Methods and systems for microsatellite analysis - Patents.com

A computer-implemented method using microsatellite classifiers and genetic algorithms addresses the unreliability of current health condition detection methods, enabling early and accurate prediction and characterization of diseases like cancer and neurological disorders.

JP7590344B2Active Publication Date: 2024-11-26ORBIT GENOMICS INC
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2021563182
Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
Priority Date
2019-04-22
Filing Date
2020-04-21
Publication Date
2024-11-26
Estimated Expiration
2040-04-21

AI Technical Summary

Technical Problem

Current methods for predicting, detecting, and characterizing health conditions associated with microsatellite alterations are unreliable and difficult to implement early in the progression of diseases such as cancer and neurological disorders, leading to challenges in diagnosis, prognosis, and treatment selection.

Method used

A computer-implemented method for constructing an optimized classifier using microsatellites, involving ranking and weighting microsatellite subsets through genetic algorithms and ROC analysis to identify an optimized subset for predicting health conditions, and determining genomic age based on microsatellite minor allele signatures.

Benefits of technology

Enhances the accuracy and reliability of early detection and characterization of health conditions by providing sensitive and specific classifiers for conditions like cancer and neurological diseases, improving prognosis and treatment outcomes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007590344000001
    Figure 0007590344000001
  • Figure 0007590344000002
    Figure 0007590344000002
  • Figure 0007590344000003
    Figure 0007590344000003
Patent Text Reader

Abstract

The present disclosure provides methods and systems for classifying microsatellites and minor alleles in a sample. The present disclosure also provides methods and systems for creating classifiers for conditions based on microsatellite loci and for performing pan-cancer assays. The methods and systems can involve next-generation sequencing of a nucleic acid sample from a subject and genotyping the microsatellite loci in the sample. Microsatellites (MS) and their alterations and instability can be the genetic driving forces behind many complex multigenic health conditions, including cancer, neurological diseases, or cardiovascular diseases.
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] cross reference This application claims the benefit of U.S. Provisional Patent Application No. 62 / 837,109, filed April 22, 2019, which is incorporated by reference in its entirety. [Background technology]

[0002] Microsatellites (MS) and their alterations and instability can be the genetic driving force behind many complex multi-gene health conditions, including cancer, neurological disease or cardiovascular disease. Currently, predicting, detecting, diagnosing and characterizing these health conditions by microsatellites can involve matching the microsatellite profile of a patient to a database of microsatellites associated with these health conditions. Such approaches may only be applicable at a later stage in the progression of a health condition, which may lead to unreliability and difficulties in detection, prognosis, diagnosis, selection of treatment and treatment outcome. Therefore, there remains a need for improved methods of predicting, detecting and characterizing these health conditions at both early and late stages by analyzing microsatellite loci. Summary of the Invention [Means for solving the problem]

[0003] In an aspect, the disclosure provides a computer-implemented method for constructing an optimized classifier for a condition, comprising ranking a subset of a plurality of microsatellites as a classifier for the condition in a plurality of optimization cycles, wherein the subset of the plurality of microsatellites comprises microsatellites in an initial population of microsatellites that correlate with the condition, thereby identifying an optimized subset of the subset of the plurality of microsatellites as an optimized classifier for the condition. In some aspects, the computer-implemented method further comprises comparing microsatellites in a first set of samples from subjects having the condition with microsatellites in a second set of samples from subjects not having the condition, thereby identifying the initial population of microsatellites.

