by an autoimmune disease. The exact causes of autoimmune diseases remain unclear and are likely multifactorial, involving both genetic and environmental influences. Autoimmune diseases represent a vast and diverse category of disorders that, despite their differences, share some common symptomatic threads. These shared symptoms occur as a result of the body's immune system mistakenly attacking its own cells and tissues, causing inflammation and damage. Diagnosing autoimmune diseases can be challenging due to their diverse presentations and the transient nature of many symptoms. Treatment of autoimmune diseases can be challenging due to variable patient responses to immunosuppressants used to treat autoimmune diseases. In rheumatoid arthritis (RA), as many as 12-18% of patients initiating a disease-modifying antirheumatic drug (DMARD) for treatment will switch their therapy within 1 year of initiation.SUMMARY

[0003] In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject identified as having an autoimmune condition and optionally having been administered an immunosuppressant drug, the method comprising: (a) generating data by detecting a plurality of cell-free nucleic acid (cfNA) fragments derived from a sample obtained from the subject, wherein cfNA fragments of the plurality of the cfNA fragments are indicative of a presence or absence of expression of a plurality of genomic regions associated with response or non-response to the immunosuppressant drug; and (b) applying a machine learning model to the data, wherein the machine learning model has been trained with training data that comprises: (i) data indicative of the presence or absence of the expression of the plurality of genomic regions in a plurality subjects having the autoimmune condition; and (ii) data indicativeWSGR Docket No. 55066-711.601of response or non-response of each subject of the plurality of subjects to the immunosuppressant drug. In some embodiments, the cfNA fragments are cell-free deoxyribonucleic acid (cfDNA) fragments. In some embodiments, the immunosuppressant drug is a disease-modifying antirheumatic drug (DMARD). In some embodiments, the subject is undergoing treatment or has undergone treatment with another immunosuppressant drug different from the immunosuppressant drug before the detecting in (a). In some embodiments, the DMARD is a Janus kinase (JAK) inhibitor. In some embodiments, the DMARD is a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD is a TNFa inhibitor or a JAK inhibitor, further comprising classifying the subject as responsive to the DMARD. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug with at least 80% accuracy. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug with at least 85% specificity and at least 80% sensitivity. In some embodiments, the autoimmune condition is Rheumatoid Arthritis. In some embodiments, the response to the immunosuppressant drug is a decrease in Clinical Disease Index (CD Al) of at least about 10 or a decrease in at least one CD Al category. In some embodiments, sample obtained from the subject is a blood sample. In some embodiments, the plurality of cfNA fragments comprise a predominance of cfNA fragments derived from active chromatin. In some embodiments, the method further comprises prior to (a): (i) contacting cfNA fragments, the cfNA fragments comprising the plurality of cfNA fragments and derived from the sample obtained from the subject, to an anionic surface to selectively enrich for the plurality of cfNA fragments, or (ii) performing a size selection on cfNA fragments, comprising the plurality of cfNA fragments and derived from the sample obtained from the subject, to enrich for the plurality of cfNA fragments. In some embodiments, the plurality of cfNA fragments comprises fragments of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length, wherein the cfNA fragments are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments, the genomic regions comprise transcription factor binding sites, exons, gene body exons, promoters, insulators, DNAse-I hypersensitive sites, or RNA polymerase II pausing sites, or any combination thereof. In some embodiments, the method further comprises, after (b), using the machine learning model to classify the subject as responsive to the immunosuppressant drug. In some embodiments, the method further comprises administering the immunosuppressant drug to the subject. In some embodiments, the method further comprises reporting an identification of the subject as responsive to the immunosuppressant drug to a physician, a caregiver, or the subject. In some embodiments, the detecting the plurality of cfNA fragments comprisesWSGR Docket No. 55066-711.601performing next-generation sequencing. In some embodiments, detecting the plurality of cfNA fragments comprises performing a quantitative reverse polymerase chain reaction (qPCR), a reverse transcriptase polymerase chain reaction (rtPCR), a digital droplet polymerase chain reaction (ddPCR), an isothermal amplification reaction, or any combination thereof. In some embodiments, the machine learning model comprises a logistic regression algorithm, a random forest algorithm, a lasso regression algorithm, a ridge regression algorithm, an elastic-net regression algorithm, a gradient boosting algorithm, an extreme gradient boosting algorithm, a group LASSO algorithm, a neural network algorithm, a Naive Bayes algorithm, a support vector machine (SVM), or any combination thereof. In some embodiments, (a) further comprises measuring counts of cfNA fragments of the plurality of cfNA fragments to generate count data, wherein (b) further comprises applying the machine learning model to the count data, and wherein the machine learning model has been trained on training data comprising cfNA count data from a plurality of subjects (e.g. those responding to the immunosuppressant drug and those not responding to the immunosuppressant drug). In some embodiments, the method further comprises: (i) identifying a set of features of the data to be input to the machine learning model; (ii) preparing a feature vector of feature values from the data, each feature value corresponding to a feature of the set of features and including one or more measured values; (iii) loading, into memory of a computer system, the machine learning model, the machine learning model having been trained using training vectors obtained from training samples, a first subset of the training samples identified as being from a responder to the immunosuppressant drug and a second subset of the training samples identified as being from a non-responder to the immunosuppressant drug; and (iv) inputting the feature vector into the machine learning model to obtain an output value. In some embodiments, the output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real -valued score. In some embodiments, the output value denotes that the subject is likely to be or is a responder to the immunosuppressant drug.

[0004] In some aspects, the present disclosure provides for a method of processing a sample of cell-free DNA (cfDNA) fragments from a subject identified as having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a plurality of genomic regions, of which expression is associated with activity of a T-cell costimulatory blocker from cfDNA fragments derived from a blood sample of the subject to generate data; and (b) applying a machine learning model to the data, wherein the machine learning model has been trained on training data that comprises: (i) data indicative of presence or absence of the expression for the plurality of genomic regions in a plurality of subjects having the autoimmune condition; and (ii) data indicative of response of each subject of the plurality of subjects to the T-cell costimulatoryWSGR Docket No. 55066-711.601blocker. In some embodiments, a genomic region of the genomic regions corresponds to a promoter, an exon, or a transcription factor binding site. In some aspects, the present disclosure provides for a method of processing a sample of cell-free DNA (cfDNA) fragments from a subject identified as having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a plurality of genomic regions including a transcription factor binding site (TFBS), of which expression is associated with activity of an immunosuppressant drug, from cell-free DNA fragments derived from a blood sample of the subject to generate data; and (b) applying a machine learning model to the data, wherein the machine learning model has been trained with training data that comprises: (i) data indicative of presence of absence of the expression for the plurality of genomic regions in a plurality of subjects having the autoimmune condition; and (ii) data indicative of response of each subject of the plurality of subjects to the immunosuppressant drug. In some embodiments, the immunosuppressant drug comprises a disease-modifying antirheumatic drug (DMARD). In some embodiments, the subject has received or is undergoing treatment with a previous immunosuppressant drug different from the immunosuppressant drug prior to the detecting. In some embodiments, the DMARD comprises a Janus Kinase (JAK) inhibitor. In some embodiments, the DMARD comprises a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD comprises a TNFa inhibitor or a JAK inhibitor, further comprising classifying the subject as responsive to the DMARD. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug or the T-cell costimulatory blocker with at least 80% accuracy. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug or the T-cell costimulatory blocker with at least 85% specificity at at least 80% sensitivity. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting in (a). In some embodiments, the activity is a decrease in Clinical Disease Index (CD Al) of at least about 10 or at least one CD Al category.

[0005] In some aspects, the present disclosure provides for a method of processing a sample comprising nucleic acids from a subject having or suspected of having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a Tumor Necrosis Factor Alpha (TNFa) inhibitor from nucleic acid (NA) derived from a sample of the subject to generate data; and (b) classifying the subject as a responder or a non-responder to the TNFa inhibitor with an accuracy of at least 64% based on the data. In some embodiments, the nucleic acid is a cell-free nucleic acid (cfNA). In some embodiments, the classifying comprises applying a machine learning model to the data, wherein the machineWSGR Docket No. 55066-711.601learning model has been trained on training data comprising: (i) data indicative of the presence or absence of the expression for genomic region in a plurality of subjects having the autoimmune condition; and (ii) data indicative of response or non-response of each subject of the plurality of subjects to the TNFa inhibitor. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder or a non-responder to the T-cell costimulatory blocker with a sensitivity of at least 80% at greater than 86% specificity based on the presence or absence data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate data; and (b) classifying the subject as a responder to the T-cell costimulatory blocker with a specificity greater than 60% at greater than 50% sensitivity based on the data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate data; and (b) classifying the subject as a responder to the T-cell costimulatory blocker with an accuracy of at least about 75% based on the data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate data; and (b) classifying the subject as a responder to the T-cell costimulatory blocker with a positive predictive value (PPV) of at least about 55% based on the data. In some embodiments, the classifying comprises applying a machine learning model to the data, wherein the machine learning model has been trained on training data that comprises: (i) data indicative of presence or absence of the expression for the plurality of genomic regions in a plurality of subjects having the autoimmune condition; and (ii) data indicative of response of each subject of the plurality ofWSGR Docket No. 55066-711.601subjects to the T-cell costimulatory blocker. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi) from nucleic acid (NA) derived from a blood sample of the subject to generate data; and (b) classifying the subject as a responder respond to the JAKi with a specificity greater than 55% at greater than 67% sensitivity based on the data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi) from nucleic acid (NA) derived from a blood sample of the subject to generate data; and (b) classifying the subject as a responder or a non-responder to the JAKi with an accuracy greater than 60% based on the data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region, of which expression is associated with activity of an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi) from nucleic acid (NA) derived from a blood sample of the subject to generate data; and (b) classifying the subject as a responder or a non-responder to the JAKi with a PPV of at least about 69% based on the data. In some embodiments, the classifying comprises applying a machine learning model to the data, wherein the machine learning model has been trained on training data comprising: (i) data indicative of the presence or absence of expression of the genomic region in a plurality of subjects having the autoimmune condition; and (ii) data indicative of response or non-response of each subject of the plurality of subjects to the JAKi. In some embodiments, the genomic region comprises a transcription factor binding site, an exon, a promoter, or any combination thereof. In some embodiments, the nucleic acid is cell-free nucleic acid (cfNA). In some embodiments, the cfNA is cell-free DNA (cfDNA).

[0006] In some aspects, the present disclosure provides for a method of processing a sample comprising cell-free nucleic acid (cfNA) fragments from a subject identified as having an autoimmune condition, and optionally having been administered an immunosuppressant drug, the method comprising: (a) obtaining sequence data of cfNA fragments obtained from a subject that overlap with a genomic region, of which expression of the genomic region is associated with activity of the immunosuppressant drug, wherein the cfNA fragments are at least about 200 bases in length or the sequences range in length from 100 to 1500 bases; (b) from the sequence data, calculating a test sample value indicative of a shape of a distribution of sequence countsWSGR Docket No. 55066-711.601versus sequence length for the genomic region over a range of sequence data obtained in (a); and (c) classifying the subject as a responder or non-responder to the immunosuppressant drug at least partially based on the test sample value calculated in (b). In some embodiments, the method further comprises calculating the test sample value in part by fitting, via a computer system comprising one or more processors and system memory, a multivariate model to a distribution of sequence counts versus length for the genomic region to generate output parameters. In some embodiments, fitting the multivariate model comprises (i) preparing a sequence matrix of sequence lengths, via the computer system, from the data in system memory of the computer system; and (ii) fitting, by the one or more processors, the multivariate model to the sequence matrix by iteratively applying an expectation-maximization algorithm for each sequence length value in the sequence matrix to calculate at least one output variable of the multivariate model, thereby calculating the test sample value. In some aspects, the present disclosure provides for a method of processing a sample of cell-free nucleic acids from a subject identified as having an autoimmune condition, and optionally having been administered an immunosuppressant drug, the method comprising: (a) obtaining sequence data of nucleic acid (NA) fragments derived from a sample of the subject that overlap with a genomic region, of which expression of the genomic region is associated with activity of an immunosuppressant drug; (b) from the sequence data, performing a wavelet transform on a distribution of sequence counts versus sequence length to compute a relative contribution of a cfNA nucleic acid population greater than 200 nucleotides in size, about 100 to about 800 nucleotides in size, or about 100 to about 1500 nucleotides in size; and (c) classifying the subject as a responder or non-responder to the immunosuppressant drug at least partially based on the relative contribution. In some embodiments, applying the wavelet transform on the distribution of the sequence counts versus sequence length comprises: (i) preparing, via a computer system and from the sequence data, a sequence matrix of sequence length values in system memory of the computer system versus sequence counts; (ii) applying, by one or more processors of the computer system, a wavelet transform to the matrix by convolving the sequence length values with a predetermined wavelet function parameterized by at least one scale parameter and at least one translation parameter to generate a plurality of wavelet coefficients; and (iii) computing, by one or more processors of the computer system, at least one output value from the plurality of wavelet coefficients, thereby calculating the relative contribution. In some embodiments, applying the wavelet transform on the distribution of sequence counts versus sequence length further comprises, between (i) and (ii), computing, by one or more processors of a computer system, a windowed signal by applying a window function over sequence lengths greater than 200 nucleotides and applying the wavelet transform to the windowed signal. In someWSGR Docket No. 55066-711.601embodiments, applying the wavelet transform on the distribution of sequence counts versus sequence length further comprises, after (ii), selecting, by the one or more processors of a computer system, a subset of the wavelet coefficients whose translation parameter lies within a fragment length interval greater than 200 nucleotides, and wherein the computing in (iii) comprises computing the at least one output variable from the subset of wavelet coefficients. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject identified as having an autoimmune condition, the method comprising: (a) obtaining sequence data of nucleic acid (NA) fragments derived from a blood sample of the subject that overlap with a genomic region, of which expression of the genomic region is associated with activity of an immunosuppressant drug; (b) from the sequencing data, performing a wavelet transform on a distribution of sequence counts versus sequence length to determine a relative contribution of at least one non-nucleosomal population; and (c) classifying the subject as a responder or non-responder to the immunosuppressant drug at least partially based on the relative contribution. In some embodiments, the classifying further comprises measuring counts of the NA fragments to generate count data and applying a machine learning model to the count data, wherein the machine learning model has been trained on NA count data for a plurality of individuals. In some embodiments, the classifying further comprises: (i) identifying a set of features of the data to be input to the machine learning model; (ii) preparing a feature vector of feature values from the data, each feature value corresponding to a feature of the set of features and including one or more measured values; (iii) loading, into memory of a computer system, the machine learning model, the machine learning model having been trained using training vectors obtained from training samples, a first subset of the training samples identified as being from a responder to the immunosuppressant drug and a second subset of the training samples identified as being from a non-responder to the immunosuppressant drug; and (iv) inputting the feature vector into the machine learning model to obtain an output value. In some embodiments, the output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real-valued score. In some embodiments, the output value denotes that the subject is likely to be or is a responder to the immunosuppressant drug. In some embodiments, the genomic region corresponds to a promoter, exon, transcription factor binding site, gene body exon, insulator, DNAse-I hypersensitive site, RNA polymerase II pausing site, or any combination thereof. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting or the obtaining sequences. In some embodiments, the activity is a decrease in Clinical Disease Index (CD Al) of at least about 10 or at least one CD Al category. In some embodiments, the subject is classified as responsive to the immunosuppressant drug with at least 80% accuracy. In some embodiments, the subject isWSGR Docket No. 55066-711.601classified as responsive to the immunosuppressant drug with at least 85% specificity at at least 80% sensitivity. In some embodiments, the immunosuppressant drug is a DMARD. In some embodiments, the DMARD is a biological disease-modifying antirheumatic drug (bDMARD). In some embodiments, the DMARD is a targeted synthetic disease-modifying antirheumatic drug (csDMARD). In some embodiments, the subject is undergoing or has undergone treatment with another immunosuppressant drug different to the immunosuppressant drug before the detecting. In some embodiments, the DMARD is a Janus Kinase (JAK) inhibitor. In some embodiments, the DMARD is a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD is a TNFa inhibitor or a JAK inhibitor.

[0007] In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments comprise a predominance of nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments corresponding to active chromatin. In some embodiments of any of the aforementioned aspects, the method further comprises prior to the detecting: (i) contacting the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments to an anionic surface to selectively enrich nucleic acids, NA fragments, cfNA fragments, or cfDNA fragments corresponding to active chromatin prior to the detecting, or (ii) performing a size selection on the nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments derived from the blood sample obtained from the subject to NA nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments corresponding to active chromatin prior to the detecting. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length, optionally wherein the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments of any of the aforementioned aspects, the active chromatin comprises chromatin comprising an H3K4mel modification, a H3K27ac modification, an H4K4me2 modification, an H3K4me3 modification, or any combination thereof. In some embodiments of any of the aforementioned aspects, the method further comprises contacting the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments to the anionic surface to selectively enrich nucleic acids, NA fragments, cfNA fragments, or cfDNA fragments corresponding to active chromatin prior to the detecting. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments corresponding to active chromatin are enriched at least 5-fold relative to nucleic acid, the NA fragments, the cfNA fragments, or cfDNA fragments corresponding to inactive chromatin. In some embodiments of any of the aforementionedWSGR Docket No. 55066-711.601aspects: (i) the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are contacted to the anionic surface in the presence of a chaotropic agent; (ii) unbound nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments not enriched for active chromatin is removed from the anionic surface; and (iii) nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments enriched in nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments corresponding to active chromatin are obtained bound to the anionic solid surface, wherein the enrichment does not involve the use of a biomolecule binding agent. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are contacted to the anionic surface in the presence of no more than about 30% alcohol. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are contacted to the anionic surface in the presence of no more than about 15% alcohol. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are contacted to the anionic surface in the presence of about 3.5% to about 11% alcohol. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are contacted to the anionic surface in the presence of about 1.5 molar (M) chaotropic agent. In some embodiments of any of the aforementioned aspects, the nucleic acid, the NA fragments, the cfNA fragments, or the cfDNA fragments are contacted to the anionic surface in the presence of about 1.5 M-2.0 M chaotropic agent. In some embodiments of any of the aforementioned aspects, (ii) further comprises washing the anionic solid surface with a wash solution comprising less than about 30% alcohol. In some embodiments of any of the aforementioned aspects, (ii) further comprises washing the anionic solid surface with a wash solution having a pH of about 8.0. In some embodiments of any of the aforementioned aspects, (iii) further comprises eluting nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments enriched in nucleic acid, NA fragments, cfNA fragments, or cfDNA fragments corresponding to active chromatin from the solid surface using a solution with a pH of about 9.0. In some embodiments of any of the aforementioned aspects, the detecting comprises performing next-generation sequencing. In some embodiments of any of the aforementioned aspects, the next-generation sequencing is configured to obtain sequences of at an average coverage greater than 3 OX. In some embodiments of any of the aforementioned aspects, the next-generation sequencing is configured to obtain sequences at an average coverage of at most about 100X. In some embodiments of any of the aforementioned aspects, the genomic region or the genomic regions are at least about 500 bases in length. In some embodiments of any of the aforementioned aspects, the genomic region or the genomic regions are at most about 4 megabases (Mb) in length. In some embodiments of any of theWSGR Docket No. 55066-711.601aforementioned aspects, the genomic region or the genomic regions are less than about 100 kilobases (Kb) in length. In some embodiments of any of the aforementioned aspects, the genomic region or the genomic regions are less than about 2 kilobases (Kb) in length. In some embodiments of any of the aforementioned aspects, the genomic region or genomic regions do not comprise a transcription factor binding site of ASCL1, NEURODI, POUF23, or YAP. In some embodiments of any of the aforementioned aspects, the genomic region or the genomic regions do not comprise a transcription factor binding site.

[0008] In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information from the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on presence or absence data for a plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the immunosuppressant drug; and not responding to the immunosuppressant drug. In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on a test sample value that is indicative of a shape of a distribution of sequence counts versus length for cfDNA sequences of at least 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length that overlap with a genomic region, of which expression of the genomic region is associated with activity of an immunosuppressant drug, optionally wherein the cfDNA sequences are at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequenceWSGR Docket No. 55066-711.601information from the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on a wavelet coefficient calculated on a distribution of counts of the sequences versus length for cfDNA sequences of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length that overlap with a genomic region, of which expression of the genomic region is associated with activity of an immunosuppressant drug, wherein the cfDNA sequences are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information from the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on a relative contribution of at least one non-nucleosomal population of cfDNA counts calculated via a wavelet transform for cfDNA sequences of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length that overlap with a genomic region of which expression is associated with activity of an immunosuppressant drug, wherein the cfDNA sequences are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments, the machine learning model comprises a logistic regression algorithm, a random forest algorithm, a lasso regression algorithm, a ridge regression algorithm, an elastic-net regression algorithm, a gradient boosting algorithm, an extreme gradient boosting algorithm, a group LASSO algorithm, a neural network algorithm, a Naive Bayes algorithm, a support vector machine (SVM), or any combination thereof. In some embodiments, the training dataset further comprises (3) clinical or demographic characteristics for the plurality of individuals having the autoimmune disease, and the method further comprises training a propensity model using clinical or demographic characteristics for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a propensity score; and in (b), weighing the samples during the training inversely to the propensity score. In some embodiments, the training dataset further comprises (3) clinicalWSGR Docket No. 55066-711.601characteristics for the plurality of individuals having the autoimmune disease, and the method further comprises training a propensity model using clinical or demographic characteristics for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a propensity score; and in (b), including the propensity score as a feature during the training. In some embodiments, the propensity model comprises a logistic regression model, a generalized additive model, a gradient boosting machine, a Bayesian logistic regression, or any combination thereof. In some embodiments, the clinical or demographic characteristics comprise at least one of age, sex, BMI, ancestry, smoking status, glucocorticoid use, methotrexate dose, number of previous DMARDs used, number of previous bDMARDs used, C-reactive protein, erythrocyte sedimentation rate, neutrophil count, lymphocyte count, platelet count, hemoglobin, LDL-C, HDL-C, triglycerides, ApoA, ApoB, GM-CSF titer, IFN-y titer, TNF-a titer, IL-1 titer, IL-2 titer, IL-6 titer, IL-8 titer, or IL- 10 titer. In some embodiments, the training dataset further comprises (4) a demographic variable, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals having the autoimmune disease, and the method further comprises training a sampling / enrollment bias model using demographic variables, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a sampling / enrollment bias score; and in (b), weighing the samples during the training inversely to the sampling / enrollment bias score. In some embodiments, the training dataset further comprises (4) a demographic variable, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals having the autoimmune disease, and the method further comprises training a sampling / enrollment bias model using demographic variables, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a sampling / enrollment bias score; and in (b), including the sampling / enrollment bias score as a feature during the training. In some embodiments, the training dataset further comprises (5) a mode of processing, date of processing, or location of processing for dataset entries for the plurality of individuals having the autoimmune disease, and the method further comprises training a measurement error / batch-effect model using demographic variables, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a sampling / enrollment bias score; and in (b), stratifying samples according to the sampling / enrollment bias score during the training.WSGR Docket No. 55066-711.601

[0009] In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject identified as having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a plurality of genomic regions associated with an autoimmune disease response to an immunosuppressant drug from cell-free DNA (cfDNA) fragments derived from a blood sample of the subject to generate presence or absence data; and (b) applying a machine learning model to the presence or absence data for the plurality of genomic regions associated with the autoimmune disease response to the immunosuppressant drug, wherein the machine learning model has been trained on training presence or absence data for the plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the immunosuppressant drug; and not responding to the immunosuppressant drug. In some embodiments, the plurality of genomic regions comprise a genomic region corresponding to a promoter, exon, or transcription factor binding site. In some embodiments, the immunosuppressant drug is a disease-modifying antirheumatic drug (DMARD). In some embodiments, the subject is undergoing treatment or has undergone treatment with a previous immunosuppressant drug different to the immunosuppressant drug before the detecting. In some embodiments, the DMARD is a Janus kinase (JAK) inhibitor. In some embodiments, the DMARD is a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD is a TNFa inhibitor or a JAK inhibitor, further comprising classifying the subject as responsive to the DMARD. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug with at least 80% accuracy. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug with at least 85% specificity and at least 80% sensitivity. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting in (a). In some embodiments, the autoimmune disease response to the immunosuppressant drug is a decrease in Clinical Disease Index (CD Al) of at least about 10 or a decrease in at least one CD Al category. In some embodiments, the plurality of genomic regions comprise transcription factor binding sites, exons, promoters, or any combination thereof. In some embodiments, the cfDNA fragments comprise a predominance of cfDNA fragments derived from active chromatin. In some embodiments, the method further comprises: (i) contacting the cfDNA fragments derived from the blood sample obtained from the subject to an anionic surface to selectively enrich the active chromatin prior to the detecting, or (ii) performing a size selection on the cfDNA fragments derived from the blood sample obtained from the subject to enrich the active chromatin prior to the detecting. In some embodiments, the active chromatin comprises cfDNA fragments of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length, wherein the cfDNAWSGR Docket No. 55066-711.601fragments are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments, the genomic regions comprise transcription factor binding sites, exons, gene body exons, promoters, insulators, DNAse-I hypersensitive sites, or RNA polymerase II pausing sites, or any combination thereof. In some embodiments, the method further comprises classifying the subject as responsive to the immunosuppressant drug. In some embodiments, the method further comprises administering the immunosuppressant drug to the subject identified as responsive to the immunosuppressant drug. In some embodiments, the method further comprises reporting an identification of the subject as responsive to the immunosuppressant drug to a physician, a caregiver, or the subject, wherein the identification is performed at least in part based on the machine learning model. In some embodiments, the detecting the presence or absence of the plurality of genomic regions comprises performing a next-generation sequencing procedure. In some embodiments, the detecting the presence of absence of the plurality of genomic regions comprises performing a quantitative reverse polymerase chain reaction (qPCR), a reverse transcriptase polymerase chain reaction (rtPCR), a digital droplet polymerase chain reaction (ddPCR), an isothermal amplification reaction, or any combination thereof. In some embodiments, the machine learning model comprises a logistic regression algorithm, a random forest algorithm, a lasso regression algorithm, a ridge regression algorithm, an elastic-net regression algorithm, a gradient boosting algorithm, an extreme gradient boosting algorithm, a group LASSO algorithm, a neural network algorithm, a Naive Bayes algorithm, a support vector machine (SVM), or any combination thereof. In some embodiments, (a) further comprises determining counts of the plurality of genomic regions to generate count data, and wherein (b) further comprises applying the machine learning model to the count data, and wherein the machine learning model has been trained on the count data from a plurality of subjects: responding to the immunosuppressant drug; and not responding to the immunosuppressant drug. In some embodiments, applying the machine learning model to the presence or absence data or the count data comprises: (i) identifying a set of features corresponding to a plurality of the genomic regions to be input to the machine learning model; (ii) preparing a feature vector of feature values from the plurality of genomic regions, each feature value corresponding to a feature of the set of features and including one or more measured values; (iii) loading, into memory of a computer system, the machine learning model, the machine learning model having been trained using training vectors obtained from training biological samples, a first subset of the biological samples identified as being from a responder to the immunosuppressant drug and a second subset of the training biological samples identified as being from a non-responder to the immunosuppressant drug; and (iv) inputting the feature vector into the machine learning model to obtain an output value. In some embodiments, theWSGR Docket No. 55066-711.601output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real -valued score. In some embodiments, the output value denotes that the subject is likely to be or is a responder to the immunosuppressant drug.