[0004] The ranking step may include comparing microsatellites in a first set of samples from subjects with the condition to microsatellites in a second set of samples from subjects without the condition, thereby identifying an initial population of microsatellites. The computer-implemented method may include an initialization including randomly selecting an initial subset population of microsatellites from the initial population of microsatellites for use in the ranking step in an optimization cycle of the plurality of optimization cycles. A population of at least about 100 subsets of the initial population of microsatellites may be used in the plurality of optimization cycles. The minimum number of microsatellites in the subset of subsets of microsatellites may be 8. The maximum number of microsatellites in the subset of subsets of microsatellites may be 64. In some cases, overlapping microsatellites are not allowed in the subset of subsets of microsatellites. The ranking step may include performing a receiver operating characteristic (ROC) analysis using (i) the subsets of microsatellites, (ii) the microsatellites in the samples from subjects with the condition, and (iii) the microsatellites in the samples from subjects without the condition. The ranking step in an optimization cycle of the plurality of optimization cycles may include determining the sum of the sensitivity and specificity of the microsatellites in each subset of the subsets as a classifier for the condition. An optimization cycle of the plurality of optimization cycles may include adding 10 new subsets of the initial population of microsatellites to the subsets from a previous optimization cycle of the plurality of optimization cycles. Seven of the 10 new subsets may be created by randomly splitting and recombining two randomly selected subsets from a previous optimization cycle, and three of the 10 new subsets may be created by randomly selecting microsatellites from the initial population of microsatellites.The method may further include discarding 10 subsets of the subsets in the optimization cycle based, at least in part, on having the lowest ranking in the optimization cycle. In some cases, the condition may be the presence or absence of a health condition in the subject. The condition may be an increased or decreased likelihood of developing a health condition in the subject. The condition may be an increased or decreased likelihood of the subject benefiting from treatment of a health condition. In some cases, the condition may be an increased or decreased likelihood of the subject having an increased risk of an adverse event from treatment of a health condition. The condition may be the responsiveness of the subject to treatment for a health condition. In some cases, the condition may be a prognosis of a health condition in the subject. In some cases, the health condition may be cancer. The cancer may be lung cancer. In other cases, the health condition may be a neurological disease or a cardiovascular disease.

[0005] In another aspect, the disclosure provides a computer-implemented method comprising determining a classifier value for a condition from a sample from a subject using a plurality of parameters, wherein each parameter of the plurality of parameters is a statistical measure of correlation of each of a plurality of microsatellites from samples from subjects having the condition and / or from samples from subjects not having the condition.

[0006] The plurality of weights can include a plurality of optimal weights. In some aspects, the computer-implemented method can include determining a plurality of optimal weights. Determining a plurality of optimal weights can include applying a standard regression analysis to the plurality of weights. Determining a plurality of optimal weights can include using a genetic algorithm. Determining a classifier can include using minor allele frequency data. The plurality of microsatellites can include at least 10 microsatellites. In some instances, each of the plurality of microsatellites correlates with the presence of a condition. The value of the classifier can further include comparing the classifier to a threshold value. In some aspects, the condition can be the presence or absence of a health condition in the subject, an increased or decreased likelihood of developing a health condition in the subject, an increased or decreased likelihood that the subject will benefit from treatment of a health condition, an increased or decreased likelihood that the subject has an increased risk of an adverse event from treatment of a health condition, the responsiveness of the subject to a treatment for a health condition, or a combination thereof. In some cases, the health condition is cancer, a cardiovascular disease, or a neurological disease. If the health condition is cancer, the cancer can be lung cancer.

[0007] In another aspect, the disclosure provides a computer-implemented method for determining a genomic age of a subject, the method comprising: determining a microsatellite minor allele signature in a first sample from the subject; processing the microsatellite minor allele signature with a reference; and determining the genomic age of the subject based on the processing step.

[0008] In some cases, the processing step includes comparing the microsatellite minor allele signature to a reference. The minor allele signature can be a number of minor alleles at the locus. The number of minor alleles can be supported by at least three next generation sequencing sequence reads. The minor allele signature can be a total number of minor allele reads normalized to a total number of major allele reads at the locus. The method can further include performing next generation sequencing of a first sample from the subject to generate sequence reads of the microsatellites of the subject. The first sample can include blood, saliva, or tumor. The method can further include determining a minor allele signature in a second sample from the subject after determining the first genomic age. The method can include evaluating the minor allele signature in the first sample from the subject and the minor allele signature in the second sample from the subject, and determining a rate of genomic aging of the subject based on the evaluating step.

[0009] In another aspect, the disclosure provides a computer-implemented method comprising: determining a plurality of classifiers for a sample from a subject using microsatellites in the sample from the subject; processing the plurality of classifiers with a plurality of reference classifiers for a plurality of conditions; and determining at least one condition for the subject from among the plurality of conditions based on the processing.

[0010] The processing step can include comparing the plurality of classifiers to a plurality of reference classifiers for the plurality of conditions. In some cases, the at least one condition of the plurality of conditions includes the presence or absence of at least one health condition from among the plurality of health conditions of the subject. In some cases, the at least one condition of the plurality of conditions includes an increased or decreased likelihood of developing at least one health condition from among the plurality of health conditions of the subject. The at least one condition of the plurality of conditions can include an increased or decreased likelihood that the subject will benefit from a treatment for at least one health condition from among the plurality of health conditions of the subject. The at least one condition of the plurality of conditions can include an increased or decreased likelihood that the subject will have an increased risk of an adverse event from a treatment for at least one health condition from among the plurality of health conditions of the subject. The at least one condition of the plurality of conditions can include the responsiveness of the subject to a treatment for at least one health condition from among the plurality of health conditions of the subject. The plurality of health conditions can include a plurality of cancers, the plurality of cancers including ovarian cancer, breast cancer, low-grade glioma, glioblastoma, lung cancer, prostate cancer, or melanoma. In some cases, the multiple health conditions may include multiple neurological diseases or multiple cardiovascular diseases.