[0010] In some aspects, the present disclosure provides for a method of processing a sample of cell-free DNA (cfDNA) fragments from a subject identified as having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a plurality of genomic regions associated with an autoimmune disease response to a T-cell costimulatory blocker from cfDNA fragments derived from a blood sample of the subject to generate presence or absence data; and (b) applying a machine learning model to the presence or absence data for the plurality of genomic regions associated with the autoimmune disease response to the T-cell costimulatory blocker, wherein the machine learning model has been trained on presence or absence data for the plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the T-cell costimulatory blocker; and not responding to the T-cell costimulatory blocker. In some embodiments, a genomic region of the genomic regions corresponds to a promoter, an exon, or a transcription factor binding site.

[0011] In some aspects, the present disclosure provides for a method of processing a sample of cell-free DNA (cfDNA) fragments from a subject identified as having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a plurality of genomic regions including a transcription factor binding site (TFBS) associated with an autoimmune disease response to an immunosuppressant drug from cell-free DNA fragments derived from a blood sample of the subject to generate presence or absence data; and (b) applying a machine learning model to the presence or absence data for the plurality of genomic regions including the TFBS associated with the autoimmune disease response to the immunosuppressant drug, wherein the machine learning model has been trained on presence or absence data for the plurality of genomic regions including the TFBS from a plurality of autoimmune condition subjects: responding to the immunosuppressant drug; and not responding to the immunosuppressant drug. In some embodiments, the immunosuppressant drug comprises a disease-modifying antirheumatic drug (DMARD). In some embodiments, the subject is undergoing treatment with a previous immunosuppressant drug different to the immunosuppressant drug prior to the detecting. In some embodiments, the DMARD comprises a Janus Kinase (JAK) inhibitor. In some embodiments, the DMARD comprises a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD comprises a TNFa inhibitor or a JAK inhibitor, further comprising classifying the subject as responsive to the DMARD. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug or the T-cell costimulatory blocker with at least 80%WSGR Docket No. 55066-711.601accuracy. In some embodiments, the machine learning model is capable of classifying the subject as responsive to the immunosuppressant drug or the T-cell costimulatory blocker with at least 85% specificity at at least 80% sensitivity. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting in (a). In some embodiments, the autoimmune disease response to the immunosuppressant drug is a decrease in Clinical Disease Index (CD Al) of at least about 10 or at least one CD Al category. In some embodiments, the subject is undergoing or has undergone treatment with a previous immunosuppressant drug different to the immunosuppressant drug before the detecting. In some embodiments, the nucleic acid comprises a predominance of nucleic acids derived from active chromatin or the cfNA fragments comprise a predominance of cfNA fragments derived from active chromatin.

[0012] In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune condition, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a Tumor Necrosis Factor Alpha (TNFa) inhibitor from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder or a non-responder to the TNFa inhibitor with an accuracy of at least 64% based on the presence or absence data. In some embodiments, nucleic acid is a cell-free nucleic acid (cfNA). In some embodiments, the classifying comprises applying a machine learning model to the presence or absence data, wherein the machine learning model has been trained on training presence or absence data for the plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the Tumor Necrosis Factor Alpha (TNFa) inhibitor; and not responding to the Tumor Necrosis Factor Alpha (TNFa) inhibitor. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder or a non-responder to the T-cell costimulatory blocker with a sensitivity of at least 80% at greater than 86% specificity based on the presence or absence data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a T-cellWSGR Docket No. 55066-711.601costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder to the T-cell costimulatory blocker with a specificity greater than 60% at greater than 50% sensitivity based on the presence or absence of data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises an T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder to the T-cell costimulatory blocker with an accuracy of at least about 75% based on the presence or absence data. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a T-cell costimulatory blocker from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder to the T-cell costimulatory blocker with a positive predictive value (PPV) of at least about 55% based on the presence or absence data. In some embodiments, the classifying comprises applying a machine learning model to the presence or absence data, wherein the machine learning model has been trained on training presence or absence data for the plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the T-cell costimulatory blocker; and not responding to the T-cell costimulatory blocker. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi) from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder respond to the JAKi with a specificity greater than 55% at greater than 67% sensitivity based on the presence or absence data, wherein the subject has an autoimmune disease. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi) from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder or a non-responder to theWSGR Docket No. 55066-711.601JAKi with an accuracy greater than 60% based on the presence or absence data, wherein the subject has an autoimmune disease. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising: (a) detecting a presence or absence of a genomic region associated with an autoimmune disease response to an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi) from nucleic acid (NA) derived from a blood sample of the subject to generate presence or absence data; and (b) classifying the subject as a responder or a non-responder to the JAKi with a PPV of at least about 69% based on the presence or absence data, wherein the subject has an autoimmune disease. In some embodiments, the classifying comprises applying a machine learning model to the presence or absence data for the plurality of genomic regions associated with an autoimmune disease response to the JAKi, wherein the machine learning model has been trained on training presence or absence data for the plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the JAKi; and not responding to the JAKi. In some embodiments, the genomic region comprises a transcription factor binding site, an exon, a promoter, or any combination thereof. In some embodiments, nucleic acid is cell-free nucleic acid (cfNA). In some embodiments, cfNA is cell-free DNA (cfDNA). In some embodiments, the classifying further comprises detecting an abundance of the genomic region associated with the autoimmune disease response to the immunosuppressant drug to generate abundance data and applying a machine learning model to the abundance data, wherein the machine learning model has been trained on training abundance data for the plurality of individuals having predetermined autoimmune disease responses to the immunosuppressant drug. In some embodiments, the classifying further comprises: (i) identifying a set of features corresponding to a plurality of genomic regions including the genomic region to be input to the machine learning model; (ii) preparing a feature vector of feature values from the plurality of genomic regions, each feature value corresponding to a feature of the set of features and including one or more measured values; (iii) loading, into memory of a computer system, the machine learning model, the machine learning model having been trained using training vectors obtained from training biological samples, a first subset of the biological samples identified as being from a responder to the immunosuppressant drug and a second subset of the training biological samples identified as being from a non-responder to the immunosuppressant drug; and (iv) inputting the feature vector into the machine learning model to obtain an output value. In some embodiments, the output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real -valued score. In some embodiments, the output value denotes that the subject is likely to be or is a responder to the immunosuppressant drug. In some embodiments, the genomic region corresponds to aWSGR Docket No. 55066-711.601promoter, exon, transcription factor binding site, gene body exon, insulator, DNAse-I hypersensitive site, RNA polymerase II pausing site, or any combination thereof. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting or the obtaining sequences. In some embodiments, the autoimmune disease response to the immunosuppressant drug is a decrease in Clinical Disease Index (CD Al) of at least about 10 or at least one CD Al category. In some embodiments, the subject is classified as responsive to the immunosuppressant drug with at least 80% accuracy. In some embodiments, the subject is classified as responsive to the immunosuppressant drug with at least 85% specificity at at least 80% sensitivity.

[0013] In some aspects, the present disclosure provides for a method of processing a sample of cell-free nucleic acid (cfNA) fragments from a subject identified as having an autoimmune condition, the method comprising: (a) obtaining sequences of cfNA fragments derived from a blood sample of the subject that overlap with a genomic region associated with an autoimmune disease response to an immunosuppressant drug, wherein the sequences are at least about 200 bases in length or the sequences range in length from 100 to 1500 bases; (b) for the genomic region, calculating a test sample value indicative of a shape of a distribution of sequence counts versus length for the genomic region over the range of data for the genomic region; and (c) classifying the subject as a responder or non-responder to the immunosuppressant drug at least partially based on the test sample value. In some embodiments, the test sample value is calculated in part by fitting, via a computer system comprising one or more processors and system memory, a multivariate formulation to a distribution of cfNA fragment counts versus length for the genomic region to generate output parameters. In some embodiments, fitting the multivariate formulation comprises: (i) preparing a sequence matrix of cfNA fragment lengths, via the computer system, from the sequences that overlap with the genomic region in the system memory of the computer system; and (ii) fitting, by the one or more processors, the multivariate formulation to the sequence matrix by iteratively applying an expectation-maximization algorithm for each cfNA fragment length value in the matrix to calculate at least one output variable of the multivariate formulation, thereby calculating the test sample value. In some embodiments, the present disclosure provides for a method of processing a sample of cell-free nucleic acids from a subject identified as having an autoimmune condition, the method comprising: (a) obtaining sequences of nucleic acid (NA) fragments derived from a blood sample of the subject that overlap with a genomic region associated with an autoimmune disease response to an immunosuppressant drug; (b) for the genomic region, performing a wavelet transform on a distribution of counts of the sequences versus length to compute a relative contribution of a cfNA nucleic acid population greater than 200 nucleotides in size, about 100 toWSGR Docket No. 55066-711.601about 800 nucleotides in size, or about 100 to about 1500 nucleotides in size; and (c) classifying the subject as a responder or non-responder to the immunosuppressant drug at least partially based on the relative contribution. In some embodiments, applying the wavelet transform on the distribution of the counts of the sequences versus length comprises: (i) preparing, via a computer system, a sequence matrix of fragment length values of the sequences that overlap with the genomic region in system memory of the computer system versus counts of fragments that overlap with the genomic region; (ii) applying, by one or more processors of the computer system, a wavelet transform to the matrix by convolving the fragment length values with a predetermined wavelet function parameterized by at least one scale parameter and at least one translation parameter to generate a plurality of wavelet coefficients; and (iii) computing, by one or more processors of the computer system, at least one output value from the plurality of wavelet coefficients, thereby calculating the relative contribution. In some embodiments, applying the wavelet transform to the distribution sequence matrix further comprises, between (i) and (ii), computing, by one or more processors of a computer system, a windowed signal by applying a window function over fragment lengths greater than 200 nucleotides and applying the wavelet transform to the windowed signal. In some embodiments, applying the wavelet transform to the distribution of sequence counts versus length further comprises, after (ii), selecting, by the one or more processors of a computer system, a subset of the wavelet coefficients whose translation parameter lies within a fragment length interval greater than 200 nucleotides, and wherein the computing in (iii) comprises computing the at least one output variable from the subset of wavelet coefficients. In some aspects, the present disclosure provides for a method of processing a sample of nucleic acids from a subject identified as having an autoimmune condition, the method comprising: (a) obtaining sequences of nucleic acid (NA) fragments derived from a blood sample of the subject that overlap with a genomic region associated with an autoimmune disease response to an immunosuppressant drug; (b) for the genomic region, performing a wavelet transform on a distribution of counts of the sequences versus length to determine a relative contribution of at least one non-nucleosomal population; and (c) classifying the subject as a responder or non-responder to the immunosuppressant drug at least partially based on the relative contribution. In some embodiments, the classifying further comprises detecting an abundance of the genomic region associated with the autoimmune disease response to the immunosuppressant drug to generate abundance data and applying a machine learning model to the abundance data, wherein the machine learning model has been trained on training abundance data for the plurality of individuals having predetermined autoimmune disease responses to the immunosuppressant drug. In some embodiments, the classifying further comprises: (i) identifying a set of features correspondingWSGR Docket No. 55066-711.601to a plurality of genomic regions including the genomic region to be input to the machine learning model; (ii) preparing a feature vector of feature values from the plurality of genomic regions, each feature value corresponding to a feature of the set of features and including one or more measured values; (iii) loading, into memory of a computer system, the machine learning model, the machine learning model having been trained using training vectors obtained from training biological samples, a first subset of the biological samples identified as being from a responder to the immunosuppressant drug and a second subset of the training biological samples identified as being from a non-responder to the immunosuppressant drug; and (iv) inputting the feature vector into the machine learning model to obtain an output value. In some embodiments, the output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real-valued score. In some embodiments, the output value denotes that the subject is likely to be or is a responder to the immunosuppressant drug. In some embodiments, the genomic region corresponds to a promoter, exon, transcription factor binding site, gene body exon, insulator, DNAse-I hypersensitive site, RNA polymerase II pausing site, or any combination thereof. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting or the obtaining sequences. In some embodiments, the autoimmune disease response to the immunosuppressant drug is a decrease in Clinical Disease Index (CD Al) of at least about 10 or at least one CD Al category. In some embodiments, the subject is classified as responsive to the immunosuppressant drug with at least 80% accuracy. In some embodiments, the subject is classified as responsive to the immunosuppressant drug with at least 85% specificity at at least 80% sensitivity. In some embodiments, the immunosuppressant drug is a DMARD. In some embodiments, the DMARD is a biological disease-modifying antirheumatic drug (bDMARD). In some embodiments, DMARD is a targeted synthetic disease-modifying antirheumatic drug (csDMARD). In some embodiments, the subject is undergoing or has undergone treatment with a previous immunosuppressant drug different to the immunosuppressant drug before the detecting. In some embodiments, DMARD is a Janus Kinase (JAK) inhibitor. In some embodiments, the DMARD is a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD is a TNFa inhibitor or a JAK inhibitor. In some embodiments, the nucleic acid comprises a predominance of nucleic acids derived from active chromatin or the cfNA fragments comprise a predominance of cfNA fragments derived from active chromatin.

[0014] In some embodiments of any of the foregoing aspects, the method further comprises: (i) contacting the cfNA fragments or the nucleic acid to an anionic surface to selectively enrich active chromatin prior to the detecting, or (ii) performing a size selection on the nucleic acid or the cfNA fragments derived from the blood sample obtained from the subject to enrich activeWSGR Docket No. 55066-711.601chromatin prior to the detecting. In some embodiments of any of the foregoing aspects, the active chromatin comprises cfDNA fragments of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length, wherein the cfDNA fragments are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments of any of the foregoing aspects, the active chromatin comprises chromatin comprising an H3K4mel modification, a H3K27ac modification, an H4K4me2 modification, an H3K4me3 modification, or any combination thereof. In some embodiments of any of the foregoing aspects, the method further comprises contacting the cfNA fragments or the nucleic acid to an anionic surface to selectively enrich the active chromatin prior to the detecting or the obtaining sequences. In some embodiments of any of the foregoing aspects, the active chromatin is enriched at least 5-fold relative to inactive chromatin. In some embodiments of any of the foregoing aspects: (i) the cfNA fragments or the nucleic acid are contacted to the anionic surface in the presence of a chaotropic agent; (ii) unbound cfNA fragments or nucleic acid not enriched for active chromatin is removed from the anionic surface; and (iii) cfNA fragments or nucleic acid enriched in active chromatin is obtained bound to the anionic solid surface, wherein the enrichment does not involve the use of a biomolecule binding agent. In some embodiments of any of the foregoing aspects, the cfNA fragments or the nucleic acid are contacted to the anionic surface in the presence of no more than about 30% alcohol. In some embodiments of any of the foregoing aspects, the cfNA fragments or the nucleic acid are contacted to the anionic surface in the presence of no more than about 15% alcohol. In some embodiments of any of the foregoing aspects, the cfNA fragments or the nucleic acid are contacted to the anionic surface in the presence of about 3.5% to about 11% alcohol. In some embodiments of any of the foregoing aspects, the cfNA fragments or the nucleic acid are contacted to the anionic surface in the presence of about 1.5 molar (M) chaotropic agent. In some embodiments of any of the foregoing aspects, the cfNA fragments or the nucleic acid are contacted to the anionic surface in the presence of about 1.5 M-2.0 M chaotropic agent. In some embodiments of any of the foregoing aspects, (ii) further comprises washing the anionic solid surface with a wash solution comprising less than about 30% alcohol. In some embodiments of any of the foregoing aspects, (ii) further comprises washing the anionic solid surface with a wash solution having a pH of about 8.0. In some embodiments of any of the foregoing aspects, (iii) further comprises eluting the cfNA or nucleic acid enriched in active chromatin from the solid surface using a solution with a pH of about 9.0. In some embodiments of any of the foregoing aspects, detecting the presence or the absence of the genomic region or the genomic regions comprises a nextgeneration sequencing operation, or wherein obtaining the sequences comprises a nextgeneration sequencing operation. In some embodiments of any of the foregoing aspects, theWSGR Docket No. 55066-711.601next-generation sequencing operation is configured to obtain sequences of the cfNA fragments or the nucleic acid at an average coverage greater than 3 OX. In some embodiments of any of the foregoing aspects, the next-generation sequencing operation is configured to obtain sequences of the cfNA or the nucleic acid at an average coverage of at most about 100X. In some embodiments of any of the foregoing aspects, the genomic region or the genomic regions are at least about 500 bases in length. In some embodiments of any of the foregoing aspects, the genomic region or the genomic regions are at most about 4 megabases (Mb) in length. In some embodiments of any of the foregoing aspects, the genomic region or the genomic regions are less than about 100 kilobases (Kb) in length. In some embodiments of any of the foregoing aspects, the genomic region or the genomic regions are less than about 2 kilobases (Kb) in length. In some embodiments of any of the foregoing aspects, the genomic region or genomic regions do not comprise a transcription factor binding site of ASCL1, NEURODI, POUF23, or YAP. In some embodiments of any of the foregoing aspects, the genomic region or the genomic regions do not comprise a transcription factor binding site.

[0015] In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information from the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on presence or absence data for presence or absence data for a plurality of genomic regions from a plurality of autoimmune condition subjects: responding to the immunosuppressant drug; and not responding to the immunosuppressant drug. In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on a test sample value that is indicative of a shape of a distribution of sequence counts versus length for cfDNA sequences ofWSGR Docket No. 55066-711.601at least 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length that overlap with a genomic region associated with an autoimmune disease response to an immunosuppressant drug, optionally wherein the cfDNA sequences are at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information from the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on a wavelet coefficient calculated on a distribution of counts of the sequences versus length for cfDNA sequences of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length that overlap with a genomic region associated with an autoimmune disease response to an immunosuppressant drug, wherein the cfDNA sequences are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some aspects, the present disclosure provides for a method for generating an immunosuppressant drug response model, comprising: (a) receiving, into computer memory, a training dataset comprising, for each of a plurality of individuals having an autoimmune disease: (1) cell-free nucleic acid (cfDNA) fragment sequence information from the plurality of individuals generated at a first time point; and (2) treatment response of the plurality of individuals to an immunosuppressant drug determined at a second later time point; and (b) using the training dataset to subject a classification model to training to yield a trained machine learning model, wherein the trained machine learning model is configured to predict an immunosuppressant drug response of a subject based at least partly on a relative contribution of at least one non-nucleosomal population of cfDNA counts calculated via a wavelet transform for cfDNA sequences of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length that overlap with a genomic region associated with an autoimmune disease response to an immunosuppressant drug, wherein the cfDNA sequences are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments, the classification model comprises a logistic regression algorithm, a random forest algorithm, a lasso regression algorithm, a ridge regression algorithm, an elastic-net regression algorithm, a gradient boosting algorithm, an extreme gradient boosting algorithm, a group LASSO algorithm, a neural network algorithm, a Naive Bayes algorithm, a support vector machine (SVM), or any combination thereof. In someWSGR Docket No. 55066-711.601embodiments, the training dataset further comprises (3) clinical or demographic characteristics for the plurality of individuals having the autoimmune disease, further comprising training a propensity model using clinical or demographic characteristics for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a propensity score; and in (b), weighing the samples during the training inversely to the propensity score. In some embodiments, the training dataset further comprises (3) clinical characteristics for the plurality of individuals having the autoimmune disease, further comprising training a propensity model using clinical or demographic characteristics for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a propensity score; and in (b), including the propensity score as a feature during the training. In some embodiments, the propensity model comprises a logistic regression model, a generalized additive model, a gradient boosting machine, a Bayesian logistic regression, or any combination thereof. In some embodiments, the clinical or demographic characteristics comprise at least one of age, sex, BMI, ancestry, smoking status, glucocorticoid use, methotrexate dose, number of previous DMARDs used, number of previous bDMARDs used, C-reactive protein, erythrocyte sedimentation rate, neutrophil count, lymphocyte count, platelet count, hemoglobin, LDL-C, HDL-C, triglycerides, ApoA, ApoB, GM-CSF titer, IFN-y titer, TNF-a titer, IL-1 titer, IL-2 titer, IL-6 titer, IL-8 titer, or IL- 10 titer. In some embodiments, the training dataset further comprises (4) a demographic variable, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals having the autoimmune disease, further comprising training a sampling / enrollment bias model using demographic variables, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a sampling / enrollment bias score; and in (b), weighing the samples during the training inversely to the sampling / enrollment bias score. In some embodiments, the training dataset further comprises (4) a demographic variable, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals having the autoimmune disease, further comprising training a sampling / enrollment bias model using demographic variables, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a sampling / enrollment bias score; and in (b), including the sampling / enrollment bias score as a feature during the training. In some embodiments, the training dataset further comprises (5) a mode of processing, date of processing, or location ofWSGR Docket No. 55066-711.601processing for dataset entries for the plurality of individuals having the autoimmune disease, further comprising training a measurement error / batch-effect model using demographic variables, insurance type, socioeconomic index, or distance to sampling site for the plurality of individuals and treatment response of the plurality of individuals to the immunosuppressant drug; for each individual in the training dataset, generating a sampling / enrollment bias score; and in (b), stratifying samples according to the sampling / enrollment bias score during the training.

[0016] In some aspects, the present disclosure provides for a method, comprising: detecting a presence or absence of at least one of the unique genes listed in Tables Bl, B2, B3, B4, B5, B6, B7, B8, or any combination thereof from nucleic acid obtained from a sample or a fraction thereof from a subject identified as having an autoimmune disease to generate presence or absence data, wherein the method involves detecting fewer than about 10,000, fewer than about 9,000, fewer than about 8,000, fewer than about 7,000, fewer than about 8,000, fewer than about 7,000, fewer than about 6,000, fewer than about 5,000, fewer than about 4,000, fewer than about 3,000, fewer than about 2,000, fewer than about 1,000, fewer than about 900, fewer than about 800, fewer than about 700, fewer than about 600, fewer than about 500, fewer than about 400, or fewer than about 300 genes. In some embodiments, the sample is a fluid sample. In some embodiments, the fluid sample comprises blood or a fraction thereof. In some embodiments, the method comprises contacting cfNA from a sample from the subject with a bait of about 80 to about 120 bases in length complementary to at least one of the unique genes. In some embodiments, the method comprises contacting the cfNA from the sample from the subject with a plurality of baits complementary to tiled across at least one of the unique genes. In some embodiments, the method further comprises detecting an abundance of the at least one of the unique genes to generate abundance data for the unique genes or the unique loci. In some embodiments, the method further comprises applying a machine learning model to the presence or absence data or the abundance data, wherein the machine learning model has been trained on training presence or absence data or abundance data from a plurality of autoimmune condition subjects: responding to an immunosuppressant drug; and not responding to the immunosuppressant drug. In some embodiments, applying the machine learning model further comprises: (i) identifying a set of features corresponding to a plurality of the at least one of the unique genes to be input to the machine learning model; (ii) preparing a feature vector of feature values from the plurality of the at least one of the unique genes, each feature value corresponding to a feature of the set of features and including one or more measured values; (iii) loading, into memory of a computer system, the machine learning algorithm, the machine learning model having been trained using training vectors obtained from training biologicalWSGR Docket No. 55066-711.601samples, a first subset of the biological samples identified as being from a responder to the immunosuppressant drug and a second subset of the training biological samples identified as being from a non-responder to the immunosuppressant drug; and (iv) inputting the feature vector into the machine learning model to obtain an output value. In some embodiments, the output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real -valued score. In some embodiments, the output value indicates that the subject is likely to be or is a responder to the immunosuppressant drug. In some embodiments, the subject has been identified as having Rheumatoid Arthritis prior to the detecting. In some embodiments, the immunosuppressant drug is a disease-modifying antirheumatic drug (DMARD). In some embodiments, the DMARD is a biological disease-modifying antirheumatic drug (bDMARD). In some embodiments, the subject is undergoing or has undergone treatment with a previous immunosuppressant drug different to the immunosuppressant drug before the detecting. In some embodiments, DMARD is a Janus Kinase (JAK) inhibitor. In some embodiments, the DMARD is a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker. In some embodiments, the DMARD is a TNFa inhibitor or a JAK inhibitor. In some embodiments, the nucleic acid comprises nucleic acids predominantly derived from active chromatin. In some embodiments, the method further comprises: (i) contacting the nucleic acid to an anionic surface to selectively enrich the active chromatin prior to the detecting, or (ii) performing a size selection on the nucleic acid derived from the blood sample obtained from the subject to enrich the active chromatin prior to the detecting. In some embodiments, the active chromatin comprises cfDNA fragments of at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length, wherein the cfDNA fragments are optionally at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments, the active chromatin comprises chromatin comprising an H3K4mel modification, a H3K27ac modification, an H4K4me2 modification, an H3K4me3 modification, or any combination thereof.

[0017] Additional aspects and advantages of the present disclosure will become readily apparent to those skilled in this art from the following detailed description, wherein only illustrative embodiments of the present disclosure are shown and described. As will be realized, the present disclosure is capable of other and different embodiments, and its several details are capable of modifications in various obvious respects, all without departing from the disclosure.Accordingly, the drawings and description are to be regarded as illustrative in nature, and not as restrictive.WSGR Docket No. 55066-711.601INCORPORATION BY REFERENCE

[0018] All publications, patents, and patent applications mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference.BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of inventions according to the disclosure will be obtained by reference to the following detailed description that sets forth illustrative embodiments, in which the principles of the invention are utilized, and the accompanying drawings of which:

[0020] FIG. 1 shows an example workflow utilizing a therapeutic assignment classifier as described herein. A sample 105 can be obtained from a subject 101 with an autoimmune disease and subjected to processing procedures 110 to isolate cfDNA fragments (which can include selective enrichment of cfDNAac fragments). The cfDNA can be subjected to an evaluation procedure 115 to identify cfDNA fragment overlap (or counts thereof) with particular genomic regions (which can include promoters, enhancers, exons, transcription factor binding sites, or insulators). The cfDNA genomic region information can be used to compile a feature vector 125 which represents predetermined features to be input into a machine learning model 125, which machine learning model has been trained on feature information for autoimmune patients that respond and do not respond to a particular DMARD. The machine learning model 125 outputs a classification or score 130 which relates to whether the subject will respond or not respond to the DMARD.

[0021] FIG. 2 shows aspects of cfDNA active chromatin enrichment according to some aspects of the disclosure. FIG. 2 panel A shows an example workflow for such cfDNA active chromatin enrichment wherein plasma samples subjected to a lysis procedure 205 are incubated with nanoparticles or beads under conditions configured to selectively bind 210 and (after separation and wash procedures 215 and 220) to selectively elute 225 a non-canonical or “active chromatin” fraction of cfDNA. FIG. 2 panel B provides an example trace showing the output of such an active chromatin capture workflow compared to an off-the-shelf workflow configured for non-specific capture of cfDNA fragments. The trace shows that the active chromatin capture workflow selectively enriches fragments greater than e.g. 200, 250, 400, or 600 bp.

[0022] FIG. 3 shows example data analysis workflows according to the disclosure that involve functional analyses of cfDNA fragments prior to machine learning. FIG. 3 panel A shows anWSGR Docket No. 55066-711.601example where a fit to a multivariate formulation (aka multivariate model) can be used to extract test values related to a cfDNA fragment size distribution; in this workflow a histogram or bins of fragment count versus length for a genomic region 301 can be subjected to an optional preprocessing procedure (e.g. a density estimation procedure) 305 followed by a fit to a multivariate formulation 310, which can be e.g. by a maximum-likelihood algorithm. The multivariable fit can result in a set of fit variable values of the multivariable formulation that serve as test values characteristic of the fragment length distribution, of which individual values or subsets of values can be input into a feature vector 320 for future processing by machine learning models. FIG. 3 panel B shows an example where a transform (e.g. a wavelet transform) can be used to extract coefficients related to a cfDNA fragment size distribution; in this workflow a histogram or bins of fragment count versus length for a genomic region 301 can be subjected to an optional preprocessing procedure (e.g. a density estimation procedure) 305 followed by a wavelet transform 312 (e.g. a continuous wavelet transform with a morlet wavelet of defined parameters). The wavelet transform can result in a set of wavelet coefficients 317, of which individual coefficients or subsets of coefficients can be input into a feature vector 320 for future processing by machine learning models.

[0023] FIG.4 depicts a study design schematic for the study described in Example 1. The study was designed to identify markers of subsequent response to biologic disease-modifying antirheumatic drugs (bDMARDs) or targeted synthetic disease-modifying antirheumatic drugs (tsDMARDs) in subjects with an autoimmune disease (e.g. Rheumatoid Arthritis). Subjects with prior treatment with at least one conventional synthetic disease-modifying antirheumatic drug (csDMARD), biologic disease-modifying antirheumatic drug (bDMARD), or targeted synthetic disease-modifying antirheumatic drug (tsDMARD) with insufficient response for which initiation of a new b / ts-DMARD was planned were included in the study. A blood draw and disease activity assessments were performed on these subjects to establish a baseline; the disease activity assessments included Clinical Disease Activity Index (CD Al), Disease Activity Score using 28 joints and C-reactive protein (DAS-28 CRP), Simplified Disease Activity Index (SDAI), European Alliance of Associations for Rheumatology (EULAR) Response), and visual analog scale (VAS). Patients who actually initiated a new b / ts-DMARD within 30 days of the baseline were subjected to a second blood draw and disease activity assessment. The primary endpoint for the study was change in baseline from Clinical Disease Activity Index (CD Al) at 12 weeks after initiating a new DMARD. A response was defined by minimally clinically important differences in CD Al: e.g. if disease activity was high (e.g. CDIA > 22) at baseline then a response was counted if the subject had a decrease of > 12 points; if disease activity wasWSGR Docket No. 55066-711.601“moderate” at baseline (e.g. CDIA >10 and < 22) then a response was counted if the subject had a decrease of > 6 points.

[0024] FIG. 5 depicts a study flow diagram for the study of Example 1, demonstrating stats for enrollment, exclusions, follow-up, and analysis of the study of Example 1. 830 patients were initially enrolled and received a blood test and baseline disease activity assessments, of which n=167 were later excluded for not receiving an on-study b / ts-DMARD within the timeframe. Of the 663 patients that did receive an on-study b / ts-DMARD, 297 had not yet received a follow-up visit so were excluded from this analysis, and 28 patients discontinued after DMARD initiation. Of the 338 patients who completed the study follow-up, 150 had data pending, 25 were clinically non-evaluable, and 2 had an assay / pipeline QC failure, so these were excluded from the analysis. Thus, 161 patients were included in the analysis, of which 86 received a new TNF inhibitor (of which 49 were determined to respond and 37 were determined to not respond), 37 received an T-cell costimulatory blocker (of which 19 were determined to respond and 18 were determined to not respond), and 38 received a JAK inhibitor (of which 25 were determined to respond and 13 were determined to not respond).

[0025] FIG. 6 depicts aspects of treatment and enrollment for subjects from the study of Example 1. Panel A shows a Sankey plot showing that the study of Example 1 has enrolled b / ts-DMARD naive and experienced patients with diverse treatment histories. The plot illustrates prior and current on-study lines of therapy for Example 1 study participants who initiated a new b / tsDMARD (n=663). Node sizes represent the number of participants per b / tsDMARD class, while connecting lines show cycling and switching patterns. For the 1st line of therapy, participant counts are: TNFi (n=492), Anti T-cell (n=71), JAKi (n=67), Anti IL-6R (n=23), and Anti B-cell (n=10). Participants naive to b / tsDMARDs lack transition lines. Those starting their 2nd line at enrollment have transitions linking their 1st and 2nd lines of therapy, while those starting their 3rd line show transitions linking 1st, 2nd, and 3rd lines of therapy. Participants starting 4th line or greater show transitions linking prior lines of therapy. Panel B depicts cumulative enrollment and geographic distribution for participants in the study of Example 1. Shown are a geographic site distribution map (top) showing the geographic location of study centers and a stacked area chart (bottom) showing the number of participants in the study enrolled over time. At the time of the analysis in Example 1, the study had enrolled participants across 22 states.

[0026] FIG. 7 illustrates aspects of data collection for the study of Example 1. Panel A depicts a workflow of how blood samples collected according to the study of Example 1 were processed: (a) cell-free DNA (cfDNA) was extracted from ImL of blood plasma using a proprietary chromatin capture workflow; (b) fragment libraries were prepared, (c) next-WSGR Docket No. 55066-711.601generation sequencing was performed on the libraries, (d) fragment sequences were aligned to the human genome and deconvoluted into cfDNAnuc and cfDNAac fragments; and (e) fragment sequences were quantified across an atlas of -895,000 features comprising exons (-285K), promoters (-37K), and transcription factor binding sites (-610K). Panel B depicts an illustration showing the origin of different cfDNA populations in the body: unlike solid tissue biopsies, cfDNA released into the bloodstream upon cell death provides a non-invasive and dynamic source of molecular information and comprises distinct cfDNA populations such as nucleosomal fragments (cfDNAnuc) and active chromatin fragments (cfDNAac); cfDNAac fragments in the blood enable the assessment of gene regulation and expression in the absence of mRNA.

[0027] FIG. 8 depicts distribution of features from the sequencing analysis of Example 1 associated with response to JAK inhibitors (“JAKi”), T-cell costimulatory blockers (“anti T-cell”), or TNFa inhibitors (“TNFi”). Starting from the cfDNA fragments quantified across exons, promoters, and transcription factor binding sites, empirical Bayes moderated t-statistics and corresponding p-values were used to identify features exhibiting differential signals associated with therapy responses across various drug classes. Shown are Venn diagrams showing overlaps of top ranking differentially regulated active chromatin features — exon, promoter, tissue-specific TFBS — associated with TNFi, JAKi, and T-cell costimulatory blocker response. Top genes representing these features are shown in the cell-type abundance phenotype (CTAP) analysis Tables 2, 3, and 4.

[0028] FIG. 9 depicts representative genetic features out of 1.6K features that were identified in a linear mixed-effects model analysis of responder status and drug class-specific effects that were identified to have significant contextual or therapy-specific response effects (adj p-value < 0.01, p-values having been corrected for multiple comparisons using the Benjamini -Hochberg procedure). Shown are box-and-whisker plots of normalized cfDNAac score vs non-responders / responders for the JAKi, anti T-cell, and TNFi categories. Panel A shows an example of a JAKi-specific feature, a promoter-associated region of RAPGEF4 (“p@RAPGEF4”), which is decreased between non-responders and responders in JAKi-treated subjects. Panel B shows an example of an anti T-cell-specific feature, a promoter-associated region of MARCH1 (“p@MARCHl”), which is decreased between non-responders and responders in anti T-cell treated subjects. Panel C shows an example of a TNFi and JAKi specific feature, exon 5 of PDZD8 (“PDZD8_Exon5”), which is increased from non-responders to responders in JAKi-treated subjects. Panel D shows an example of a feature specific to TNFi, JAKi, and T-cell costimulatory blockers, a promoter-associated region of PAX9 (“p@PAX9”), which is decreased from non-responders to responders in subjects treated with TNFi, JAKi, and T-cell costimulatory blockers.WSGR Docket No. 55066-711.601

[0029] FIG. 10 shows study design and statistical aspects of the study in Example 2. Panel A shows a study design schematic for the study described in Example 2. The study was designed to identify markers of subsequent response to TNFa inhibitors (“TNFi”) or Janus kinase inhibitors (“JAKi”) in patients with an autoimmune disease (e.g. Rheumatoid Arthritis). A total of 48 RA participants undergoing a therapy switch to either TNFi (n=19) or JAKi (n=29) were selected from the CorEvitas BIO- 100 registry. The participants, aged 20-80 years (median=57), comprised 89.5% females and 10.5% males, with 89.6% identifying as white. Samples were selected based on equivalent numbers of responders and non-responder. 21 patients were classified as responders while 27 as non-responders based on their Clinical Disease Activity Index (CD Al) scores collected at baseline and 6 months after therapy initiation. Panel B shows plots of CD Al scores at baseline (left), CD Al scores at follow-up (middle), and change in CD Al scores (“ACDAI”) from baseline to follow-up (right). Panel C shows a scatter plot that has been customized for clinical outcomes analysis illustrating changes in CD Al scores over the course of the study in Example 2. CD Al is a scheme that allows treatment decisions to be based entirely on clinical criteria. Panel C depicts a scatter plot of changes in CD Al scores (“ACDAI”) for individual subjects of the study in Example 2 versus baseline CD Al score; as shown in the graph, the study encompassed subjects with a wide range of CD Al score that have a wide variety of responses to treatment.

[0030] FIG. 11 depicts a workflow of how blood samples collected according to the study of Example 2 were processed. Cell-free DNA (cfDNA) was extracted from 1 mL of plasma from subjects using a proprietary chromatin capture workflow, followed by library preparation and next-generation sequencing, after which significant features associated with drug response were used to train a machine-learning algorithm to predict drug class-specific therapy responses.

[0031] FIG. 12 shows aspects of a model that classifies patients likely to respond to TNFi per the study of Example 2 and associated performance. Panel A shows a heatmap of the top 57 features with the most discriminatory power for TNFi response (y-axis) vs. individual patients (x-axis) in the study of Example 2. The heatmap displays features identified through hybrid unsupervised clustering during feature selection. Upper tracks denote response status, disease activity, and change in CD Al from baseline to 24 weeks post-therapy initiation. Panel B shows performance of a machine learning model used to identify relevant variables to construct a predictive model using epigenetic data for patients switching therapies to TNFi and JAKi drug classes. Panel C shows a Receiver Operating Characteristic (ROC) curve of the performance of the classifier which achieved 80% sensitivity and 85% specificity for the dataset; the workflow incorporated 10,000 classification trees and a 5-repeated 5-fold cross-validation.WSGR Docket No. 55066-711.601

[0032] FIG. 13 demonstrates that classifier features from Example 2 predicting response are linked to RA pathobiology. Panel A shows a dot plot of the results of a WebCSEA (bioinfo.uth.edu / webcsea / ) cell-type specific enrichment analysis performed on the feature set of FIG. 9. The plot shows enrichment across top 20 general cell types predicted by the CSEA with raw and Bonferroni -corrected p-values. Bonferroni-corrected significance (p = 3.69 x 1 O’5) is indicated by the dashed horizontal line, and nominal significance (p = 1 x 1 O’3) by the solid horizontal line. X-axis represents cell types and y-axis represents the significance level. Panel B shows a diagram of joint synovium with cell types identified in enrichment analysis shown.

[0033] FIG. 14 shows features that preliminarily classify patients likely to respond to JAKi per the study of Example 2. Shown is a heatmap of the top 39 features with the most discriminatory power for JAKi response (y-axis) vs. individual patients (x-axis) in the study of Example 2. These features were used in a machine learning model which demonstrated a preliminary sensitivity of 80% with a specificity of 83% for JAKi response.

[0034] FIG. 15 shows an example computer system 1501 configured to implement methods according to the disclosure, comprising a CPU 1505, memory 1510, electronic storage 1515, communications interface 1520, and peripheral devices 1525 connected to a network 1530 and a processing system 1535.

[0035] FIG. 16 depicts a flow chart illustrating one example of training of a machine learning model as described herein.

[0036] FIG. 17 depicts a flow chart describing patient cohorts in Example 3.

[0037] FIG. 18 depicts a workflow for a repeated 5-Fold Nested Consensus Cross Validationbased feature selection as described in Example 3.DETAILED DESCRIPTIONOverview

[0038] Rheumatoid arthritis (RA) is documented as the most common chronic systemic autoimmune disease; however, despite its prevalence, there evidently remains a great deal of disease variability due to inconsistent responses of RA patients to available treatments, leading to unpredictable treatment odysseys for RA patients. Conventional synthetic DMARDs like methotrexate remain first-line treatment for RA; however, as few as about 73% of DMARD-naive patients reach an ACR20 response (which is considered roughly equivalent to a CDAI change of about 13.9) in response to these drugs by 24 weeks of treatment. TNFa inhibitors such as adalimumab, etanercept, infliximab, golimumab, or certolizumab are often used as first-line biologies when csDMARDs fail; however, as few as about 59% of patients reach an ACR20 response by 24 weeks when treated with such TNFa inhibitors. T-cell costimulatory blockersWSGR Docket No. 55066-711.601such as abatacept are often chosen in situations where TNFa inhibitors are not effective or contraindicated; however, as few as about 62% of patients reach an ACR20 when treated with these agents by 24 weeks. JAK inhibitors like tofacitinib, baricitinib, upadacitinib, and filgotinib are often chosen when a patient has shown an inadequate response to a biologic DMARD such as a TNFa inhibitor; however, as few as about 65% of patients reach an ACR20 response by 24 weeks in response to such synthetic molecules (see e.g. Konzett et al. Ann Rheum Dis. 2024 Jan 2;83(l):58-64 and Schiff et al. “AB0317 Association of ACR clinical responses with CD Al (clinical disease activity index) and RAPID3 (routine assessment of patient index data 3) indices of disease activity in rheumatoid arthritis patients treated with certolizumab pegol plus methotrexate.” Ann. Rheum. Dis. 71(supplement 3) (2012): 655. doi: 10.1136 / annrheumdis-2012-eular.317, both of which are incorporated by reference herein in their entireties).

[0039] Accordingly, there is a need for improved methods and systems that predict the responses (rather than non-responses) of autoimmune disease patients to the multiple classes of DMARDs (e.g. TNFa inhibitors, T-cell costimulatory blockers, or Janus Kinase inhibitors) commonly used to treat autoimmune diseases, as the response of individual patients to such DMARDs can be variable. Particularly, there is a need for non-invasive tests suitable for repeat administration. In some aspects, the methods and systems presented herein address this need via the use of blood-derived cfDNA (particularly, non-canonical cfDNA of increased size) to detect loci associated with response to particular DMARDs in patients with autoimmune diseases.

[0040] There is also a need for methods and systems that improve the detection of epigenetic signals from non-invasive samples (e.g. cfNA from blood). One specific example is that there is a need for methods and systems that are sensitive to length-based signals in cfNA beyond nucleosomal signals that appear at predictable size ranges. In some cases, said methods are capable of detecting genes and loci associated with physiological phenotypes (e.g. autoimmune disease response to an immunosuppressant) with improved accuracy, sensitivity, versus other methods, or detecting genes and loci associated with such length-based signals that are not apparent via other detection methods.

[0041] FIG. 1 shows an example workflow or system utilizing a therapeutic assignment classifier as described herein suitable for predicting a patient response to a DMARD (e.g. a TNFa inhibitor, a T-cell costimulatory blocker, or a JAK inhibitor). A sample 105 can be obtained from a subject 101 with an autoimmune disease and subjected to processing procedure 110 to isolate cfDNA fragments (which can include selective enrichment of cfDNAac fragments). The cfDNA can be subjected to an evaluation procedure 115 to identify cfDNA fragment overlap (or counts thereof) with particular genomic regions (which can includeWSGR Docket No. 55066-711.601promoters, enhancers, exons, transcription factor binding sites, insulators). The cfDNA genomic region information can be used to compile a feature vector 125 which represents predetermined features to be input into a machine learning model 125, which machine learning model has been trained on feature information for autoimmune patients that respond and do not respond to a particular DMARD. The machine learning model 125 outputs a classification or score 130 which relates to whether the subject will respond or not respond to the DMARD.

[0042] As part of the workflow, in some instances, the sample can be processed to obtain cfDNAac fragments. The processing can involve a procedure according to FIG. 2 panel A, wherein plasma samples subjected to a lysis procedure 205 are incubated with nanoparticles or beads under conditions configured to selectively bind 210 and (after separation and wash procedures 215 and 220) to selectively elute 225 a non-canonical or “active chromatin” fraction of cfDNA. An example trace showing the output of such a cfDNAac enrichment procedure is shown in FIG. 2 panel B.

[0043] Referring again to FIG. 1, in some cases after or concurrent with the evaluation procedure 115 (which can be performed to identify cfDNA fragment overlap with particular genomic regions which can include promoters, enhancers, exons, transcription factor binding sites, and insulators), an optional approximation operation of an individual or a subset of genomic regions can be performed prior to machine learning procedures. Without wishing to be bound by theory, it is hypothesized that while nucleosomal epigenetic length signals (“homogenous signals”) can occur at predictable length bands due to the winding of DNA around nucleosomes (e.g. -147 bp for mononucleosomes, -320 bp for dinucleosomes), non-nucleosomal epigenetic length signals due to processes such as active transcriptional complex assembly on DNA can occur at varying lengths greater than 200 base pairs (“heterogenous signals”). According to this theory, workflows that capture overall length distributions (e.g. functional analysis-type approaches) can be better for capturing heterogenous non-nucleosomal signals. In some embodiments, the approximation operation may involve a length-based analysis. The length-based analysis may be implemented on individual genomic regions or may be generic to genomic regions (e.g. it may encompass reads derived from sequencing without assigning them to genomic regions or bins). In one example, the approximation operation can comprise a fitting procedure to fit a multivariate formulation to a distribution of fragment counts versus length for a genomic region, as depicted in FIG. 3 panel A, in which a histogram or bins of fragment count versus length for a genomic region 301 can be subjected to an optional preprocessing procedure (e.g. a density estimation procedure) 305 followed by a fit to a multivariate formulation 310, which can be e.g. by a maximum-likelihood algorithm. The multivariable fit can result in a set of fit variable values of the multivariable formulation, of whichWSGR Docket No. 55066-711.601individual variable values or subsets of variable values can be input into a feature vector 320 for future processing by machine learning models. Alternatively or additionally, the approximation operation can involve a transform, such as wavelet transform, as depicted in FIG. 3 panel B, in which in which a histogram or bins of fragment count versus length for a genomic region 301 can be subjected to an optional preprocessing procedure (e.g. a density estimation procedure) 305 followed by a wavelet transform 312 (e.g. a continuous wavelet transform with a morlet wavelet of defined parameters). The wavelet transform can result in a set of wavelet coefficients 317, of which individual coefficients or subsets of coefficients can be input into a feature vector 320 for future processing by machine learning models.

[0044] The machine learning model 125 can be a model suited to binary classifications of samples between responder and non-responder for particular immunosuppressants (e.g. TNFa inhibitors, T-cell costimulatory blockers, or JAK inhibitors). In some cases, the machine learning model comprises multiple machine learning models (e.g. a stacked classifier comprising multiple classification models, each determining responsiveness to an individual immunosuppressant). In some cases, the machine learning model can be a tree-based ensemble model (e.g. XGBoost, LightGBM, or Random forest). In some cases, the machine learning model can be a regularized linear model (e.g. LASSO, elastic net, or ridge regression applied as regularized logistic regression models).Definitions

[0045] While various embodiments of the invention have been shown and described herein, it will be obvious to those skilled in the art that such embodiments are provided by way of example only. Numerous variations, changes, and substitutions may occur to those skilled in the art without departing from the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed.

[0046] The practice of some methods disclosed herein employ, unless otherwise indicated, techniques of immunology, biochemistry, chemistry, molecular biology, microbiology, cell biology, genomics and recombinant DNA. See for example Sambrook and Green, Molecular Cloning: A Laboratory Manual, 4thEdition (2012); the series Current Protocols in Molecular Biology (F. M. Ausubel, et al. eds.); the series Methods In Enzymology (Academic Press, Inc.), PCR 2: A Practical Approach (M.J. MacPherson, B.D. Hames and G.R. Taylor eds. (1995)), Harlow and Lane, eds. (1988) Antibodies, A Laboratory Manual, and Culture of Animal Cells: A Manual of Basic Technique and Specialized Applications, 6thEdition (R.I. Freshney, ed. (2010)) (which is entirely incorporated by reference herein).WSGR Docket No. 55066-711.601

[0047] The term “about” or “approximately” generally signifies within an acceptable error range for the particular value as determined by the application, which will depend in part on how the value is measured or determined, i.e., the limitations of the measurement system. For example, “about” can mean within one or more than one standard deviation. Alternatively, “about” can mean a range of up to 20%, up to 15%, up to 10%, up to 5%, or up to 1% of a given value.

[0048] The term “or”, as used herein, is intended to be an inclusive “or”.

[0049] The term “nucleic acid” generally refers to deoxyribonucleic acid (DNA), ribonucleic acid (RNA) or any hybrid or fragment thereof. The nucleic acid in the sample can be a cell-free nucleic acid. A sample can be a liquid sample (e.g., blood or interstitial fluid) or a solid sample (e.g., a cell or tissue sample). In some examples, the sample can be obtained from a cell-free bodily fluid, such as plasma. In such instance, the sample may include cell-free DNA or cell-free RNA. In some examples, the majority of DNA in a biological sample that may be enriched for cfDNA (e.g., a plasma sample obtained via a centrifugation protocol) can be cell-free (e.g., greater than 50%, 60%, 70%, 80%, 90%, 95%, or 99% of the DNA can be cell-free). In some examples, nucleic acid can be derived from circulating tumor cells or circulating fetal cells. Throughout the disclosure the use of cell-free nucleic acid or cfNA may generally be used interchangeably with nucleic acid or NA. Examples of nucleic acids include, but are not limited to, deoxyribonucleic acid (DNA), genomic DNA, plasmid DNA, complementary DNA (cDNA), cell-free (e.g., non-encapsulated) DNA (cfDNA), circulating tumor DNA (ctDNA), nucleosomal DNA, chromatosomal DNA, and mitochondrial DNA (miDNA). In some embodiments, a nucleic acid analyzed as described herein can be at least about 10 and at most about 15,000 nucleotides (or base pairs) in length (wherein sequence reads or ssDNA are denoted in nucleotides and corresponding double-stranded DNA can be denoted in base pairs). In some embodiments, a nucleic acid analyzed as described herein can be at least 100 and at most about 15,000 nucleotides (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 120, 150, 165, 200, 210, 211, 230, 250, 281, 300, 334, 335, 336, 350, 360, 399, 400, 451, 460, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1100, 1150, 1200, 1250, 1300, 1350, 1400, 1450, or 1500 nucleotides (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be at least about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 120, 150, 165, 200, 210, 211, 230, 250, 281, 300, 334, 335, 336, 350, 360, 399, 400, 451, 460, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1100, 1150, 1200, 1250, 1300, 1350, 1400, or 1450 nucleotides (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be at most about 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35,WSGR Docket No. 55066-711.60140, 45, 50, 60, 70, 80, 90, 100, 120, 150, 165, 200, 210, 211, 230, 250, 281, 300, 334, 335, 336, 350, 360, 399, 400, 451, 460, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1100, 1150, 1200, 1250, 1300, 1350, 1400, 1450, or 1500 (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be at least about 81, 141, 200, 211, 281, 334, 336, 400, or 460 nucleotides (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be at most about 141, 210, 280, 335, 336, 399, or 459 nucleotides (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be at most about 141, 210, 280, 335, 336, 399, or 459 nucleotides (or base pairs) in length and less than or equal to about 1500, 1400, 1300, 1200, 1100, or 1100 nucleotides (or base pairs) in length. In some embodiments, a nucleic acid analyzed as described herein may be about 81-141, 81-336, 100-280, 281-400, 401-1500, 100-210, 211-334, 335-399, or 400-1500 nucleotides (or base pairs) in length. In some cases, a nucleic acid analyzed as described herein may comprise or represent nucleosomal DNA. In some cases, a nucleic acid analyzed as described herein may comprise or represent non-nucleosomal DNA. In some cases, nucleosomal DNA may be about 100-210, 335-399, 100-280, or 401-1500 nucleotides (or base pairs) in length. In some cases, non-nucleosomal DNA may be 211-334, 400-1500, or 281-400 nucleotides (or base pairs) in length.

[0050] As used herein, the term “sequencing” generally refers to any processes (e.g. chemical, enzymatic, biochemical) that may be used to determine the order of biological macromolecules such as nucleic acids or proteins. For example, sequencing data can include all or a portion of the nucleotide bases in a nucleic acid molecule such as an mRNA transcript, a cfDNA fragment, a DNA fragment, or a genomic locus.

[0051] As used herein, the term “sequence reads” or “reads” refers to nucleotide sequences produced by a nucleic acid sequencing process. Reads can be generated from one end of nucleic acid fragments (“single-end reads”) or from both ends of nucleic acid fragments (e.g., paired-end reads, paired reads, double-end reads). In some embodiments, the sequence reads are of a mean, median or average length of about 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 120, 150, 165, 200, 210, 211, 230, 250, 281, 300, 334, 335, 336, 350, 360, 399, 400, 451, 460, 500, 550, 600, 650, 700, 750, 800, 850, 900, 950, 1000, 1100, 1150, 1200, 1250, 1300, 1350, 1400, 1450, or 1500 nucleotides (or bases) in length.

[0052] The term "subject" generally refers to a biological entity containing genetic materials. Examples of a biological entity include a plant or animal. In some examples, a subject is a mammal, e.g., a human that can be male or female. Such a human can be of various ages, e.g., from 1 day to about 1 year old, about 1 year old to about 3 years old, about 3 years old to about 12 years old, about 13 years old to about 19 years old, about 20 years old to about 40 years old,WSGR Docket No. 55066-711.601about 40 years old to about 65 years old, or over 65 years old. In various examples, a subject can be healthy or normal, abnormal, or diagnosed or suspected of being at a risk for a disease. In some cases, the disease is an autoimmune disease.

[0053] The term “biological sample” (or just “sample”) generally refers to any substance obtained from a subject. A sample may contain or be presumed to contain analytes, for example those described herein (nucleic acids, polypeptides, carbohydrates, or metabolites) from a subject. In some respects, a sample can include cells or cell-free material obtained in vivo, cultured in vitro, or processed in situ, as well as lineages including pedigree and phytogeny. Examples of a liquid sample (e.g., a bodily fluid) include whole blood, buffy coat from blood (which can include lymphocytes), urine, saliva, cerebrospinal fluid, plasma, serum, ascites, sputum, sweat, tears, buccal sample, cavity rinse, or organ rinse. In some cases, the liquid is a cell-free liquid that is an essentially cell-free liquid sample or comprises cell-free nucleic acid, e.g., cell-free DNA (cfDNA). For methods described herein, in some embodiments the methods involve directly obtaining a sample from a subject (e.g. via venipuncture, capillary collection, biopsy, etc.). For methods described herein, in some embodiments the methods involve obtaining a sample which was originally obtained from a subject by another party.

[0054] The terms "blood plasma" or "plasma", as used herein, generally refer to a straw-col ored / pale-yellow liquid component of blood that holds the blood cells in whole blood in suspension. Blood plasma can make up about 55% of total blood by volume. It can comprise up to about 93% by volume water, and can contain dissolved proteins including albumins, immunoglobulins, and fibrinogen, glucose, clotting factors, electrolytes (Na+, Ca2+, Mg2+, HCO3, CF, etc.), hormones and carbon dioxide. Blood serum generally refers to blood plasma without fibrinogen or the other clotting factors (i.e., whole blood minus both the cells and the clotting factors).

[0055] The term “input features” generally refers to variables that are used by a machine learning model to predict an output classification (label) of a sample such as e.g., a condition, or suggested treatments. Values of the variables can be determined for a sample and used to determine a classification. Examples of input features of genetic data include: aligned variables that relate to alignment of sequence data (e.g., sequence reads) to a genome, sequence (or fragment) read lengths, sequence (or fragment) read counts, GC content, or methylation.

[0056] The term “feature vector” generally refers to a set of numerical values representing features of data to be input to a machine learning model. The feature vector can be created as any data structure suitable for input to a machine learning model and can include one or more input features.WSGR Docket No. 55066-711.601

[0057] The term ‘'machine learning model” generally refers to a collection of parameters and functions, wherein the parameters are trained on a set of training samples, which can be used to make predictions or classifications. The parameters and functions may be a collection of linear algebra operations, non-linear algebra operations, and tensor algebra operations. The parameters and functions may include statistical functions, tests, and probability models. The training samples can correspond to samples having measured properties of the sample (e.g., genomic data and other subject data, such as images or health records), as well as predetermined classifications / labels (e.g., phenotypes or treatments) for the subject. The model can learn from the training samples in a training process that optimizes the parameters (and potentially the functions) to provide an optimal quality metric (e.g., accuracy) for classifying new samples. The training function can include expectation maximization, maximum likelihood, Bayesian parameter estimation methods such as markov chain monte carlo, gibbs sampling, hamiltonian monte carlo, and variational inference, or gradient based methods such as stochastic gradient descent and the Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm. Example parameters include weights (e.g., vector or matrix transformations) that multiply values, e.g., in regression or neural networks, families of probability distributions, or a loss, cost or objective function that assigns scores and guides model training. Example parameters include weights that multiple values, e.g., in regression or neural networks. A model can include multiple submodels, which may be different layers of a model or independent model, which may have a different structural form, e.g., a combination of a neural network and a support vector machine (SVM). Examples of machine learning models include deep learning models, neural networks (e.g., deep learning neural networks), tree-based ensemble models, regularized linear models, kernel-based regressions, adaptive basis regression or classification, Bayesian methods, ensemble methods, logistic regression and extensions, Gaussian processes, support vector machines (SVMs), probabilistic models, and probabilistic graphical models. A machine learning model can further include feature engineering (e.g., gathering of features into a data structure such as a 1, 2, or greater dimensional vector) and feature representation (e.g., processing of data structure of features into transformed features to use in training for inference of a classification).

[0058] The term “training sample” generally refers to samples for which a classification may be predetermined. Training samples can be used to train the model. The values of the features for a sample can form an input vector, e.g., a training vector for a training sample. Each element of a training vector (or other input vector) can correspond to a feature that includes one or more variables. For example, an element of a training vector can correspond to a matrix. The value of the label of a sample can form a vector that contains strings, numbers, bytecode, or any collection of the aforementioned datatypes in any size, dimension, or combination.WSGR Docket No. 55066-711.601

[0059] As used herein, the terms “autoimmune disorder”, “autoimmune disease”, or “autoimmune condition” generally refer to any disorder, disease, or condition in which the body produces an immunogenic (e.g., immune system) response to some constituent of its own tissue. In other words, the immune system loses its ability to recognize some tissue or system within the body as “self’ and targets and attacks it as if it were foreign. Autoimmune disorders, diseases, or conditions can be classified into those in which predominantly one organ is affected (e.g., hemolytic anemia and anti-immune thyroiditis), and those in which the autoimmune disorder, disease, or condition process is diffused through many tissues (e.g., Systemic Lupus Erythematosus). Examples of autoimmune disorders, diseases, or conditions include, but are not limited to, rheumatoid arthritis, multiple sclerosis, lupus erythematosis, myasthenia gravis, scleroderma, Crohn's Disease, ulcerative colitis, Hashimoto's Disease, Graves' Disease, Sjogren's Syndrome, poly endocrine failure, vitiligo, peripheral neuropathy, autoimmnune polyglandular syndrome type I, acute glomerulonephritis, Addison's Disease, adult-onset idiopathic hypoparathyroidism (AOIH), Alopecia Totalis, Amyotrophic Lateral Sclerosis, Ankylosing Spondylitis, Psoriatic Arthritis, autoimmune aplastic anemia, autoimmune hemolytic anemia, Behcet's Disease, Celiac Disease, chronic active hepatitis, CREST syndrome, dermatomyositis, dilated cardiomyopathy, eosinophilia-myalgia syndrome, epidermolisis bullosa acquisita (EBA), giant cell arteritis, Goodpasture's syndrome, Guillain-Barre syndrome, hemochromatosis, Henoch-Schonlein purpura, idiopathic IgA nephropathy, insulin-dependent diabetes mellitus (IDDM), Juvenile Rheumatoid Arthritis, Lambert-Eaton syndrome, linear IgA dermatosis, myocarditis, narcolepsy, necrotizing vasculitis, Neonatal Lupus Syndrome (NLE), nephrotic syndrome, pemphigoid, pemphigus, polymyositis, primary sclerosing cholangitis, psoriasis, rapidly-progressive glomerulonephritis (RPGN), Reiter's Syndrome, stiff-man syndrome, silent thyroiditis, or postpartum thyroiditis. The autoimmune disease can comprise Rheumatoid Arthritis, Systemic Lupus Erythematosus, Sjogren's Syndrome, Graves’ Disease, Hashimoto’s Disease, Celiac Disease, Psoriatic Arthritis, and Ankylosing Spondylitis.

[0060] The term “rheumatoid arthritis” (RA) generally refers to a chronic autoimmune disease that mostly affects joints. RA occurs when the immune system, which normally helps protect the body from infection and disease, attacks its own tissues. The disease causes pain, swelling, stiffness, and loss of function in joints. Examples of RA diagnosis and treatment are described in e.g. "Handout on Health: Rheumatoid Arthritis". National Institute of Arthritis and Musculoskeletal and Skin Diseases. August 2014. (www.niams.nih.gov / health-topics / rheumatoid-arthritis, which is incorporated by reference herein for all purposes) and Wasserman. Am Fam Physician. 2011 Dec 1;84(11): 1245-52 (which is incorporated by reference herein for all purposes). Symptoms of RA can include: (1) pain, swelling, stiffness,WSGR Docket No. 55066-711.601and tenderness in more than one joint; (2) stiffness, especially in the morning or after sitting for long periods; (3) pain and stiffness in the same joints on both sides of the body; (4) fatigue; (5) weakness; (6) fever; (7) elevated erythrocyte sedimentation rate (ESR); and elevated C-reactive protein (CRP). In some cases, a patient with RA can be positive for anti-rheumatic factor (RF) antibodies (e.g. IgM, IgG, or IgA antibodies) or anti-cyclic citrullinated peptide (CCP) antibodies. In some cases, a patient with RA can be not positive for anti-rheumatic factor (RF) or anti-cyclic citrullinated peptide (CCP) antibodies.

[0061] The term “CD Al” generally refers to the Clinical Disease Activity Index (CD Al), which is a comprehensive measure of RA disease activity that integrates swollen joint counts, tender joint counts, patient global assessment of disease activity, and physician global assessment of disease activity (as described in e.g. Takanashi et al., CD Al and DAS28 in the management of rheumatoid arthritis in clinical practice, Ann Rheum Dis. 2020, which is incorporated by reference herein for all purposes). (CDAI=SJC+TJC+PGA+EGA).

[0062] The term “DAS” generally refers to the Disease Activity Score, a measure of the activity of RA in a subject (as described e.g. in D. van der Heijde et al., Ann. Rheum. Dis. 1990, 49(11): 916-920, which is incorporated by reference herein for all purposes). The “DAS28” generally involves the evaluation of 28 specific joints (e.g. bilateral proximal interphalangeal joints (10 joints), metacarpophalangeal joints (10), wrists (2), elbows (2), shoulders (2) and knees (2)). A DAS28 can be calculated for an RA subject according to the standard as outlined at the das-score.nl website, maintained by the Department of Rheumatology of the University Medical Centre in Nijmegen, the Netherlands. The number of swollen joints, or swollen joint count out of a total of 28 (SJC28), and tender joints, or tender joint count out of a total of 28 (TJC28) in each subject can be assessed.

[0063] The term “DAS-28 CRP” generally refers to a DAS28 assessment calculated using CRP measurements added to the score. CRP is generally produced in the liver. As there is normally little or no CRP circulating in an individual's blood serum and CRP is generally present in the body during episodes of acute inflammation or infection, a high or increasing amount of CRP in blood serum can be associated with acute infection or inflammation. The DAS28-CRP can be calculated according to either of the formulas below, with or without the GH factor, where “CRP” represents the amount of this protein present in a subject's blood serum in mg / L, “sqrt” represents the square root, and “In” represents the natural logarithm:DAS28-CRP with GH (or DAS28-CRP4)=(0.56*sqrt(TJC28)+0.28*sqrt(SJC28)+0.36*ln(CRP+l))+(0.014*GH)+0.96; or, (a) DAS28-CRP without GH (or DAS28- CRP3)=(0.56*sqrt(TJC28)+0.28*sqrt(SJC28)+0.36*ln(CRP+l))*1.10+1.15.WSGR Docket No. 55066-711.601

[0064] The term “EULAR response” generally refers to a classification system for evaluating a patient's response to treatment for inflammatory arthritis, like rheumatoid arthritis (RA). It categorizes patients into "good," "moderate," or "non-response" based on both the magnitude of their improvement (change in disease activity) and their absolute level of disease activity at the end of a treatment period, calculated using the Disease Activity Score (DAS). A “good” EULAR response can be defined as a significant decrease in the DAS (e.g., >1.2) AND a low level of disease activity (e.g., a “changed” DAS28 score < 2.4). A “moderate” EULAR response can be defined as a decrease in DAS between 0.6 and 1.2) where the “changed” DAS is still > 3.7, or a smaller decrease in DAS where the “changed” DAS is low, or a significant decrease where the “changed” DAS is not low. A EULAR non-response can be defined as a minimal change in DAS (e.g., decrease < 0.6) or a decrease between 0.6 and 1.2 where the “changed” DAS is still high (e.g., DAS > 3.7).

[0065] The term “cyclic citrullinated peptide antibodies” generally refers to autoantibodies that are directed against peptides and proteins that are citrullinated. Such autoantibodies are often present in subjects with rheumatoid arthritis and can be used as part of diagnostic criteria for RA.

[0066] The term “C-reactive protein” or “CRP” generally refers to refers to a circular pentameric protein found on the first chromosome, exemplified by Swiss Prot Accession No. NP 000558. It is generally documented as an acute-phase protein synthesized predominantly in the liver, as well as in other cells locally such as endothelial cells, in response to inflammation.

[0067] The term “immunosuppressant” generally refers to a compound or drug which possesses immune response inhibitory activity. Immunosuppressants may comprise DMARDs, mTOR inhibitors, azathioprine, or mycophenolate mofetil.

[0068] The term “disease-modifying antirheumatic drug” or “DMARD” generally refers to a drug in a class of drugs used to treat inflammatory or autoimmune diseases toslow the progression of the disease. DMARDs generally include synthetic diseasemodifying antirheumatic drugs (sDMARDs) and biological diseasemodifying antirheumatic drugs. sDMARDs can comprise conventional synthetic diseasemodifying antirheumatic drugs (csDMARDs) and targeted synthetic diseasemodifying antirheumatic drugs (tsDMARDs). Targeted synthetic DMARDs can comprise Janus kinase inhibitors, SYK inhibitors, IRAKI inhibitors, IRAK4 inhibitors, RIP IK inhibitors and apremilast. Biological disease-modifying antirheumatic drugs (bDMARDs) can comprise original and biological disease-modifying antirheumatic drugs (boDMARDs) and biosimilar disease-modifying antirheumatic drug (bsDMARDs). Examples of conventional syntheticWSGR Docket No. 55066-711.601disease-modifying antirheumatic drugs (csDMARDs) can include methotrexate, azathioprine, cyclophosphamide, cyclosporine, leflunomide, mycophenolate mofetil, sulfasalazine, hydroxychloroquine, hydroxycycloquine, minocycline, and gold salts.

[0069] The term “seropositive RA” generally refers to a common subtype of rheumatoid arthritis characterized by the presence of specific autoantibodies in the blood. In some cases, the specific autoantibodies may comprise anti-citrullinated protein antibodies (ACPAs) (see e.g. Lancet. 388 (10055): 2023-2038. doi: 10.1016 / 80140-6736(16)30173-8, which is incorporated by reference herein for all purposes) or anti-rheumatoid factor (RF) antibodies.

[0070] The term “seronegative RA” generally refers to a subtype of rheumatoid arthritis in which individuals affected by the disease do not test positive for certain autoantibodies commonly associated with seropositive RA. The autoantibodies are not present in detectable levels in seronegative rheumatoid arthritis, making the diagnosis more challenging and requiring consideration of other clinical criteria. Despite the lack of detectable autoantibodies, seronegative RA is still considered an autoimmune disease. The exact cause of seronegative RA has not been defined, but it is documented as resulting from an abnormal immune response that leads to chronic inflammation and joint damage. The diagnosis of seronegative RA can rely on a combination of clinical evaluation, physical examination, medical history, laboratory tests, and imaging studies. Both seronegative RA and non-RA subjects may experience joint pain, swelling, and stiffness. However, the pattern and distribution of joint involvement can provide important clues. In some embodiments, the small joints of the hands and feet can be commonly affected in seronegative RA subjects, often in a symmetrical manner (affecting both sides of the body). In some embodiments, larger joints and the presence of other systemic symptoms in subjects may also suggest RA. RA can be a chronic and progressive condition, so a history of recurrent or persistent joint symptoms may raise suspicion for seronegative RA.

[0071] The term “alcohol” generally refers to an organic compound that carries at least one hydroxyl functional group ( — OH) bound to a saturated carbon atom. Examples include mono, di, tri, tetra or pentaalcohols. Examples further include ethanol or isopropanol. Examples still further include methanol, ethanol, isopropanol, propanol, butanol, ethylene glycol, propylene glycol, butylene glycol, pentylene glycol, glycerol, tetritol, pentitol, 1,3 propanediol, and the like, or mixtures thereof.

[0072] The term “chaotropic agent” generally refers to a compound that disrupts hydrogen bonds between water molecules and disrupts the tertiary structure of biopolymers. In some embodiments, the chaotropic agent comprises a guanidinium ion (e.g. guanidinium chloride or guanidine thiocyanate), thiourea, or urea. In some embodiments, the chaotropic agent is a sulfur atom-free agent.WSGR Docket No. 55066-711.601

[0073] As used herein, a “cell” generally refers to a biological cell. A cell may be the basic structural, functional or biological unit of a living organism. A cell may originate from any organism having one or more cells. Some non-limiting examples include: a prokaryotic cell, eukaryotic cell, a bacterial cell, an archaeal cell, a cell of a single-cell eukaryotic organism, a protozoa cell, a cell from a plant, an algal cell, an animal cell, a cell from an invertebrate animal, a cell from a vertebrate animal (e.g., fish, amphibian, reptile, bird, mammal), or a cell from a mammal (e.g., a pig, a cow, a goat, a sheep, a rodent, a rat, a mouse, a non-human primate, a human, etc.).

[0074] The term “binding agent” generally refers to a molecule that has selective affinity for a biomolecule. Binding agents may comprise aptamers, antibodies, lectins and enzymes.

[0075] The term "antibody" generally includes polyclonal antiserum, monoclonal antibodies, fragments of antibodies (e.g. single chain or Fab fragments), and engineered derivatives carrying binding components of antibodies (e.g. single chain variable fragments or ScFvs)

[0076] The term "aptamer" generally includes affinity agents with selectivity for a specific predetermined molecule, which affinity agents are polymers of nucleic acids.

[0077] The term “chromatin" generally refers to nucleoprotein complexes with nucleic acid which can compact and organize great lengths of cellular genetic material to contain it within cells. The nucleosome, in which double-stranded DNA (dsDNA) is wound approximately twice around a core of conserved histone or histone-like proteins, can comprise a primary level of chromatin organization in the nucleus of eukaryotic cells. Higher-order chromatin organization can involve further compaction of nucleosomes around additional chromatin-associated proteins and can employ a variety of chromatin assembly factors. The basic structural unit of chromatin is the nucleosome: which can comprise a core octamer of histone proteins (e.g. two copies of each of H2A, H2B, H3 and H4) as well as a linker histone and about 180 base pairs of DNA.

[0078] The term “heterochromatin” or “inactive chromatin” generally refers to a subset of chromatin that is most densely compacted and is generally transcriptionally silent.

[0079] As used herein, “euchromatin” or “active chromatin” generally refers to more extended chromatin domains that are often transcriptionally active, accessible portions of the genome. In some embodiments, active chromatin can comprise cfNA fragments at least 200, 211, 281, 334, 336, 400, or 460 base pairs in length.

[0080] As used herein, the term “H3K4mel” generally refers to monomethylation of lysine 4 in the histone H3 protein. In some embodiments, H3K4mel is associated with active and primed enhancer elements when encountered in a region of cellular chromatin.WSGR Docket No. 55066-711.601

[0081] As used herein, the term “H3K4me2” generally refers to demethylation of lysine 4 in the histone H3 protein. In some embodiments, H3K4me3 is associated with active transcription of nearby genes when encountered in a region of cellular chromatin.

[0082] As used herein, the term “H3K4me3” generally refers to trimethylation of lysine 4 in the histone H3 protein. In some embodiments, H3K4me3 is associated with active transcription of nearby genes when encountered in a region of cellular chromatin.

[0083] As used herein, the term “H3K27ac2” or “H3K27ac” generally refers to acetylation of the lysine residue atN-terminal position 27 of the histone H3 protein. In some embodiments, H3K27ac is associated with increased transcription when encountered in a region of cellular chromatin.

[0084] As used herein, the term “H3K36me3” generally refers to trimethylation at the 36thlysine residue of the histone H3 protein. In some embodiments, H3K36me3 is associated with gene bodies when encountered in a region of cellular chromatin.

[0085] The methods in some embodiments may be conducted on a single sample from a subject, for instance, a blood, serum, plasma, urine, or tissue sample, or a sample obtained by a non-invasive, minimally invasive, or invasive procedure. Such a “single sample” generally denotes herein a sample that is obtained from the recipient at one time, such as during one blood draw or phlebotomy appointment or during one other diagnostic or medical appointment. Accordingly, the “single sample” is not required to be present in the same sample container but instead is merely drawn from the patient at the same time, during the context of one diagnostic or medical appointment.

[0086] In some embodiments, the sample is obtained from a non-invasive procedure, such as a throat swab, buccal swab, bronchial lavage, urine collection, skin or epidermal scraping, feces collection, menses collection, or semen collection. In other cases, a minimally invasive procedure may be used such as a blood draw, e.g., by venipuncture methods. In other cases, a sample may be obtained by an invasive procedure such as a biopsy, alveolar or pulmonary lavage, or needle aspiration.

[0087] In some embodiments, the sample is a blood, serum, or plasma sample. As used herein obtaining a sample includes obtaining a sample directly or indirectly. In some embodiments, the sample is taken from the subject by the same party (e.g. a testing laboratory) that subsequently acquires biomarker data from the sample. In some embodiments, the sample is received (e.g. by a testing laboratory) from another entity that collected it from the subject (e.g. a physician, nurse, phlebotomist, or medical caregiver). In some embodiments, the sample is taken from the subject by a medical professional under direction of a separate entity (e.g. a testing laboratory) and subsequently provided to said entity (e.g. the testing laboratory). InWSGR Docket No. 55066-711.601some embodiments, the sample is taken by the subject or the subject’s caregiver at home and subsequently provided to the party that acquires biomarker data from the sample (e.g. a testing laboratory). A variety of kits suitable for self or home collection of biological samples have been described commercially and in the literature such as e.g., US20170023446A1 and US4777964A.I. Sample AcquisitionSubjects

[0088] The term subject can include human or non-human animals. Thus, the methods described herein are applicable to both human and veterinary disease and animal models. In some cases, subjects are “patients,” e.g., living humans that are receiving medical care for a disease or condition (e.g. an autoimmune disease). These methods can be particularly useful for human subjects who have rheumatoid arthritis, although they can also be used for subjects who have been identified as having other autoimmune diseases. The subjects may be mammals or nonmammals. Preferably, the subject is a human but in some cases, the subject is a non-human mammal, such as a non-human primate (e.g., ape, monkey, chimpanzee), cat, dog, rabbit, goat, horse, cow, pig, rodent, mouse, SCID mouse, rat, guinea pig, or sheep. The subject may be male or female; and, in some cases, the subject may be an infant, child, adolescent, teenager or adult.

[0089] In some cases, the subject is a patient or other individual undergoing a treatment regimen, or being evaluated for a treatment regimen (e.g., immunosuppressive or DMARD therapy). In some cases, the subject is receiving a pharmacological treatment according to Table A. However, in some instances, the subject is not undergoing a treatment regimen (e.g. is treatment naive for autoimmune disease). In some embodiments, the subject is receiving a csDMARD. In some embodiments, the subject is receiving methotrexate. In some embodiments, the subject is receiving a bDMARD. In some embodiments, the subject is receiving a TNFa inhibitor. In some embodiments, the subject is receiving a T-cell costimulatory blocker. In some embodiments, the subject is receiving a tsDMARD. In some embodiments, the subject is receiving a JAK inhibitor.Table A: Example pharmacological treatments for subjects according to the disclosure.WSGR Docket No. 55066-711.601WSGR Docket No. 55066-711.601><WSGR Docket No. 55066-711.601WSGR Docket No. 55066-711.601

[0090] In some embodiments, the patient has been identified as having an autoimmune disease. The autoimmune disease can comprise rheumatoid arthritis, multiple sclerosis, lupus erythematosus, myasthenia gravis, scleroderma, Crohn's disease, ulcerative colitis, Hashimoto's disease, Graves' disease, Sjogren's syndrome, poly endocrine failure, vitiligo, peripheral neuropathy, autoimmune polyglandular syndrome type I, acute glomerulonephritis, Addison's disease, adult-onset idiopathic hypoparathyroidism (AOIH), alopecia totalis, amyotrophic lateral sclerosis, ankylosing spondylitis, psoriatic arthritis, autoimmune aplastic anemia, autoimmune hemolytic anemia, Behcet's disease, Celiac disease, chronic active hepatitis, CREST syndrome, dermatomyositis, dilated cardiomyopathy, eosinophilia-myalgia syndrome, epidermolysis bullosa acquisita (EBA), giant cell arteritis, Goodpasture's syndrome, Guillain-Barre syndrome, hemochromatosis, Henoch-Schonlein purpura, idiopathic IgA nephropathy, insulin-dependent diabetes mellitus (IDDM), juvenile rheumatoid arthritis, Lambert-Eaton syndrome, linear IgA dermatosis, myocarditis, narcolepsy, necrotizing vasculitis, neonatal lupus syndrome (NLE), nephrotic syndrome, pemphigoid, pemphigus, polymyositis, primary sclerosing cholangitis, psoriasis, rapidly-progressive glomerulonephritis (RPGN), Reiter's syndrome, stiff-man syndrome, silent thyroiditis, or postpartum thyroiditis. The autoimmune disease can comprise Rheumatoid Arthritis, Systemic Lupus Erythematosus, Sjogren's Syndrome, Graves’ Disease, Hashimoto’s Disease, Celiac Disease, Psoriatic Arthritis, or Ankylosing Spondylitis. TheWSGR Docket No. 55066-711.601autoimmune disease can comprise Rheumatoid Arthritis, Systemic Lupus Erythematosus, Sjogren's Syndrome, Graves’ Disease, or Hashimoto’s Disease. The autoimmune disease can comprise Rheumatoid Arthritis.

[0091] When the subject has Rheumatoid Arthritis, the disease severity of the subject can be variable. In some embodiments, the subject’s Clinical Disease Activity Index (CD Al) score can be at least 1, at least 5, at least 10, at least 15, at least 20, at least 25, at least 30, at least 35, at least 40, at least 45, at least 50, at least 55, or at least 60. In some embodiments, the subject’s Clinical Disease Activity Index (CD Al) score can be at most 1, at most 5, at most 10, at most 15, at most 20, at most 25, at most 30, at most 35, at most 40, at most 45, at most 50, at most 55, at most 60, at most 65, or at most 70. In some embodiments, the subject’s CD Al can be about 1-10, about 10-20, or about 20-70.

[0092] In various embodiments, the subjects suitable for methods described herein are patients who have been identified as having an autoimmune disease within 6 hours, 12 hours, 1 day, 2 days, 3 days, 4 days, 5 days, 10 days, 15 days, 20 days, 25 days, 1 month, 2 months, 3 months, 4 months, 5 months, 7 months, 9 months, 11 months, 1 year, 2 years, 4 years, 5 years, 10 years, 15 years, 20 years or longer.Samples

[0093] In some embodiments, the NA is extracted from blood, urine, stool, interstitial fluid, any subfraction of the aforementioned, or any combination thereof. In some embodiments, the NA is extracted from tissue. In some embodiments, the cfNA is extracted from blood, urine, stool, interstitial fluid, any subfraction of the aforementioned, or any combination thereof. In some embodiments, the NA is extracted from blood. In some embodiments, the NA is extracted from blood plasma. In some embodiments, the cfNA is extracted from blood. In some embodiments, the cfNA is extracted from blood plasma. In some embodiments, the blood is collected by venipuncture, a finger stick, arterial draw, or any combination thereof. In some embodiments, the blood is collected in a clinical setting. Non-limiting examples of clinical settings include hospital, urgent care, doctor’s office, and laboratories. In some embodiments, the blood is collected in a blood collection tube. In some embodiments, the blood collection tube is a STRECK tube. In some cases, the cfNA is extracted from serum or plasma.

[0094] In some embodiments, the NA is collected from urine. In some embodiments, the cfNA is collected from urine. In some embodiments, the urine is collected by a subject urinating in a container or through use of a foley catheter. In some embodiments, the container is sterile. In some embodiments, the container is non-sterile. In some embodiments, the container has a lid or seal. The lid may be a cover, a screw top, or pop top. In some embodiments, the seal may be aWSGR Docket No. 55066-711.601plug. In some embodiments, the seal may be a zip top. In some embodiments, urine is collected directly from the foley catheter. In some embodiments, the urine is collected from a urinary drainage bag. Urine may be collected in a clinical setting or a personal setting of the subject. Non-limiting examples of clinical settings include hospital, urgent care, doctor’s office, and laboratories. Non-limiting examples of personal settings of the subject may include a house, an office, or a business. In some embodiments, the subject brings the collected urine sample to the clinical setting. In some embodiments, the subject collects the urine by themselves. In some embodiments, the urine is collected by the subject with the help of another person. In some embodiments, the urine is collected by another person. Non-limiting examples of another person include medical staff such as a nurse, doctor, orderly or a personal contact of the subject such as a family member or friend. In some embodiments, the NA is extracted from stool. In some embodiments, the cfNA is extracted from stool. In some embodiments, the stool is collected by a subject excreting in a container or on to a holder. In some embodiments, the container is sterile. In some embodiments, the container is non-sterile. In some embodiments, the container has a lid or seal. The lid may be a cover, a screw top, or pop top. In some embodiments, the seal may be a plug. In some embodiments, the seal may be a zip top. In some embodiments, the holder is a stick. Stool may be collected in a clinical setting or a personal setting of the subject. Nonlimiting examples of clinical settings include hospital, urgent care, doctor’s office, and laboratories. Non-limiting examples of personal settings of the subject may include a house, an office, or a business. In some embodiments, the subject brings the collected stool sample to the clinical setting. In some embodiments, the subject collects the stool by themselves. In some embodiments, the stool is collected by the subject with the help of another person. In some embodiments, the stool is collected by another person. Non-limiting examples of another person include medical staff such as a nurse, doctor, orderly or a personal contact of the subject such as a family member or friend.

[0095] In some embodiments, the NA is extracted from interstitial fluid. In some embodiments, the cfNA is extracted from interstitial fluid. In some embodiments, the interstitial fluid is collected using a needle. In some embodiments, the interstitial fluid is collected in a clinical setting. Non-limiting examples of clinical settings include hospital, urgent care, doctor’s office, and laboratories. In some embodiments, the interstitial fluid is collected in a collection tube.II. Size enrichment

[0096] In some embodiments, a composition nucleic acid (NA) obtained according to methods of the disclosure is subjected to an operation to selectively enrich particular sizes or types of DNA (e.g. active chromatin, non-nucleosomal DNA).WSGR Docket No. 55066-711.601

[0097] In some embodiments, an operation to selectively enrich particular sizes or types of DNA involves contacting a NA in solution with an anionic solid surface in the presence of a chaotropic agent. In some embodiments, a composition comprising cfNA in solution is contacted with an anionic solid surface in the presence of a chaotropic agent. The anionic solid surface can be any anionic surface. In some embodiments, the anionic solid surface is planar or spherical. In some embodiments, the anionic solid surface is a slide, a flow cell, a well, a bead, or any combination thereof.

[0098] An anionic solid surface may be a planar surface. The planar surface may be a well among a plurality of wells. The plurality of wells may comprise at least two wells. The plurality of wells may comprise at least 1,000 wells. There may be 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 200, 300, 400, 500, 1000, 1500, 2000, 3000, 4000, 5000, 10,000, 100,000, 1,000,000 or more than 1,000,000 wells in a plurality of wells.

[0099] An anionic solid surface may be a bead or nanoparticle. A bead may be a polymer such as a polystyrene bead or polystyrene cross-linked with divinylbenzene. In some embodiments, the bead comprises a single composition. In some embodiments, the bead comprises a bead core and an outer shell.

[0100] In some embodiments, the bead core comprises metal. In some embodiments, the metal bead core is magnetic. In some embodiments, the magnetic metal bead core comprises metal oxide. In some embodiments, the metal oxide is iron oxide. In some embodiments, the iron oxide is iron (II, III) oxide.

[0101] In some embodiments, an outer shell covers the bead core. In some embodiments, the outer shell of the bead is the anionic solid surface. In some embodiments, the outer shell comprises polymer, such as polyethylene glycol, polyacrylic acid, polyacrylamide, polyvinyl alcohol, poly-methyl methacrylate, polystyrene, poly-4-vinylphenol, polyester, polyimide, polyethylene, polypropylene, polyethylene vinyl acetate, polyacrylates, polysaccharide, etc. In some embodiments, the polymer has different molecular weight. In various embodiments, the polymer has an average molecular weight of 100 to 500000 Daltons, such as 200 to 100,000 Daltons, 300 to 50,000 Daltons, 400 to 20,000 Daltons, 500 to 10,000 Daltons, or 600 to 5,000 Daltons. In some embodiments, the polymer comprises one monomer unit. In some of these embodiments, the monomer is ethylene glycol, acrylic acid, acrylamide, or styrene. In some embodiments, the polymer is a copolymer comprising two or more different monomer units. In some of these embodiments, the two or more different monomer units are selected from ethylene glycol, acrylic acid, acrylamide, and styrene. In some embodiments, the polymer has a linear structure. In some embodiments, the polymer has a branched structure.WSGR Docket No. 55066-711.601

[0102] In some embodiments, the shell is a single layer. In some embodiments, the shell comprises a plurality of layers. In certain embodiments, the shell comprises a layer of silicon dioxide and a layer of titanium dioxide. In certain embodiments, the shell comprises a layer of silicon dioxide and a layer of polymer. In certain embodiments, the shell comprises a layer of titanium dioxide and a layer of polymer.

[0103] In some embodiments, the outer shell contains a coating. In some embodiments, the outer shell contains functional groups. In some embodiments, the functional group is attached to the outer shell. In some embodiments, the functional group is covalently attached to the outer shell. In some embodiments, the functional group is attached to the outer shell non-covalently. In some embodiments, the functional group is capable of binding directly to a predetermined molecule, such as a nucleic acid, a protein, a peptide, a carbohydrate, a lipid, or an organic molecule. In certain embodiments, the functional group is capable of binding to a molecular probe, such as a nucleic acid probe or a protein probe. In some of these latter embodiments, the molecular probe is capable, in turn, of binding to a predetermined molecule, such as a nucleic acid, a protein, a peptide, a carbohydrate, or a lipid. In various embodiments, the functional group is carboxyl, hydroxyl, epoxy, carbonyl, aldehyde, amine, maleimide, N-hydroxysuccinimide, carbodiimide, anhydride, hydrazide, polyethylene glycol, azide, nitrile, sulfhydryl, thiocyanate, phosphate, borono, thioester, cysteine, disulfide, alkyl and acyl halide, glutathione, maltose, isocyanate, sulfonyl chloride, tosylate ester, carbonate, arylating agent, imidoester, fluorophenyl ester, or Schiff base. In some embodiments, the functional group can be carboxyl, hydroxyl, epoxy, carbonyl, aldehyde, amine, maleimide, N-hydroxysuccinimide, carbodiimide, anhydride, hydrazide, or biotin. In some embodiments, the magnetic particle comprises a plurality of functional group species. In some embodiments, each of the plurality of functional groups is attached to the outer shell. In various embodiments, the functional group lead to the creation of surface charge of the magnetic nanoparticle. In some embodiments, the surface charge of the magnetic nanoparticle is positive. In some embodiments, the surface charge of the magnetic nanoparticle is negative. In some embodiments, the surface charge of the magnetic nanoparticle can be tuned by changing the pH of the solution.

[0104] In some embodiments, selective enrichment of NAs utilizes a binding mixture in conjunction with an anionic solid surface according to the disclosure. In some embodiments, the binding mixture comprises a chaotropic agent. The chaotropic agent can be any chaotropic agent. Non-limiting examples of chaotropic agents include guanidinium -based solutions (e.g., guanidinium hydrochloride, guanidinium chloride, guanidinium thiocyanate, guanidinium isothiocyanate), urea-based solutions (e.g., urea, thiourea), lithium-based solutions (e.g., lithium perchlorate, lithium acetate), propylene glycol, phenol, and dimethylsulfoxide (DMSO). In someWSGR Docket No. 55066-711.601embodiments, the chaotropic agent is a salt. In some embodiments, the chaotropic agent is a guanidinium salt. In some embodiments, the guanidinium salt is guanidinium isothiocyanate. In some embodiments, the resultant concentration of the chaotropic agent is about 2.5 molar (M) to about 1.5 M. In some embodiments, the resultant concentration of the chaotropic agent is about 2 M to about 1.5 M. In some embodiments, the resultant concentration of the chaotropic agent is about 1.8 M. In some embodiments, the resultant concentration of the chaotropic agent is about 1.5 M. In some embodiments, the binding mixture has a pH of 6.5 or less or 6.0 or less. In some embodiments, lysis buffer has a pH of 8. In some embodiments, the lysis buffer has a pH of between 6.0 and 8.5, between 6.0 and 8.0, between 6.0 and 7.5, between 6.0 and 7.0, or between 6.0 and 6.5. In some embodiments, the binding mixture comprises an alcohol. The alcohol can be any alcohol, such as ethyl alcohol or isopropyl alcohol. In some embodiments, the alcohol is isopropyl alcohol. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is no more than about 15 %, 14%, 13%, 12%, 11%, 10%, 9.5%, 9%, 8.5%, 8%, 7.5%, 7%, 6.5%, 6%, 5.5%, 5%, 4.5%, 4%, 3.5%, 3%, 2.5%, 2%, 1.5% or 1%, or any range between these values. In some embodiments, the resultant concentration of alcohol mixed with the sample is no less than about 10%, 9.5%, 9%, 8.5%, 8%, 7.5%, 7%, 6.5%, 6%, 5.5%, or 5%. In some embodiments, the resultant concentration of alcohol mixed with the sample is about 10 % to about 5 %. In some embodiments, the resultant concentration of alcohol mixed with the sample is about 8 % to about 5 %. In some embodiments, the resultant concentration of alcohol mixed with the sample is about 9 % to about 6 %. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 10%, 9.5%, 9%, 8.5%, 8%, 7.5%, 7%, 6.5%, 6%, 5.5%, or 5%In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 5%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 5.5%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 6%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 6.5%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 7%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 7.5%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 8%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 8.5%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 9%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding is about 9.5%. In some embodiments, the resultant concentration of alcohol mixed with the sample prior to binding isWSGR Docket No. 55066-711.601about 10%. In some embodiments, the binding mixture comprises a detergent. The detergent can be any suitable detergent used to solubilize biological samples. In some embodiments, the detergent is an ionic detergent (e.g. an anionic or cationic detergent). Non-limiting examples of ionic detergents include dodecyl sulfide salts (e.g. sodium dodecylsulfide or SDS), bile acid salts (e.g. Sodium cholate and sodium deoxycholate), and quaternary ammonium detergents (e.g. cetyltrimethylammonium bromide or CTAB). In some embodiments, the detergent is a nonionic detergent. Non-limiting examples of classes of nonionic detergents include hydrophilic polyethylene oxide derivatives (e.g. Triton™ X-100, Triton™ X-102, Triton™ X-114, Triton™ CG-110, Triton™ X-405, Triton™ X-165, Triton™ X-45, Triton™ N-57, Triton™ N-60), lipid-like nonionic detergents (e.g. n-Dodecyl-beta-Maltoside or DDM), steroidal nonionic detergents (e.g. digitonin), nonionic polyoxyethylene detergents (e.g. Tween®-20, Tween®40, Tween®-60, Tween®-80, Tween®65, Tween®-85, Tagat TO, Cremophore RH 40, Cremophore EL, Alpha-tocopherol TGPS, Brij®-96, Brij®-S20, Brij®-S100, Brij®-35, Brij®-58, Brij®-020, Brij®-L23, Brij®-S10, Brij®-010, Brij®-C10, Brij®-93, Brij®-L4, SP Brij® C2 MBAL-SO-(SG), SP Brij® S2 (MB AL), and nonionic ethoxylated nonylphenol detergents (e.g. NP-40). In some embodiments, the nonionic detergent is Tween®-20. In some embodiments, the nonionic detergent is Tween®40. In some embodiments, the nonionic detergent is Tween®- 60. In some embodiments, the nonionic detergent is Tween®-80. In some embodiments, the nonionic detergent is Tween®65. In some embodiments, the nonionic detergent is Tween®-85. In some embodiments, the nonionic detergent is Brij®-96. In some embodiments, the nonionic detergent is Brij®-58. In some embodiments, the nonionic detergent is Brij®-S20. In some embodiments, the nonionic detergent is Brij ®-S 100. In some embodiments, the nonionic detergent is Brij®-35. In some embodiments, the nonionic detergent is Brij®-58. In some embodiments, the nonionic detergent is Brij®-020. In some embodiments, the nonionic detergent is Brij®-L23. In some embodiments, the nonionic detergent is Brij®-S10. In some embodiments, the nonionic detergent is Brij®-010. In some embodiments, the nonionic detergent is Brij®-C10. In some embodiments, the nonionic detergent is Brij®-93. In some embodiments, the nonionic detergent is Brij®-L4. In some embodiments, the nonionic detergent is SP Brij® C2 MBAL-SO-(SG). In some embodiments, the nonionic detergent is SP Brij® S2 (MB AL). In some embodiments the nonionic detergent is a hydrophilic polyethylene oxide derivative. In some embodiments, the nonionic detergent is a lipid-like nonionic detergent. In some embodiments, the nonionic detergent is a steroidal nonionic detergent. In some embodiments, the nonionic detergent is a polysorbate-type nonionic detergent. In some embodiments, the nonionic detergent is a nonionic polyoxyethylene detergent. In some embodiments, the nonionic detergent is an ethoxylated nonylphenol detergent (e.g. NP-40). In some embodiments, the resultantWSGR Docket No. 55066-711.601concentration of detergent mixed with the sample prior to binding is no more than about 6 %. In some embodiments, the resultant concentration of detergent mixed with the sample is no less than about 2 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 6 % to about 2 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 5 % to about 3 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 4 % to about 3 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 3.5 % to about 3 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 3.25 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 3.3 %. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is no more than about 125 mM, 120 mM, 115 mM, 110 mM, 105 mM, 100 mM, 95 mM, 90 mM, 85 mM, 80 mM, 75 mM, 70 mM, 65 mM, 60 mM, 55 mM, 50 mM, 45 mM, 40 mM, 35 mM, 30 mM, 25 mM, 20 mM, 15 mM, 10 mM, or 5mM, or any range between these values. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is at least about 120 mM, 115 mM, 110 mM, 105 mM, 100 mM, 95 mM, 90 mM, 85 mM, 80 mM, 75 mM, 70 mM, 65 mM, 60 mM, 55 mM, 50 mM, 45 mM, 40 mM, 35 mM, 30 mM, 25 mM, 20 mM, 15 mM, 10 mM, or 5mM, or any range between these values. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 120 mM, 115 mM, 110 mM, 105 mM, 100 mM, 95 mM, 90 mM, 85 mM, 80 mM, 75 mM, 70 mM, 65 mM, 60 mM, 55 mM, 50 mM, 45 mM, 40 mM, 35 mM, 30 mM, 25 mM, 20 mM, 15 mM, 10 mM, or 5mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 90 millimolar (mM). In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 85 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 80 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 75 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 70 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 65 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 60 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 55 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 50 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 45 mM. In some embodiments, theWSGR Docket No. 55066-711.601resultant concentration of detergent mixed with the sample prior to binding is about 40 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 35 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 30 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 25 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 20 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 15 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 10 mM. In some embodiments, the resultant concentration of detergent mixed with the sample prior to binding is about 5mM. In some embodiments, the binding mixture has a pH from about 6 to about 9. In some embodiments, the binding mixture has a pH from about 6 to about 8.5. In some embodiments, the binding mixture has a pH of about 8. In some embodiments, the binding mixture has a pH of 7.8. In some embodiments, the binding mixture incubates with the sample at a temperature no higher than about 25°C. In some embodiments, the binding mixture incubates with the sample at a temperature no lower than 15°C. In some embodiments, the binding mixture incubates with the sample at a temperature from about 25°C to about 15°C. In some embodiments, the binding mixture incubates with the sample at a temperature from about 22°C to about 18°C. In some embodiments, the binding mixture incubates with the sample at a temperature of about 20°C. In some embodiments, the binding mixture incubates with the sample at room temperature. In some embodiments, the binding mixture incubates with the sample for no more than 60 minutes. In some embodiments, the binding mixture incubates with the sample for no less than 15 minutes. In some embodiments, the binding mixture incubates with the sample for about 60 minutes to about 15 minutes. In some embodiments, the binding mixture incubates with the sample for about 40 minutes to about 20 minutes. In some embodiments, the binding mixture incubates with the sample for about 35 minutes to about 25 minutes. In some embodiments, the binding mixture incubates with the sample for about 30 minutes. In some embodiments, the binding mixture incubates with the sample undisturbed. In some embodiments, the binding mixture incubates with the sample with disruption. In some embodiments, the disruption is shaking. In some embodiments, the disruption is on a tube rotator.

[0105] In some embodiments, NA not bound to the anionic solid surface is removed. In some embodiments, cfNA not bound to the anionic solid surface is removed. In some embodiments, the unbound NA is removed with a wash buffer comprising a wash operation. In some embodiments, the unbound cfNA is removed with a wash buffer comprising a wash operation. In some embodiments, more than one wash operation is performed. In someWSGR Docket No. 55066-711.601embodiments, more than one wash buffer is used. In some embodiments, the extraction of NAs utilizes a wash buffer. In some embodiments, the extraction of cfNAs utilizes a wash buffer. In some embodiments, the extraction of cfNAs utilizes more than one wash buffer. In some embodiments, the extraction of cfNAs utilizes two different wash buffers. In some embodiments, the first wash buffer comprises a chaotropic agent. The chaotropic agent can be any chaotropic agent. Non-limiting examples of chaotropic agents include guanidinium-based solutions (e.g., guanidinium hydrochloride, guanidinium chloride, guanidinium thiocyanate, guanidinium isothiocyanate), urea-based solutions (e.g., urea, thiourea), lithium-based solutions (e.g., lithium perchlorate, lithium acetate), propylene glycol, phenol, and DMSO. In some embodiments, the chaotropic agent is a salt. In some embodiments, the chaotropic agent is a guanidinium salt. In some embodiments, the guanidinium salt is guanidinium isothiocyanate. In some embodiments, the resultant concentration of the chaotropic agent mixed with the anionic surface in the wash buffer is about 4M (molar), about 3.75M, about 3.5M, about 3.25M, about 3.0 M, about 2.75M, about 2.5M, about 2.0M, about 1.75M, about 1.5M, about 1.25M, about 1 M, about 0.75M, about 0.5 M, or about 0.25M, or any range between these values. In some embodiments, the resultant concentration of the chaotropic agent mixed with the anionic surface in the wash buffer is about 3.5 M to about 1.5 M. In some embodiments, the resultant concentration of the chaotropic agent mixed with the anionic surface in the wash buffer is about 3 M to about 2 M. In some embodiments, the resultant concentration of the chaotropic agent mixed with the anionic surface in the wash buffer is about 2.5 M. In some embodiments, the first wash buffer comprises an alcohol. The alcohol can be any alcohol, such as ethyl alcohol or isopropyl alcohol. In some embodiments, the alcohol is ethyl alcohol. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is no more than about 30%, 25%, 20%, 15%, 10%, 5%, or 1%, or any range between these values. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is at least 25%, 20%, 15%, 10%, 5%, or 1%, or any range between these values. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 30 %, 25%, 20%, 15%, 10%, 5%, or 1%, or any range between these values. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 30 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 25 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 20 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 15 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 10 %. In some embodiments, the resultantWSGR Docket No. 55066-711.601concentration of alcohol mixed with anionic surface in the wash buffer is about 5 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is no less than about 5 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 30 % to about 5 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 25 % to about 5 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 20 % to about 10 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 16 % to about 14 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 15 %. In some embodiments, the first wash buffer comprises a detergent. The detergent can be any suitable detergent used to solubilize biological samples. In some embodiments, the detergent is an ionic detergent (e.g. an anionic or cationic detergent). Non-limiting examples of ionic detergents include dodecylsulfide salts (e.g. sodium dodecylsulfide or SDS), bile acid salts (e.g. Sodium cholate and sodium deoxycholate), and quaternary ammonium detergents (e.g. cetyltrimethylammonium bromide or CTAB). In some embodiments, the detergent is a nonionic detergent. Non-limiting examples of classes of nonionic detergents include hydrophilic polyethylene oxide derivatives (e.g. Triton™ X-100, Triton™ X-102, Triton™ X-l 14, Triton™ CG-110, Triton™ X-405, Triton™ X-165, Triton™ X-45, Triton™ N-57, Triton™ N-60), lipid-like nonionic detergents (e.g. n-Dodecyl-beta-Maltoside or DDM), steroidal nonionic detergents (e.g. digitonin), nonionic polyoxyethylene detergents (e.g. Tween®-20, Tween®40, Tween®- 60, Tween®-80, Tween®65, Tween®-85, Tagat TO, Cremophore RH 40, Cremophore EL, Alpha-tocopherol TGPS, Brij®-96, Brij®-S20, Brij®-S100, Brij®-35, Brij®-58, Brij®-020, Brij®-L23, Brij®-S10, Brij®-010, Brij®-C10, Brij®-93, Brij®-L4, SP Brij® C2 MBAL-SO-(SG), and SP Brij® S2 MB AL), and nonionic ethoxylated nonylphenol detergents (e.g. NP-40). In some embodiments the nonionic detergent is a hydrophilic polyethylene oxide derivative. In some embodiments, the nonionic detergent is a lipid-like nonionic detergent. In some embodiments, the nonionic detergent is a steroidal nonionic detergent. In some embodiments, the nonionic detergent is a polysorbate-type nonionic detergent. In some embodiments, the nonionic detergent is a nonionic polyoxyethylene detergent. In some embodiments, the nonionic detergent is an ethoxylated nonylphenol detergent (e.g. NP-40). In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is no more than about 125 mM, 120 mM, 115 mM, 110 mM, 105 mM, 100 mM, 95 mM, 90 mM, 85 mM, 80 mM, 75 mM, 70 mM, 65 mM, 60 mM, 55 mM, 50 mM, 45 mM, 40 mM, 35 mM, 30 mM, 25 mM, 20 mM, 15 mM, 10 mM, or 5mM, or any range between these values. In someWSGR Docket No. 55066-711.601embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is at least about 120 mM, 115 mM, 110 mM, 105 mM, 100 mM, 95 mM, 90 mM, 85 mM, 80 mM, 75 mM, 70 mM, 65 mM, 60 mM, 55 mM, 50 mM, 45 mM, 40 mM, 35 mM, 30 mM, 25 mM, 20 mM, 15 mM, 10 mM, or 5mM, or any range between these values. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 120 mM, 115 mM, 110 mM, 105 mM, 100 mM, 95 mM, 90 mM, 85 mM, 80 mM, 75 mM, 70 mM, 65 mM, 60 mM, 55 mM, 50 mM, 45 mM, 40 mM, 35 mM, 30 mM, 25 mM, 20 mM, 15 mM, 10 mM, or 5mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 90 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 85 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 80 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 75 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 70 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 65 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 60 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 55 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 50 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 45 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 40 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 35 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 30 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 25 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 20 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 15 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 10 mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 5mM. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is no more than about 5 %, 4.5%, 4%, 3.5%, 3%, 2.5%, 2%, 1.5%, 1%, or 0.5%, or any range between these values. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the washWSGR Docket No. 55066-711.601buffer is no less than about 1 %. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 5 % to about 1 %. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 4 % to about 2 %. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 3.5 % to about 2.5 %. In some embodiments, the resultant concentration of detergent mixed with anionic surface in the wash buffer is about 3 %. In some embodiments, the first wash buffer has a pH from about 6 to about 9. In some embodiments, the first wash buffer has a pH from about 6 to about 8.5. In some embodiments, the first wash buffer has a pH of about 8, 7.75, 7.5. 7.25, 7.0, 6.75, 6.5, or 6, or any range between these values. In some embodiments, the first wash buffer has a pH of 8.1. In some embodiments, the first wash buffer has a pH of greater than or equal to about 6.5 or 6.0. In some embodiments, the first wash buffer is used to wash the sample once. In some embodiments, the second wash buffer comprises an alcohol. The alcohol can be any alcohol, such as ethyl alcohol or isopropyl alcohol. In some embodiments, the alcohol is ethyl alcohol. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is no more than about 90 %, 88%, 86%, 84%, 82%, 80%, 78%, 76%, 74%, 72%, or 70%, or any range between these values. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is no less than about 70 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 90 % to about 70 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 85 % to about 75 %. In some embodiments, the resultant concentration of alcohol mixed with anionic surface in the wash buffer is about 80 %. In some embodiments, the second wash buffer is used to wash the sample at least once. In some embodiments, the second wash buffer is used to wash the sample twice. In some embodiments, the second wash buffer has a pH from about 6 to about 9. In some embodiments, the second wash buffer has a pH of 6.0, 6.25, 6.5, 6.75, 7.0, 7.25, 7.5, 8.0, 8.25, 8.5, 8.75, or 9.0, or any range between these values.

[0106] In some embodiments, the second wash buffer has a pH from about 6 to about 8.5. In some embodiments, the second wash buffer has a pH of about 8.

[0107] In some embodiments, the NA bound the anionic solid surface are eluted. In some embodiments, the cfNA bound the anionic solid surface are eluted. In some embodiments, the NA bound is eluted with an elution operation. In some embodiments, the cfNA bound is eluted with an elution operation. In some embodiments, the elution operation utilizes an elution buffer. In some embodiments, the elution solution is basic. In some embodiments, the extraction of cfNAs utilizes an elution buffer. In some embodiments, the elution buffer comprises tris-baseWSGR Docket No. 55066-711.601buffer. In some embodiments, the elution buffer has a pH from about 6 to about 9. In some embodiments, the elution buffer has a pH of 6.0, 6.25, 6.5, 6.75, 7.0, 7.25, 7.5, 8.0, 8.25, 8.5, 8.75, or 9.0, or any range between these values. In some embodiments, the elution buffer has a pH from about 6.5 to about 9. In some embodiments, the elution buffer has a pH of about 9. In some embodiments, the elution buffer has a pH of 9. In some embodiments, the elution buffer has a pH of 8.

[0108] In some embodiments, the buffers used for NA extraction have a pH that is basic. In some embodiments, the buffers used for NA extraction have a pH that is slightly acidic. In some embodiments, the buffers used for NA extraction have a pH that is neutral. In some embodiments, the pH range of the buffers used for NA extraction have a pH from about 6 to about 9. In some embodiments, the pH of the buffers used for NA extraction is 6.0, 6.25, 6.5, 6.75, 7.0, 7.25, 7.5, 8.0, 8.25, 8.5, 8.75, or 9.0, or any range between these values. In some embodiments, all the buffers used for NA extraction have a same pH. In some embodiments, none of the buffers used for NA extraction have a same pH. In some embodiments, one or more buffers used for NA extraction have a same pH. Non-limiting examples of buffers that may be used in cfNA extraction include lysis buffer, binding buffer, wash buffer, and elution buffer.

[0109] In some embodiments, the active chromatin (e.g. active chromatin enriched by any of the methods described herein) comprises chromatin comprising an H3K4mel modification, a H3K27ac modification, an H4K4me2 modification, an H3K4me3 modification, or any combination thereof.

[0110] In some embodiments, the active chromatin is selectively enriched at least 5-, 10-, 15-, 20-, 25-, 30-, 35-, 40-, 45-, 50-, 60-, 70-, 80-, 90-, or 100-fold versus fragments that are not active chromatin by a process involving contacting an anionic solid surface as described herein. In some embodiments, cfNA fragments at least 200, 211, 281, 334, 336, 400, or 460 base pairs in length are selectively enriched at least 5-, 10-, 15-, 20-, 25-, 30-, 35-, 40-, 45-, 50-, 60-, 70-, 80-, 90-, or 100-fold relative to cfNA fragments less than 200, 211, 281, 334, 336, 400, or 460 base pairs in length by a process involving contacting an anionic solid surface as described herein.

[0111] In some embodiments, an operation to selectively enrich particular sizes or types of DNA involves an electrophoretic separation procedure or an anionic capture procedure. The principle of electrophoretic separation is based on the fact that charged particles migrate when placed in an electric field. Electrophoretic separation is used to separate charged particles, such as proteins or nucleic acids, based on their size and charge. Electric field is applied causing the charged particles move through a medium, such as a gel or a capillary, at different rates based on their size and charge-to-mass ratio. This results in the separation of the particles based on theirWSGR Docket No. 55066-711.601mobility, with smaller, more negatively charged particles moving faster than larger, less charged ones. In some cases, electrophoretic separation procedure comprises gel electrophoresis, Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE), Isoelectric Focusing (IEF), or Capillary Electrophoresis.

[0112] Anion capture procedure can be used to focus and separate molecules based on their charge and size (e.g. cfDNA as described herein). In some cases, anion capture procedure is a variation of capillary electrophoresis (CE). The principle of anion capture procedure involves the use of a negatively charged chemical additive, called an anionic detergent or additive, to interact with and modify the charge of the target molecules. In this procedure, the capillary or separation column used for CE is filled with a buffer solution containing the anionic additive. In some embodiments, anion capture procedure may comprise sample preparation, sample injection, electrophoretic separation, anion capture, focusing and separation, detection or any combination.III. NA fragment analysis / Sequencing

[0113] In some embodiments, the NA fragments as described herein are analyzed. In some embodiments, the cfNA fragments are analyzed using a nucleic acid-based detection assay. In some embodiments, the eluted cfNA fragments are analyzed. In some embodiments, the cfNA fragments are analyzed using a nucleic acid-based detection assay. In some embodiments, the nucleic acid-based detection assay comprises PCR, qPCR, rtPCR, ddPCR, gel electrophoresis (including for e.g., Northern or Southern blot), immunochemistry, in situ hybridization such as fluorescent in situ hybridization (FISH), cytochemistry, or sequencing. In some embodiments, the sequencing technique comprises next generation sequencing. In some embodiments, the methods involve a hybridization assay such as fluorogenic qPCR (e.g., TaqMan™ or SYBR green), which involves a nucleic acid amplification reaction with a specific primer pair, and hybridization of the amplified nucleic acid probes comprising a detectable moiety or molecule that is specific to a predetermined nucleic acid sequence. In some embodiments, the electrophoresis is automated. In some embodiments, the electrophoresis utilizes a TapeStation system.

[0114] NA as described herein can be subjected to a sequencing procedure. Nucleic acids may be sequenced using sequencing methods such as next-generation sequencing, high-throughput sequencing, massively parallel sequencing, sequencing-by-synthesis, paired- end sequencing, single-molecule sequencing, nanopore sequencing, pyrosequencing, semiconductor sequencing, sequencing-by-ligation, sequencing-by-hybridization, RNA-Seq, Digital Gene Expression, Single Molecule Sequencing by Synthesis (SMSS), Clonal Single Molecule ArrayWSGR Docket No. 55066-711.601(Solexa), shotgun sequencing, Maxim-Gilbert sequencing, primer walking, and Sanger sequencing. In some aspects, the methodology or systems of the disclosure utilize systems such as those provided by Illumina, Inc, (including but not limited to HiSeq™ X10, HiSeq™ 1000, HiSeq™ 2000, HiSeq™ 2500, Genome Analyzers™, MiSeq™ NextSeq, NovaSeq 6000 systems), Applied Biosystems Life Technologies (SOLiD™ System, Ion PGMTM Sequencer, ion Proton™ Sequencer) or Genapsys or BGI MGI and other systems.

[0115] Sequencing methods may comprise targeted sequencing, whole-genome sequencing (WGS), lowpass sequencing, bisulfite sequencing, whole-genome bisulfite sequencing (WGBS), or a combination thereof. Sequencing methods may include preparation of suitable libraries. Sequencing methods may include amplification of nucleic acids (e.g., by targeted or universal amplification, such as PCR).

[0116] In some cases, the methods disclosed herein may further comprise performing a next generation sequencing (NGS) reaction (e.g. sequencing using reversible terminators, bridge amplification, DNA nanoballs, or any combination thereof) on the size selected cfNA fragments to obtain the sequences of a plurality of non-canonical cell-free nucleic acid (cfNA) fragments corresponding to plurality of genomic regions of the subject. Next generation sequencing (NGS) can be applied in the field of genomics to enable rapid and cost-effective sequencing of large amounts of DNA or RNA, to analyze entire genomes or transcriptomes in a single experiment. It may apply any of several high-throughput approaches to DNA sequencing using the concept of massively parallel processing. NGS parallelization of the sequencing reactions generates hundreds of megabases to gigabases of nucleotide sequence reads in a single instrument run. This has enabled a drastic increase in available sequence data and fundamentally changed genome sequencing approaches in the biomedical sciences.

[0117] In some embodiments, NGS allows sequencing of entire genomes for studying genetic variations, mutations, and structural changes related to various diseases and traits. In some embodiments, NGS enables transcriptome analysis for studying gene expression patterns, alternative splicing events, and non-coding RNA molecules in various tissues and under different conditions. In some embodiments, NGS may be applied to investigate epigenetic modifications, such as DNA methylation and histone modifications, or alteration of cell-free DNA in response to such alterations or transcription translation.

[0118] In some embodiments, a NGS reaction or operation as described herein may be configured to achieve a particular average coverage over the entire sequenced set of cfNA fragments for a sample. The term “coverage” as used herein generally refers to the abundance of sequence tags mapped to a sequence. In some embodiments, the average coverage is at least about IX, 2X, 5X, 10X, 30X, or 40X. In some embodiments, the average coverage is at leastWSGR Docket No. 55066-711.601about 30X. In some embodiments, the average coverage is at most about 50X, 60X, 70X, 80X, 90X, or 100X. In some embodiments, the average coverage is at most about 100X. In some embodiments, such configuration is accomplished by optimizing sample loading or multiplexing in a flow-cell.

[0119] In some embodiments, NA fragment analysis as described herein may be preceded by a library preparation procedure wherein NA fragments are used to prepare a library (e.g. a library for sequencing). A library can be prepared using platform-specific library preparation method or kit. The method or kit can be commercially available and can generate a sequencer-ready library. Platform-specific library preparation methods can add a predetermined sequence to the end of nucleic acid molecules; the predetermined sequence can be referred to as an adapter sequence. Optionally, the library preparation method can incorporate one or more molecular barcodes. To sequence a population of double-stranded DNA fragments using massively parallel sequencing systems, the DNA fragments can be flanked with adapter sequences compatible with a flow cell used for the sequencing methodology. A collection of such DNA fragments with adapters at either end can be considered sequencing library. Two examples of suitable methods for generating sequencing libraries from purified DNA are: (1) ligation-based attachment of predetermined adapters to either end of fragmented DNA; and (2) transposase-mediated insertion of adapter sequences. In some examples, library preparation involves modification of a nucleic acid molecule or fragment thereof, for example, by ligating a barcode or another tag to a nucleic acid molecule or fragment thereof. Ligating a barcode, or tag to one end of a nucleic acid molecule or fragment thereof may facilitate analysis of the nucleic acid molecule or fragment thereof following sequencing. In some examples, a barcode is nonunique, and barcode sequences can be used in connection with endogenous sequence information such as the start and stop sequences of a target nucleic acid (e.g., the target nucleic acid is flanked by the barcode and the barcode sequences, in connection with the sequences at the beginning and end of the target nucleic acid, creates a uniquely tagged molecule).

[0120] For methods described herein, in some embodiments the methods involve sequencing a sample obtained from a subject. For methods described herein, in some embodiments the methods involve sequencing a sample obtained from a subject by another party. For methods described herein, in some embodiments the methods involve obtaining sequencing data obtained from a sample from a subject generated by another party.

[0121] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction (e.g. a PCR, qPCR, rtPCR, or ddPCR reaction, an isothermal amplification, or any combination thereof) to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of aWSGR Docket No. 55066-711.601gene or a functional fragment thereof gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to a DMARD listed in Table Bl. The gene symbols in Table Bl are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3,4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table Bl. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table Bl.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-3 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to a DMARD listed in Table Bl.In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-5 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table Bl. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 2, 5, 7, 10, 12, 15, 17, 19, 21, 23, or 25, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table Bl. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 5-25 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites,WSGR Docket No. 55066-711.601or insulators) in cfDNA associated with response to an immunosuppressant listed in Table Bl.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 5 to about 25, from about 5 to about 23, from about 5 to about 21, from about 5 to about 19, from about 5 to about 17, from about 5 to about 15, from about 5 to about 13, from about 5 to about 11, from about 5 to about 9, or from about 5 to about 7 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table Bl. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 12 to about 25, from about 12 to about 23, from about 12 to about 21, from about 12 to about 19, from about 12 to about 19, from about 12 to about 17, or from about 12 to about 15 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table Bl. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 17 to about 25, from about 17 to about 23, or from about 17 to about 21, or from about 17 to about 19 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table Bl.Table Bl: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to TNFi or JAKi.WSGR Docket No. 55066-711.601

[0122] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction (e.g. a PCR, qPCR, rtPCR, or ddPCR reaction, an isothermal amplification, or any combination thereof) to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant listed in Table B2. The gene symbols in Table B2 are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3,4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B2. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B2.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-3 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-5 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factorWSGR Docket No. 55066-711.601binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-10 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 3, 5, 7, 10, 13, 15,17, 19, 21, 23, 25, 27, 29, 31, 33, 35, 37, or 39, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 10-39 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 10 to about 39, from about 10 to about 35, from about 10 to about 30, from about 10 to about 25, from about 10 to about 20, or from about 10 to about 15 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 17 to about 39, from about 17 to about 35, from about 17 to about 30, from about 17 to about 25, or from about 17 to about 20 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 29 to about 39, from about 29 to about 35, or from about 29 to about 30 of the genes or functionalWSGR Docket No. 55066-711.601fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B2.Table B2: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to JAKi.

[0123] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction (e.g. a PCR, qPCR, rtPCR, or ddPCR reaction, an isothermal amplification, or any combination thereof) to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant listed in Table B3. The gene symbols in Table B3 are provided corresponding to a specified human reference genome, hgl9. The sequence of theWSGR Docket No. 55066-711.601human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3,4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B3. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B3.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-2 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-5 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-7 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 2, 4, 6, 8, 10, 12, 14, 16, or 18 or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 2-18 genes orWSGR Docket No. 55066-711.601functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 2 to about 18, from about 2 to about 16, from about 2 to about 14, from about 2 to about 12, from about 2 to about 10, from about 2 to about 8, from about 2 to about 6, or from about 2 to about 4 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 8 to about 18, from about 18 to about 16, from about 8 to about 14, from about 8 to about 12, or from about 8 to about 10 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 12 to about 18, from about 14 to about 18, or from about 16 to about 18 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B3.Table B3: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to TNFi.

[0124] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction (e.g. a PCR, qPCR,WSGR Docket No. 55066-711.601rtPCR, or ddPCR reaction, an isothermal amplification, or any combination thereof) to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant listed in Table B4. The gene symbols in Table B4 are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3,4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B4. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B4.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-2 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-3 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-4 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 2,WSGR Docket No. 55066-711.6013, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 4-17 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 4 to about 17, from about 4 to about 16, from about 4 to about 15, from about 4 to about 14, from about 4 to about 13, from about 4 to about 12, from about 4 to about 11, from about 4 to about 10, from about 4 to about 9, from about 4 to about 8, from about 4 to about 7, from about 4 to about 6, or from about 4 to about 5 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 8 to about 17, from about 8 to about 16, from about 8 to about 15, from about 8 to about 14, from about 8 to about 13, from about 8 to about 12, from about 8 to about 11, from about 8 to about 10, from about 8 to about 9 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 12 to about 17, from about 12 to about 16, from about 12 to about 16, from about 12 to about 15, from about 12 to about 14, or about 12 to about 13 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B4.Table B4: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to a T-cell costimulatory blocker.WSGR Docket No. 55066-711.601

[0125] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction (e.g. a PCR, qPCR, rtPCR, or ddPCR reaction, an isothermal amplification, or any combination thereof) to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant listed in Table B5. The gene symbols in Table B5 are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3, 4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B5. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B5.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-2 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject asWSGR Docket No. 55066-711.601non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-3 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-4 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 2, 3, 4, 5, 6, 7, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 2-7 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 1 to about 2, from about 1 to about 3, from about 1 to about 4, from about 1 to about 5, from about 1 to about 6, or from about 1 to about 7 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 3 to about 7, from about 3 to about 6, from about 3 to about 5, from about 3 to about 4 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 5 toWSGR Docket No. 55066-711.601about 7 or from about 5 to about 6 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B5.Table B5: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to a JAKi.

[0126] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant (e.g. a TNFa inhibitor) listed in Table B6. The gene symbols in Table B6 are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3, 4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B6.In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-12 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, orWSGR Docket No. 55066-711.601insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-24 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-59 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 6, 12, 18, 24, 29, 35, 41, 47, 53, 59, 71, 77, 82, 88, 94, 100, 106, 112, 118, 124, 129, 135, 141, 147, 153, 159, 165, 170, 176, 182, 188, 194, 200, 206, 211, 217, 223, 229, or 235, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of at least about 1, 6, 12, 18, 24, 29, 35, 41, 47, 53, 59, 71, 77, 82, 88, 94, 100, 106, 112, 118, 124, 129, 135, 141, 147, 153, 159, 165, 170, 176, 182, 188, 194, 200, 206, 211, 217, 223, 229, or 235, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of the top at least about 1, 6, 12, 18, 24, 29, 35, 41, 47, 53, 59, 71, 77, 82, 88, 94, 100, 106, 112, 118, 124, 129, 135, 141, 147, 153, 159, 165, 170, 176, 182, 188, 194, 200, 206, 211, 217, 223, 229, or 235 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classifyWSGR Docket No. 55066-711.601a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 59-235 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 59 to about 235, from about 59 to about 223, from about 59 to about 217, from about 59 to about 200, from about 59 to about 188, from about 59 to about 176, from about 59 to about 165, from about 59 to about 153, from about 59 to about 141, from about 59 to about 129, from about 59 to about 118, from about 59 to about 106, from about 59 to about 94, from about 59 to about 82, or from about 59 to about 71 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from 118 to about 235, from about 118 to about 223, from about 118 to about 217, from about 118 to about 200, from about 118 to about 188, from about 118 to about 176, from about 118 to about 165, from about 118 to about 153, from about 118 to about 141, or from about 118 to about 129 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from 188 to about 235, from about 188 to about 223, from about 188 to about 217, or from about 188 to about 200 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B6.Table B6: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to a TNFa inhibitor, ranked by how frequently they overlap between folds during feature selection.WSGR Docket No. 55066-711.601WSGR Docket No. 55066-711.601

[0127] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant (e.g. a T-cell costimulatory blocker) listed in Table B7. The gene symbols in Table B7 are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a referenceWSGR Docket No. 55066-711.601number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3, 4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B7. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B7.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-11 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-23 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-57 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 6, 11, 17, 23, 28, 34, 40, 45, 51, 56, 66, 73, 79, 84, 90, 96, 101, 107, 113, 118, 124, 129, 135, 141, 146, 152, 158, 163, 169, 174, 180, 186, 191, 197, 203, 208, 214, 219, 225, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressantWSGR Docket No. 55066-711.601based on the presence, absence, or level of at least about 1, 6, 11, 17, 23, 28, 34, 40, 45, 51, 56, 66, 73, 79, 84, 90, 96, 101, 107, 113, 118, 124, 129, 135, 141, 146, 152, 158, 163, 169, 174, 180, 186, 191, 197, 203, 208, 214, 219, or 225 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of the top about 1, 6, 11, 17, 23, 28, 34, 40, 45, 51, 56, 66, 73, 79, 84, 90, 96, 101, 107, 113, 118, 124, 129, 135, 141, 146, 152, 158, 163, 169, 174, 180, 186, 191, 197, 203, 208, 214, 219, or 225 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 60-225 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 56 to about 225, from about 56 to about 214, from about 56 to about 203, from about 56 to about 191, from about 56 to about 180, from about 56 to about 169, from about 56 to about 158, from about 56 to about 146, from about 56 to about 135, from about 56 to about 124, from about 56 to about 113, from about 56 to about 101, from about 56 to about 90, from about 56 to about 79, or from about 56 to about 68 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from 113 to about 225, from about 113 to about 214, from about 113 to about 203, from about 113 to about 191, from about 113 to about 180, from about 113 to about 169, from about 113 to about 158, from about 113 to about 146, from about 113 to about 135, from about 113 to about 124, from about 113 to about 113, from about 113 to about 101, from about 113 to about 90, from about 113 to about 79, or from about 113 to about 68 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factorWSGR Docket No. 55066-711.601binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from 158 to about 225, from about 158 to about 214, from about 158 to about 203, from about 158 to about 191, from about 158 to about 180, or from about 158 to about 169 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B7. Table B7: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to a T-cell costimulatory blocker, ranked by how frequently they overlap between folds during feature selection.WSGR Docket No. 55066-711.601WSGR Docket No. 55066-711.601

[0128] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant (e.g. a JAK inhibitor) based on the presence, absence, or level of a gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant listed in Table B8. The gene symbols in Table B8 are provided corresponding to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), sequences of and coordinates of the genes or parts of the genes (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3, 4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B8.In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-9 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1-19 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-47 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factorWSGR Docket No. 55066-711.601binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 1, 5, 9, 14, 19, 23, 28, 33, 37, 42, 47, 56, 61, 66, 70, 75, 80, 84, 89, 94, 98, 103, 108, 112, 117, 123, 131, 136, 140, 145, 150, 154, 159, 164, 168, 173, 178, 182, or 187, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of at least about 1, 5, 9, 14, 19, 23, 28, 33, 37, 42, 47, 56, 61, 66, 70, 75, 80, 84, 89, 94, 98, 103, 108, 112, 117, 123, 131, 136, 140, 145, 150, 154, 159, 164, 168, 173, 178, 182, or 187, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of the top about 1, 5, 9, 14, 19, 23, 28, 33, 37, 42, 47, 56, 61, 66, 70, 75, 80, 84, 89, 94, 98, 103, 108, 112, 117, 123, 131, 136, 140, 145, 150, 154, 159, 164, 168, 173, 178, 182, or 187 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 47-187 genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 47 to about 187, from about 47 to about 178, from about 47 to about 168, from about 47 to about 159, from about 47 to about 150, from about 47 to about 140, from about 47 to about 131, from about 47 to about 122, from about 47 to about 112, from about 47 to about 103, from about 47 to about 94, from about 47 to about 84, from about 47 to about 75, from about 47 to about 65, orWSGR Docket No. 55066-711.601from about 47 to about 56 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from 94 to about 187, from about 94 to about 178, from about 94 to about 168, from about 94 to about 159, from about 94 to about 150, from about 94 to about 140, from about 94 to about 131, from about 94 to about 122, from about 94 to about 112, from about 94 to about 103, from about 94 to about 94, from about 94 to about 84, from about 94 to about 75, from about 94 to about 65, or from about 94 to about 56 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8.In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from 140 to about 187, from about 140 to about 178, from about 140 to about 168, from about 140 to about 159, or from about 140 to about 150 of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Table B8.Table B8: Genes (provided as gene symbols) in cfDNA associated with an autoimmune disease response to a JAK inhibitor, ranked by how frequently they overlap between folds during feature selection.WSGR Docket No. 55066-711.601WSGR Docket No. 55066-711.601

[0129] In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction (e.g. a PCR, qPCR, rtPCR, or ddPCR reaction, an isothermal amplification, or any combination thereof) to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of a gene or a functional fragment thereof gene or a functional fragment thereof (e.g. a promoter, enhancer, exon, transcription factor binding site, or insulator) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. The gene symbols in Tables Bl, B2, B3, B4, B5, B6, B7, or B8 are provided relative to a specified human reference genome, hgl9. The sequence of the human reference genome, hgl9, is available from Genome Reference Consortium with a reference number, GRCh37 / hgl9, and also available from Genome Browser provided by Santa Cruz Genomics Institute (genome.ucsc.edu), and loci and sequences corresponding to the genes can be derived therefrom. In some cases, methods or systems according to the reaction can detect at least 3,4, 5, 6, 7, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some cases, a probe can be designed to overlap any nucleotides within a sequence of a gene (or a functional fragment thereof such as a promoter, enhancer, exon, transcription factor binding site, or insulator) corresponding to the gene symbols in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-5% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, orWSGR Docket No. 55066-711.601level of about 1-10% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some cases, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 1-25% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of about 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5%, 20%, 22.5%, 25%, 30%, 32.5%, 35%, 37.5%, 40%, 52.5%, 55%, 57.5%, 60%, 62.5%, 65%, 67.5%, 70%, 72.5%, 75%, 77.5%, 80%, 82.5%, 85%, 87.5%, 90%, 92.5%, 95%, 97.5%, or 100%, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of at least about 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5%, 20%, 22.5%, 25%, 30%, 32.5%, 35%, 37.5%, 40%, 52.5%, 55%, 57.5%, 60%, 62.5%, 65%, 67.5%, 70%, 72.5%, 75%, 77.5%, 80%, 82.5%, 85%, 87.5%, 90%, 92.5%, 95%, 97.5%, or 100%, or any range between these values, of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of the top about 2.5%, 5%, 7.5%, 10%, 12.5%, 15%, 17.5%, 20%, 22.5%, 25%, 30%, 32.5%, 35%, 37.5%, 40%, 52.5%, 55%, 57.5%, 60%, 62.5%, 65%, 67.5%, 70%, 72.5%, 75%, 77.5%, 80%, 82.5%, 85%, 87.5%, 90%, 92.5%, 95%, 97.5%, or 100%, genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In someWSGR Docket No. 55066-711.601embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as responsive to an immunosuppressant based on the presence, absence, or level of about 25-100% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 25% to about 100%, from about 25% to about 95%, from about 25% to about 90%, from about 25% to about 85%, from about 25% to about 80%, from about 25% to about 75%, from about 25% to about 70%, from about 25% to about 65%, from about 25% to about 60%, from about 25% to about 55%, from about 25% to about 50%, from about 25% to about 45%, from about 25% to about 40%, from about 25% to about 35%, or from about 25% to about 30% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 50% to about 100%, from about 50% to about 95%, from about 50% to about 90%, from about 50% to about 85%, from about 50% to about 80%, from about 50% to about 75%, from about 50% to about 70%, from about 50% to about 65%, from about 50% to about 60%, or from about 50% to about 55% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof. In some embodiments, the methods disclosed herein comprise performing a sequencing reaction, a DNA hybridization reaction, or a DNA amplification reaction to classify a subject as non-responsive to an immunosuppressant based on the presence, absence, or level of from about 75% to about 100%, from about 75% to about 95%, from about 75% to about 90%, from about 75% to about 85%, or from about 75% to about 80% of the genes or functional fragments thereof (e.g. promoters, enhancers, exons, transcription factor binding sites, or insulators) in cfDNA associated with response to an immunosuppressant listed in Tables Bl, B2, B3, B4, B5, B6, B7, or B8, or any combination thereof.WSGR Docket No. 55066-711.601IV. Fragment length acquisition / Pre-processing

[0130] Sequence reads obtained by sequencing may be subjected to procedures suitable for processing paired-end sequencing data. Such procedures can involve starting with separate raw sequencing read files corresponding to forward reads (Rl) and reverse reads (R2) and performing a quality control or trimming procedure (e.g. utilizing tools like FastQC or Trimmomatic / Cutadapt) to remove low-quality bases, trim sequencing adapters, or remove reads that are too short after trimming. Such procedures can further comprise using an aligner (e.g. Bowtie2 or BWA-MEM) to map forward and reverse reads simultaneously to a reference genome. Such procedures can yet further comprise filtering the forward and reverse aligned reads for properly paired reads in the correct orientation. Finally, such procedures can involve taking the genomic span from the start of the forward read to the end of the reverse read to obtain an insert or paired-end sequence that represents the sequenced molecule (e.g. a cfNA fragment).

[0131] Sequence reads obtained by sequencing may be subjected to procedures for removing duplicate fragments using tools such as SAMtools markdup.

[0132] Sequence reads obtained by sequencing can also be filtered for mapping quality, to remove reads that cannot be uniquely assigned to a single genomic location.

[0133] Further, sequence reads can be parsed and converted into tabular data frames (e.g. comprising read start, end, strand, mapping quality, and sample metadata). Such a tabular data frame can include an interval tree, a segment tree, a nested containment list, or an augmented interval tree.

[0134] In some cases, sequence reads can be subjected to an operation to correct for GC content or mappability bias. Such procedures can involve a Bin-wise locally estimated scatterplot smoothing (LOESS) or cubic spline coupled with mappability scaling, fragment-level reweighting using empirical bias curves, GLM / GAM normalization with offsets, or GC-stratified quantile normalization.

[0135] In some cases, sequence reads can be subjected to an operation to identify a set of sequences that map to or overlap with a genomic region. In some embodiments, the genomic region may be a particular length in bases (or nucleotides). In some embodiments, the window can be about 0.1, 0.25, 0.5, 1.0, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 10, 15, 20, 25, 30, 35, 40, 45, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 250, 500, 1,000, 2,000, 3,000, 4,000, or 5,000 kilobases (Kb) in length. In some embodiments, the window can be at least about 0.1, 0.25, 0.5, 1.0, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 10, 15, 20, 25, 30, 35, 40, 45, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 250, 500, 1,000, 2,000, 3,000, 4,000, or 5,000 kilobases (Kb) in length. In some embodiments, the window can be at most about 0.25, 0.5, 1.0, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 10, 15,WSGR Docket No. 55066-711.60120, 25, 30, 35, 40, 45, 49, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 250, 500, 1,000, 2,000, 3,000, 4,000, or 5,000 kilobases (Kb) in length.

[0136] In some cases, sequence read counts identified as overlapping with a genomic region can be subjected to a mathematical processing procedure. In some embodiments, the sequence read count data can be logit transformed. In some embodiments, logit transformed data can be normalized to a mean of zero and a standard deviation of one. In some embodiments, the normalized data can be fit to a second-order polynomial. In some embodiments, the second-order polynomial can involve GC content and predetermined expression of the genomic region in a particular cell type (e.g. expression in hematopoietic cells). In some embodiments, the second-order polynomial involving GC content and predetermined expression of the genomic region in a particular cell type is subtracted from a distribution of sequence read counts overlapping the genomic region as a function of length.

[0137] In some cases, sequence read counts can be determined for sequence reads of a particular length. In some embodiments, the length can be at least about 200, 211, 255, 281, 334, 336, 400, or 460 nucleotides. In some embodiments, the length can be at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length. In some embodiments, the length can be at least about 200, 211, 255, 281, 334, 336, 400, or 460 nucleotides in length or at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 nucleotides in length.

[0138] In some cases, the genomic region may comprise or be limited to a particular type of genomic feature. In some embodiments, the genomic region comprise or be limited to a noncoding region, an exon, a gene body exon, an intron, an exon-intron junction, a promoter, a transcriptional start site (TSS), promoter nucleotides other than a TSS, an enhancer, an insulator, a transcription factor binding site (TFBS), a TFBS that is not a TFBS of ASCL1 / NEUROD1 / POUF23 / YAP1, a Cytosine-phosphate-Guanine (CpG) island, a CpG shore, a RNA Polymerase II pausing site, or any combination thereof. In some embodiments, the genomic region may comprise or be limited to a non-coding region, a promoter, promoter nucleotides other than a TSS, an enhancer, an insulator, a transcription factor binding site (TFBS), a TFBS that is not a TFBS of ASCL1 / NEUROD1 / POUF23 / YAP1, a Cytosine-phosphate-Guanine (CpG) island, a CpG shore, or any combination thereof. In some embodiments, the genomic region comprise or be limited to a non-coding region. In some embodiments, the genomic region may comprise or be limited to a promoter. In some embodiments, the genomic region may comprise or be limited to promoter nucleotides other than a transcriptional start site (TSS). In some embodiments, the genomic region may comprise or be limited to an enhancer. In some embodiments, the genomic region may comprise or be limited to an insulator. In some embodiments, the genomic region may comprise or be limitedWSGR Docket No. 55066-711.601to a transcription factor binding site (TFBS). In some embodiments, the genomic region may comprise or be limited to a TFBS of ASCL1 / NEUROD1 / POUF23 / YAP1. In some embodiments, the genomic region may comprise or be limited to a CpG island. In some embodiments, the genomic region may comprise or be limited to a CpG shore.

[0139] In some embodiments, the pre-processing operation can comprise first preparing a sequence matrix or histogram of fragment lengths from sequences obtained from cfNA fragments that overlap with bins, features, or functional genomic regions within a genome. A density estimation procedure can then be performed on the sequence matrix or histogram to produce a population density function (PDF), such as a kernel density estimation (KDE), a spline density estimation, an orthogonal expansion, a Bayesian KDE, a nearest-neighbor density estimation, an adaptive or variable-bandwidth KDE, or a diffusion-based density estimation.

[0140] Sequencing reads and data may be processed using methods such as demultiplexing, adapter-trimming, quality filtering, GC correction, amplification bias correction, correction of batch effects, depth normalization, removal of sex chromosomes, and removal of poor-quality genomic bins.V. Optional approximation operation on fragment length data

[0141] In some cases, sequence reads may be subjected to approximation operation to define an overall behavior of sequence count data mapping to particular bins, features, or functional genomic regions. In some embodiments, the approximation operation may involve a length-based analysis. The length-based analysis may be implemented on individual genomic regions or may be generic to genomic regions (e.g. encompass reads derived from sequencing without assigning them to genomic regions or bins). In some cases, such an approximation operation can be performed on distributions of sequence counts versus length for particular genomic bins or features to determine overall behavior of distributions of fragments mapping to the genomic bins or features. In particular cases, the approximation operation can involve decomposition operation to determine relative contributions of subpopulations to an overall distribution. For example, a decomposition operation can provide information on relative contributions of nucleosomal or non-nucleosomal populations to cfNA fragment distributions such that each can be examined individually (e.g. to isolate regions where one or the other is a dominant method contributing to cfNA fragment generation).

[0142] The approximation operation can comprise a fitting procedure (e.g. an expectation-maximization (EM) fit, a least-squares fit, a maximum likelihood fit, a Bayesian estimator) to a multivariate formulation (aka multivariate model) applied to sequence read vs length data to produce an approximate multivariate formulation summarizing the data. TheWSGR Docket No. 55066-711.601multivariate formulation can comprise a polynomial function, an exponential function, a gaussian function, an exponential function, an arctangent function, or any combination thereof to all (or a subset of) lengths for sequence count versus length data. The multivariate formulation, when used in a decomposition-type analysis, can involve a hierarchical formulation, a Gaussian mixture formulation (e.g. a GMM), or a split Gaussian framework (e.g. a GSMM). Such analyses, rather than attempt to describe the data with a single function, start from a hypothesis that the data derives from several underlying sub-distributions or sub-populations. A gaussian split mixed framework (e.g. a GSMM) fit can be performed by fitting to any of the following equations:1) P(X = x) = E; Pjfj(Aej)wherein P(X=x) represents the overall observed histogram values, x represents the observed fragment length, / / (x|07) is the component density function for the jthsubpopulation in amixture and is a gaussian (e.g. / / wherein |Jj is the mean and Oj is<the variance of an individual gaussian), and p7is the mixing proportion of component); or 2) the maximum likelihood representation3) the log-likelihood representation

[0143] In the case when the approximation operation comprises a fitting procedure, the method can comprise a density estimation procedure appropriate to the algorithm used for fitting. For example, in the cases where least squares or maximum likelihood is used as the fitting algorithm, the density estimation may be skipped; but when an expectation-maximization algorithm is used a density estimation or smoothing may be performed to avoid poor local optima.

[0144] In some cases, the approximation operation can be performed on cfNA fragment reads of a particular size. In some embodiments, the cfNA fragment sequences can be at least about 200, 211, 255, 281, 334, 336, 400, or 460 nucleotides in length. In some embodiments, the cfNA fragment sequences can be or at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 nucleotides in length. In some embodiments, the cfNA fragment sequences can be at least about 200, 211, 255, 281, 334, 336, 400, or 460 base pairs in length and at most about 800, 900, 1000, 1100, 1200, 1300, 1400, or 1500 base pairs in length.WSGR Docket No. 55066-711.601

[0145] In some cases, a split Gaussian framework (e.g. a GSMM) can be fit to the fragment size distribution or array for sample data in the set using an Expectation-Maximalization (EM) approach: the fragment-length data xtas in(wherein , p is the mean and Oj is the variance of anindividual gaussian, and p;is the mixing proportion of component j) was subjected to an optimization procedure to find djthat maximize i=i log (Xj; P> (where 0 representsthe collection of all component parameters).

[0146] Alternatively or additionally, the approximation operation can comprise a mathematical transform. Such transforms can provide information on global or local frequency content or rate of change in distributions or information (e.g. fragment counts as a function of length) over varying scales. The transform can comprise a Fourier transform or a wavelet transform.

[0147] In the case of wavelet transforms, the wavelet transform can involve a variety of wavelet functions. The wavelet transform can comprise a continuous wavelet transform (CWT), or a discrete wavelet transform (DWT). The wavelet functions can be characterized by at least one scale parameter or at least one translation parameter. The wavelet functions can comprise a Haar, Daubechies, Coiflet, Symlet, Meyer, Morlet, or Mexican Hat wavelet function. Examples of implementation of these wavelets are described, e.g. in Lee et al. Journal of Open Source Software, 2019. 10.21105 / joss.01237 (which is incorporated by reference herein in its entirety).

[0148] Parameters that can be extracted from a wavelet transform analysis include but are not limited to the following:• Maximum power: e.g. the maximum power across the power spectrum of the signal • Dominant periodicity (local): e.g. the periodicity at the point of the maximum power • Dominant global periodicity: e.g. the periodicity that corresponds to the maximum power among the power at each periodicity. Unlike the Dominant local periodicity that corresponds to a combination of a given periodicity and a fragment bin length, the global periodicity is calculated by calculating the power across all fragment bin lengths for a given periodicity and selecting the periodicity with the maximum total power.• Mono power: the total wavelet power contained within the mononucleosomal periodicity band (e.g. from about 160 nucleotides in length to about 200 nucleotides in length) • Di power: the total wavelet power contained within the dinucleosomal periodicity band (e.g. from about 320 nucleotides in length to about 400 nucleotides in length)WSGR Docket No. 55066-711.601• Tri power: the total wavelet power contained within the trinucleosomal periodicity band (e.g. about 420 nucleotides in length to about 600 nucleotides in length).

[0149] The approximation operation can comprise performing a dimensionality reduction procedure separate from a machine learning procedure. The dimensionality reduction procedure can comprise a supervised or unsupervised dimensionality reduction technique.Unsupervised dimensionality reduction techniques can include, for example, Principal Component Analysis (PCA), Kernel Principal Component Analysis (KPCA), Isomap, Locally Linear Embedding (LLE), Laplacian Eigenmap, Maximum Variance Unfolding (MVU), and Random Projection, and any combination thereof. Supervised dimensionality reduction techniques can include, for example, Canonical Component Analysis (CCA), Metric Learning (ML) methods, Supervised Principal Component Analysis (SPCA), Locality Sensitive Hashing (LSH), Locality Sensitive Discriminant Analysis (LSD A), Linear Discriminant Analysis, supervised UMAP, Sequential Non-negative Matrix Factorization (NMF), and Generalized Discriminant Analysis (GDA), and any combination thereof.VI. Machine learning / ComputationClassifiers and trainable algorithm

[0150] The methods described herein include using a machine learning model to analyze sample data, particularly to perform a classification between a subject having a pathological or physiological phenotype and a subject not having the pathological or physiological phenotype (e.g. a binary classification).

[0151] In supervised learning approaches (e.g. those used for a machine learning model), a group of samples from two or more groups are analyzed with a statistical classification method. Differential gene or nucleic acid level (e.g. count) data or length data can be discovered that can be used to build a classifier that differentiates between the two or more groups. A new sample can then be analyzed so that the classifier can associate the new sample with one of the two or more groups. Commonly used machine learning model algorithms include without limitation the neural network (multi-layer perceptron), support vector machines, k-nearest neighbors, Gaussian mixture model, Gaussian, naive Bayes, decision tree and radial basis function (RBF) classifiers. Linear classification methods include Fisher's linear discriminant, LDA, logistic regression, naive Bayes classifier, perceptron, and support vector machines (SVMs). Other classifiers for use with the methods according to the disclosure include quadratic classifiers, k-nearest neighbor, boosting, decision trees, random forests, neural networks, pattern recognition, Elastic Net, Golub Classifier, Parzen-window, Iterative RELIEF, ClassificationWSGR Docket No. 55066-711.601Tree, Maximum Likelihood Classifier, Nearest Centroid, Prediction Analysis of Microarrays (PAM), Fuzzy C-Means Clustering, Bayesian networks, Hidden Markov models, logistic regression algorithms, lasso regression algorithm, ridge regression, elastic-net, Naive Bayes, and any combination thereof.

[0152] Generally, using a machine learning model to solve a problem of supervised learning one can involve the following operations:

[0153] 1. Gathering a training set. These can include, for example, genetic or sequencing data from samples that are from subjects before being treated with an immunosuppressant according to the disclosure, alongside subsequent data about immunosuppressant response. The training samples can then be used to “train” the classifier.

[0154] 2. Determine an input object representation for the machine learning model. The accuracy of the machine learning model depends on how the input object is represented. The input object is transformed into a feature vector, which contains a number of features that are descriptive of the data (e.g. levels or counts within particular genomic features or bins, or data derived from levels or counts). The number of features is selected to be an optimal size to avoid overfitting, but large enough to accurately predict the output.

[0155] 3. Determine the structure of the machine learning model and corresponding algorithm. A trainable algorithm is chosen, e.g., artificial neural networks, decision trees, Bayes classifiers or support vector machines. The learning algorithm is used to build the classifier (e.g. machine learning model).

[0156] 4. Build the classifier (e.g. machine learning model). The trainable algorithm is run on the gathered training set. Parameters of the trainable algorithm may be adjusted by optimizing performance on a subset (called a validation set) of the training set, or via cross-validation. After parameter adjustment and learning, the performance of the resultant machine learning model may be measured on a test set of samples that is separate from the training set.

[0157] Once the classifier (e.g. machine learning model) is determined as described above, it can be used to classify a sample, e.g., that of a subject with an autoimmune disease analyzed by the methods of the described herein. In some instances, gene or nucleic acid levels are measured in a sample from the subject and a machine learning model is applied to the resulting data in order to detect or predict response to an immunosuppressant.

[0158] Machine learning can be used to reduce a set of data generated from all (primary sample / analytes / test) combinations into an optimal predictive set of features, e.g., which satisfy specified criteria. In various examples statistical learning, or regression analysis can be applied. Simple to complex and small to large models making a variety of modeling assumptions can be applied to the data in a cross-validation paradigm. Simple to complex includes considerations ofWSGR Docket No. 55066-711.601linearity to non-linearity and non-hi erar chi cal to hierarchical representations of the features. Small to large models include considerations of the size of basis vector space to project the data onto as well as the number of interactions between features that are included in the modelling process.

[0159] Commonly used machine learning model types or trainable algorithms include without limitation the neural network (multi-layer perceptron), support vector machines, k-nearest neighbors, Gaussian, naive Bayes, decision tree and radial basis function (RBF) classifiers. Linear classification methods include Fisher's linear discriminant, LDA, logistic regression, naive Bayes classifier, perceptron, and support vector machines (SVMs). Other classifiers for use with the methods described herein include quadratic classifiers, k-nearest neighbor, boosting, decision trees, random forests, neural networks, pattern recognition, Elastic Net, Golub Classifier, Parzen-window, Iterative RELIEF, Classification Tree, Maximum Likelihood Classifier, Nearest Centroid, Prediction Analysis of Microarrays (PAM), Fuzzy C-Means Clustering, Bayesian networks, Hidden Markov models, logistic regression algorithms, lasso regression algorithm, ridge regression, elastic-net, gradient boosting (e.g. LightGBM or XGBoost), group LASSO, Naive Bayes, and any combination thereof.

[0160] In some embodiments, a machine learning model of a method as described herein utilizes one or more neural networks. In some case, a neural network is a type of computational system that can learn the relationships between an input dataset and a target dataset. A neural network may be a software representation of a human neural system (e.g., cognitive system), intended to capture “learning” and “generalization” abilities as used by a human. In some embodiments, the machine learning algorithm comprises a neural network comprising a CNN. Non-limiting examples of structural components of machine learning algorithms described herein include: CNNs, recurrent neural networks, dilated CNNs, fully-connected neural networks, deep generative models, and Boltzmann machines. Total number of learnable or trainable parameters;

[0161] In some embodiments, the neural network comprises artificial neural networks (ANNs). ANNs may be machine learning algorithms that may be trained to map an input dataset to an output dataset, where the ANN comprises an interconnected group of nodes organized into multiple layers of nodes. For example, the ANN architecture may comprise at least an input layer, one or more hidden layers, and an output layer. The ANN may comprise any total number of layers, and any number of hidden layers, where the hidden layers function as trainable feature extractors that allow mapping of a set of input data to an output value or set of output values. As used herein, a deep learning algorithm (such as a deep neural network (DNN)) is an ANN comprising a plurality of hidden layers, e.g., two or more hidden layers. Each layer of the neuralWSGR Docket No. 55066-711.601network may comprise a number of nodes (or “neurons”). A node receives input that comes either directly from the input data or the output of nodes in foregoing layers, and performs a specific operation, e.g., a summation operation. A connection from an input to a node can be associated with a weight (or weighting factor). The node may sum up the products of all pairs of inputs and their associated weights. The weighted sum may be offset with a bias. The output of a node or neuron may be gated using a threshold or activation function. The activation function may be a linear or non-linear function. The activation function may be, for example, a rectified linear unit (ReLU) activation function, a Leaky ReLU activation function, or other function such as a saturating hyperbolic tangent, identity, binary step, logistic, arctan, softsign, parametric rectified linear unit, exponential linear unit, softplus, bent identity, softexponential, sinusoid, sine, Gaussian, or sigmoid function, or any combination thereof.

[0162] In some embodiments of a machine learning model as described herein, a machine learning model comprises a neural network such as a deep CNN. In some embodiments in which a CNN is used, the network can be constructed with any number of convolutional layers, dilated layers or fully-connected layers. In some embodiments, the number of convolutional layers is between 1-10 and the dilated layers between 0-10. The total number of convolutional layers (including input and output layers) may be at least about 1, 2, 3, 4, 5, 10, 15, 20, or greater, and the total number of dilated layers may be at least about 1, 2, 3, 4, 5, 10, 15, 20, or greater. The total number of convolutional layers may be at most about 20, 15, 10, 5, 4, 3, or less, and the total number of dilated layers may be at most about 20, 15, 10, 5, 4, 3, or less. In some embodiments, the number of convolutional layers is between 1-10 and the fully-connected layers between 0-10. The total number of convolutional layers (including input and output layers) may be at least about 1, 2, 3, 4, 5, 10, 15, 20, or greater, and the total number of fully-connected layers may be at least about 1, 2, 3, 4, 5, 10, 15, 20, or greater. The total number of convolutional layers may be at most about 20, 15, 10, 5, 4, 3, 2, 1, or less, and the total number of fully-connected layers may be at most about 20, 15, 10, 5, 4, 3, 2, 1, or less.

[0163] Alternatively, an attention mechanism (e.g., a transformer) is applied to mimic human cognitive process of selectively focusing on relevant information or data while filtering out irrelevant details. Attention mechanisms may focus on, or “attend to,” certain input regions while ignoring others. This may increase model performance because certain input regions may be less relevant. At each operation, an attention unit can compute a dot product of a context vector and the input at the operation, among other operations. The output of the attention unit may define where the most relevant information or data in the input sequence is located.

[0164] As machine learning models require manipulation of many parameters simultaneously, sometimes tens or hundreds of parameters such as cfNA levels, they can beWSGR Docket No. 55066-711.601developed and calculated using appropriate software programming methods and may be implemented on a computer.Input into Machine Learning Model

[0165] From nucleic acid sequence or count information as described herein, a large set of features can be generated to provide a feature space from which a feature vector can be determined. This feature vector from each of a set of training samples can then be used for training a current version of the machine learning model.

[0166] In some examples, at least one of the features is a count of fragments overlapping a genomic region associated with an autoimmune disease response to an immunosuppressant, a variable from a multivariable formulation fit to a distribution of sequence counts versus length for fragments overlapping the genomic region associated with the autoimmune disease response to the immunosuppressant, a wavelet coefficient calculated by applying a wavelet transform to a distribution of sequence counts versus length for fragments overlapping the genomic region associated with the autoimmune disease response to the immunosuppressant, or an engineered feature from any of the aforementioned.

[0167] The feature selection can select features that are invariant or have low variation within samples that have a same classification (e.g., have a same probability or associated risk of particular phenotype), but where such features vary among groups of samples that have different classifications. Procedures can be implemented to identify what features appear to be the most invariant within a particular population (e.g., one that shares a classification or lease has a similar classification when the classification is a real number). Procedures can also identify features that vary among populations.

[0168] The feature vector can be created as any data structure that can be reproduced for each training sample, so that corresponding data appears in the same place in the data structure across the training samples. For example, the feature vector can be associated with indices, where a particular value exists at each index. As explained above, a matrix can be stored at a particular index of the feature vector, and the matrix elements can have further sub-indices. Other elements of the feature vector can be generated from summary statistics of such a matrix.

[0169] As another example, a single element of a feature vector can correspond to the set of sequence reads across a set of windows of a genome. Thus, an element or the feature vector can itself be a vector. Such counts of reads can be of all reads or certain group (class) of reads, e.g., reads having a particular sequence complexity or entropy. A set of sequence reads can be filtered or normalized, such as for GC bias or mappability bias.WSGR Docket No. 55066-711.601

[0170] In some examples, an element of the feature vector can be the result of a concatenation of multiple features. This can differ from other examples where an element is itself an array (e.g., a vector or matrix) in that the concatenation value can be treated as a single value, as opposed to a collection of values. Thus, features can be concatenated, merged, and combined to be used as engineered features or feature representations for the machine learning model. In some embodiments, the features that are concatenated, merged, or combined into engineered features are one or more variables extracted from a multivariable formulation fit to cfDNA length vs count information generated according to the disclosure. In some embodiments, the features are variables for a gaussian model. In some embodiments, the engineered feature comprises a probability or total counts of a fragment falling in a particular size range peak. In some embodiments, the particular size range peak is a nucleosomal peak. In some embodiments, the nucleosomal peak is about 165 nucleotides or about 308 nucleotides in length. In some embodiments, the nucleosomal peak is about 120 to about 250 nucleotides in length. In some embodiments, the nucleosomal peak is about 240 or 250 nucleotides to about 400 nucleotides in length. In some embodiments, the particular size range peak is a non-nucleosomal peak. In some embodiments, the non-nucleosomal peak is at least about 400 or 401 nucleotides in length. In some embodiments, the non-nucleosomal peak is about 400 to about 1500 nucleotides in length. In some embodiments, the engineered feature comprises a probability or total counts of a fragment falling in all nucleosomal peaks. In some embodiments, the engineered feature comprises a probability or total counts of a fragment falling in all non-nucleosomal peaks. In some embodiments, the engineered feature comprises a probability or total counts of fragment falling in all nucleosomal and all non-nucleosomal peaks. In some embodiments, the engineered feature comprises a ratio of: (i) a probability or total counts of a fragment falling in all non-nucleosomal peaks to; (ii) a probability or total counts of fragment falling in all nucleosomal and all non-nucleosomal peaks. In some embodiments, the engineered feature comprises a maximum classification probability to a nucleosomal peak as determined by a k-means algorithm. In some cases, the feature is any one of : (i) probability or total counts of a fragment falling in all nucleosomal peaks; (ii) probability or total counts of a fragment falling in all non-nucleosomal peaks; (iii) the difference between (ii) and (i); (iv) the ratio of (i) and (ii); and (v) the maximum probability of classifying a fragment to a nucleosomal peaks.

[0171] In some embodiments, the features that are concatenated, merged, or combined into engineered features are one or more wavelet coefficients generated from applying a wavelet transform to cfDNA length vs count information generated according to the disclosure. In some cases, the wavelet coefficients are averaged over a nucleosomal peak. In some cases, the wavelet coefficients are averaged over a non-nucleosomal peak. In some embodiments, theWSGR Docket No. 55066-711.601nucleosomal peak is about 120 to about 250 nucleotides in length. In some embodiments, the nucleosomal peak is about 240 or 250 nucleotides to about 400 nucleotides in length. In some cases, the wavelet coefficients are averaged over a non-nucleosomal peak. In some embodiments, the non-nucleosomal peak is at least about 400 or 401 nucleotides in length. In some embodiments, the non-nucleosomal peak is about 400 to about 1500 nucleotides in length.

[0172] Multiple combinations and approaches to merging the features can be performed. For example, when different measures are counted over the same window (bin), ratios between those bins, such as inversions divided by deletions, may be a useful feature. Further, ratios of bins that are proximal in space and whose merging may convey biological information, such as dividing a transcript start site count by a gene body count, can also serve as a useful feature.

[0173] Weights can be applied to features when they are added to a feature vector. Such weights can be based on elements within the feature vector, or specific values within an element of the feature vector. For example, some or every region (window) in the genome can have a different weight. Some windows can have a weight of zero meaning that the window does not contribute to classification. Other windows can have larger weights, e.g., between 0 and 1. Thus, a weighting mask can be applied to the values for the features used to create the feature vector, e.g., different values of the mask to be applied to features for count, sequence complexity, frequency, sequence similarity in the population, etc.Training of machine learning model

[0174] Weighting factors (or related bias values, and threshold values, or other computational parameters) of the machine learning model as described herein, may be “taught” or “learned” in a training phase using one or more sets of training data involving training samples.

[0175] For supervised learning as described above, training samples (e.g., in tens, hundreds, or thousands) can include measured data (e.g., of various analytes) and predetermined labels, which may be determined via other time-consuming processes, such as imaging or clinical evaluation of the subject and analysis by a trained practitioner. Example labels can include classification of a subject, e.g., discrete classification of whether a subject has an autoimmune disease or not, responds to an immunosuppressant or not, or continuous classifications providing a probability (e.g., a risk or a score) of a discrete value. A learning module can optimize parameters of a model such that a quality metric (e.g., accuracy of prediction to predetermined label) is achieved with one or more specified criteria. Determining a quality metric can be implemented for any arbitrary function including the set of all risk, loss, utility, and decision functions. A gradient can be used in conjunction with a learning procedureWSGR Docket No. 55066-711.601(e.g., a measure of how much the parameters of the model are to be updated for a given time of the optimization process).

[0176] Training of the machine learning model can be according to FIG. 16.At block 1601, for each of the plurality of training samples, sets of measured ...

Claims

WSGR Docket No. 55066-711.601CLAIMSWHAT IS CLAIMED IS:

1. A method of processing a sample of nucleic acids from a subject identified as having an autoimmune condition, the method comprising:(a) selecting cell-free nucleic acid (cfNA) fragments corresponding to active chromatin from said nucleic acids to obtain a plurality of enriched cfNA fragments;(b) detecting enriched cfNA fragments of said plurality of enriched cfNA fragments to generate data, wherein said enriched cfNA fragments of said plurality of enriched cfNA fragments are indicative of a presence or absence of expression of a plurality of genomic regions associated with response or non-response to an immunosuppressant drug; and(c) applying a machine learning model to said data, wherein said machine learning model has been trained with training data that comprises: (i) data indicative of said presence or absence of said expression of said plurality of genomic regions in a plurality subjects having said autoimmune condition; and (ii) data indicative of response or non-response of each subject of said plurality of subjects to said immunosuppressant drug.

2. The method of claim 1, wherein said cfNA fragments are cell-free deoxyribonucleic acid (cfDNA) fragments.

3. The method of claim 1 or 2, wherein said immunosuppressant drug is a diseasemodifying antirheumatic drug (DMARD).

4. The method of any one of claims 1-3, wherein, prior to (a), said subject is undergoing treatment or has undergone treatment with said immunosuppressant drug or with another immunosuppressant drug different from said immunosuppressant drug.

5. The method of any one of claims 3-4, wherein said DMARD is a Janus kinase (JAK) inhibitor.

6. The method of any one of claims 3-4, wherein said DMARD is a Tumor Necrosis Factor alpha (TNFa) inhibitor or a T-cell costimulatory blocker.

7. The method of any one of claims 3-4, wherein said DMARD is a TNFa inhibitor or a JAK inhibitor, further comprising classifying said subject as responsive to said DMARD.

8. The method of any one of claims 1-7, wherein said machine learning model is capable of classifying said subject as responsive to said immunosuppressant drug with at least 80% accuracy.WSGR Docket No. 55066-711.6019. The method of any one of claims 1-8, wherein said machine learning model is capable of classifying said subject as responsive to said immunosuppressant drug with at least 85% specificity and at least 80% sensitivity.

10. The method of any one of claims 1-9, wherein said autoimmune condition is Rheumatoid Arthritis.

11. The method of any one of claims 1-10, wherein response to said immunosuppressant drug is a decrease in Clinical Disease Index (CD Al) of at least about 10 or a decrease in at least one CD Al category.

12. The method of any one of claims 1-11, wherein said nucleic acids are obtained from a blood sample.

13. The method of any one of claims 1-12, wherein (a) comprises obtaining a population of cfNA fragments from said nucleic acids wherein said population of cfNA fragments comprises at least 50% cfNA fragments corresponding to active chromatin.

14. The method of claim 13, wherein (a) further comprises: (i) contacting cfNA fragments of said nucleic acids to an anionic surface to select said cfNA fragments; or (ii) performing a size selection on said nucleic acids to select said cfNA fragments.

15. The method of claim 13 or 14, wherein said plurality of enriched cfNA fragments comprises cfNA fragments of at least about 200 base pairs in length.

16. The method of any one of claims 1-15, wherein said genomic regions comprise transcription factor binding sites, exons, gene body exons, promoters, insulators, DNAse-I hypersensitive sites, or RNA polymerase II pausing sites, or any combination thereof.

17. The method of any one of claims 1-16, further comprising, after (c), using said machine learning model to classify said subject as responsive to said immunosuppressant drug.

18. The method of any one of claims 1-17, further comprising administering said immunosuppressant drug to said subject.

19. The method of claim 17, further comprising reporting an identification of said subject as responsive to said immunosuppressant drug to a physician, a caregiver, or said subject.

20. The method of any one of claims 1-19, wherein said detecting said enriched cfNA fragments comprises performing next-generation sequencing.

21. The method of any one of claims 1-19, wherein detecting said enriched cfNA fragments comprises performing a quantitative reverse polymerase chain reaction (qPCR), a reverse transcriptase polymerase chain reaction (rtPCR), a digital droplet polymerase chain reaction (ddPCR), an isothermal amplification reaction, or any combination thereof.

22. The method of any one of claims 1-21, wherein said machine learning model comprises a logistic regression algorithm, a random forest algorithm, a lasso regression algorithm, a ridgeWSGR Docket No. 55066-711.601regression algorithm, an elastic-net regression algorithm, a gradient boosting algorithm, an extreme gradient boosting algorithm, a group LASSO algorithm, a neural network algorithm, a Naive Bayes algorithm, a support vector machine (SVM), or any combination thereof.

23. The method of any one of claims 1-22, wherein (b) further comprises measuring counts of enriched cfNA fragments of said plurality of cfNA fragments to generate count data, wherein (c) further comprises applying said machine learning model to said count data, and wherein said training data further comprises cfNA count data obtained from a plurality of subjects.

24. The method of any one of claims 1-23, further comprising:(i) identifying a set of features of said data to be input to said machine learning model;(ii) preparing a feature vector of feature values from said data, each feature value corresponding to a feature of said set of features and including one or more measured values;(iii) loading, into memory of a computer system, said machine learning model, said machine learning model having been trained using training vectors obtained from training samples, a first subset of said training samples identified as being from a responder to said immunosuppressant drug and a second subset of said training samples identified as being from a non-responder to said immunosuppressant drug; and(iv) inputting said feature vector into said machine learning model to obtain an output value.

25. The method of claim 24, wherein said output value is a class prediction, a majority vote classification, a class probability, a decision value, or a real-valued score.

26. The method of claim 25, wherein said output value denotes that said subject is likely to be or is a responder to said immunosuppressant drug.

27. A method of processing a sample comprising cell-free nucleic acid (cfNA) fragments from a subject identified as having an autoimmune condition, the method comprising:(a) obtaining sequence data of cfNA fragments obtained from said subject that overlap with a genomic region, of which expression of said genomic region is associated with activity of an immunosuppressant drug, wherein said cfNA fragments are at least about 200 bases in length; (b) from said sequence data, calculating a test sample value indicative of a shape of a distribution of sequence counts versus sequence length for said genomic region over a range of sequence data obtained in (a); and(c) classifying said subject as a responder or non-responder to said immunosuppressant drug at least partially based on said test sample value calculated in (b).

28. The method of claim 27, further comprising calculating said test sample value in part by fitting, via a computer system comprising one or more processors and system memory, a multivariate model to a distribution of sequence counts versus length for said genomic region to generate output parameters.WSGR Docket No. 55066-711.60129. The method of claim 28, wherein fitting said multivariate model comprises(i) preparing a sequence matrix of sequence lengths, via said computer system, from said sequence data in system memory of said computer system; and(ii) fitting, by said one or more processors, said multivariate model to said sequence matrix by iteratively applying an expectation-maximization algorithm for each sequence length value in said sequence matrix to calculate at least one output variable of said multivariate model, thereby calculating said test sample value.

30. A method of processing a sample of cell-free nucleic acids from a subject identified as having an autoimmune condition, the method comprising:(a) obtaining sequence data of nucleic acid (NA) fragments derived from a sample of said subject that overlap with a genomic region, of which expression of said genomic region is associated with activity of an immunosuppressant drug;(b) from said sequence data, performing a wavelet transform on a distribution of sequence counts versus sequence length to compute a relative contribution of a cfNA nucleic acid population about 100 to about 800 nucleotides in size; and(c) classifying said subject as a responder or non-responder to said immunosuppressant drug at least partially based on said relative contribution.

31. The method of claim 30, wherein performing said wavelet transform on said distribution of said sequence counts versus sequence length comprises:(i) preparing, via a computer system and from said sequence data, a sequence matrix of sequence length values in system memory of said computer versus sequence counts;(ii) applying, by one or more processors of said computer system, a wavelet transform to said matrix by convolving said sequence length values with a predetermined wavelet function parameterized by at least one scale parameter and at least one translation parameter to generate a plurality of wavelet coefficients; and(iii) computing, by one or more processors of said computer system, at least one output variable from said plurality of wavelet coefficients, thereby calculating said relative contribution.

32. The method of claim 30 or 31, wherein performing said wavelet transform on said distribution sequence matrix further comprises, between (i) and (ii), computing, by said one or more processors, a windowed signal by applying a window function over sequence lengths greater than 200 nucleotides and applying said wavelet transform to said windowed signal.

33. The method of any one of claims 30-32, wherein performing said wavelet transform on said distribution of sequence counts versus sequence length further comprises, after (ii), selecting, by said one or more processors, a subset of said wavelet coefficients whose translation parameter lies within a fragment length interval greater than 200 nucleotides, and wherein saidWSGR Docket No. 55066-711.601computing in (iii) comprises computing said at least one output parameter from said subset of wavelet coefficients.

34. A method of processing a sample of nucleic acids from a subject identified as having an autoimmune condition, the method comprising:(a) obtaining sequence data of nucleic acid (NA) fragments derived from a sample of said subject that overlap with a genomic region, of which expression of said genomic region is associated with activity of an immunosuppressant drug;(b) from said sequencing data, performing a wavelet transform on a distribution of sequence counts versus sequence length to determine a relative contribution of at least one non-nucleosomal population of NA fragments; and(c) classifying said subject as a responder or non-responder to said immunosuppressant drug at least partially based on said relative contribution.

35. The method of any one of claims 27-34, wherein said classifying further comprises measuring counts of cfNA fragments to generate count data and applying a machine learning model to said count data, wherein said machine learning model has been trained on cfNA count data for a plurality of individuals.

36. A method of processing a sample comprising cell-free nucleic acids from a subject identified as having an autoimmune condition, the method comprising:(a) obtaining sequence data of nucleic acid (NA) fragments obtained from said subject that overlap with a genomic region, of which expression of said genomic region is associated with activity of an immunosuppressant drug;(b) from said sequence data, performing a wavelet transform on a distribution of sequence counts versus sequence length to compute a relative contribution of a cfNA nucleic acid population greater than 200 nucleotides in size, about 100 to about 800 nucleotides in size, or about 100 to about 1500 nucleotides in size; and(c) classifying said subject as a responder or non-responder to said immunosuppressant drug at least partially based on said relative contribution computed in (b).

37. The method of any one of claims 35-36, wherein performing said wavelet transform on said distribution of said counts of said sequences versus length comprises:(i) preparing, via a computer system and from said sequence data, a sequence matrix of sequence length values in system memory of said computer system versus sequence counts; (ii) applying, by one or more processors of said computer system, a wavelet transform to said matrix by convolving said sequence length values with a predetermined wavelet function parameterized by at least one scale parameter and at least one translation parameter to generate a plurality of wavelet coefficients; andWSGR Docket No. 55066-711.601(iii) computing, by said one or more processors, at least one output variable from said plurality of wavelet coefficients, thereby calculating said relative contribution.

38. A method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune condition, the method comprising:(a) detecting a plurality of cfNA fragments from said sample to generate data, wherein a cfNA fragment of said plurality of said cfNA fragments is indicative of a presence or absence of expression of a genomic region associated with activity of an immunosuppressant drug that comprises a Tumor Necrosis Factor Alpha (TNFa) inhibitor; and(b) classifying said subject as a responder or a non-responder to said TNFa inhibitor with an accuracy of at least 64% based on said data.

39. A method of processing a sample of nucleic acids from a subject having or suspected of having an autoimmune disease, the method comprising:(a) detecting a plurality of cfNA fragments from said sample to generate data, wherein a cfNA fragment of said plurality of cfNA fragments is indicative of a presence or absence of a genomic region associated with activity of an immunosuppressant drug that comprises a T-cell costimulatory blocker; and(b) classifying said subject as a responder or a non-responder to said T-cell costimulatory blocker with a specificity greater than 60% at greater than 50% sensitivity based on said data.

40. A method of processing a sample of nucleic acids from a subject having an autoimmune disease, the method comprising:(a) detecting a plurality of cfNA fragments from said sample to generate data, wherein a cfNA fragment of said cfNA fragments is indicative of a presence or absence of a genomic region associated with activity of an immunosuppressant drug that comprises a Janus kinase inhibitor (JAKi); and(b) classifying said subject as a responder or a non-responder to said JAKi with an accuracy greater than 60% based on said data.

Images