[0011] In an aspect, the disclosure provides a non-transitory computer-readable medium comprising executable instructions that, when executed by one or more processors, cause the one or more processors to perform a method for constructing an optimized classifier for a condition, comprising ranking a subset of a plurality of microsatellites as a classifier for the condition in a plurality of optimization cycles, wherein the subset of the plurality of microsatellites comprises microsatellites in an initial population of microsatellites that correlate with the condition, thereby identifying the optimized subset of the subset of the plurality of microsatellites as the optimized classifier for the condition. The computer-implemented method may further comprise comparing microsatellites from a first set of samples from subjects having the condition and microsatellites from a second set of samples from subjects not having the condition, thereby identifying the initial population of microsatellites.

[0012] The ranking step may include comparing microsatellites in a first set of samples from subjects with the condition to microsatellites in a second set of samples from subjects without the condition, thereby identifying an initial population of microsatellites. The computer-implemented method may include an initialization including randomly selecting a population of initial subsets of microsatellites from the initial population of microsatellites for use in the ranking step in an optimization cycle of the plurality of optimization cycles. A population of at least about 100 subsets of the initial population of microsatellites may be used in the plurality of optimization cycles. The minimum number of microsatellites in the subset of subsets of microsatellites may be 8. The maximum number of microsatellites in the subset of subsets of microsatellites may be 64. In some embodiments, overlapping microsatellites are not allowed in the subset of subsets of microsatellites. The ranking step may include performing a receiver operating characteristic (ROC) analysis using (i) the subsets of microsatellites, (ii) the microsatellites in the samples from subjects with the condition, and (iii) the microsatellites in the samples from subjects without the condition. The ranking step in an optimization cycle of the plurality of optimization cycles may include determining the sum of the sensitivity and specificity of the microsatellites in each subset of the subsets as a classifier for the condition. An optimization cycle of the plurality of optimization cycles may include adding 10 new subsets of the initial population of microsatellites to the subsets from a previous optimization cycle of the plurality of optimization cycles. Seven of the 10 new subsets may be created by randomly splitting and recombining two randomly selected subsets from a previous optimization cycle, and three of the 10 new subsets may be created by randomly selecting microsatellites from the initial population of microsatellites.The method may further include discarding ten subsets of the subsets in the optimization cycle based, at least in part, on having the lowest ranking in the optimization cycle. The condition may be the presence or absence of a health condition in the subject. The condition may be an increased or decreased likelihood of developing a health condition in the subject. The condition may be an increased or decreased likelihood that the subject will benefit from treatment of a health condition. The condition may be an increased or decreased likelihood that the subject will have an increased risk of an adverse event from treatment of a health condition. The condition may be the responsiveness of the subject to treatment for a health condition. The condition may be a prognosis of a health condition in the subject. The health condition may be cancer. The cancer may be lung cancer. The health condition may be a neurological or cardiovascular disease.

[0013] In another aspect, the disclosure provides a non-transitory computer readable medium comprising executable instructions that, when executed by one or more processors, cause the one or more processors to perform a method comprising: determining a classifier value for a condition from a sample from a subject using a plurality of parameters, wherein each parameter of the plurality of parameters is a statistical measure of correlation of each of a plurality of microsatellites from samples from subjects having the condition and / or from samples from subjects not having the condition.

[0014] The plurality of weights can include a plurality of optimal weights. The computer implemented method can include determining a plurality of optimal weights. Determining a plurality of optimal weights can include applying a standard regression analysis to the plurality of weights. Determining a plurality of optimal weights can include using a genetic algorithm. Determining a classifier can include using minor allele frequency data. The plurality of microsatellites can include at least 10 microsatellites. Each of the plurality of microsatellites can be correlated with the presence of a condition. The value of the classifier can further include comparing the classifier to a threshold value. The condition can be the presence or absence of a health condition in the subject, an increased or decreased likelihood of developing a health condition in the subject, an increased or decreased likelihood that the subject will benefit from treatment of a health condition, an increased or decreased likelihood that the subject has an increased risk of an adverse event from treatment of a health condition, the responsiveness of the subject to a treatment for a health condition, or a combination thereof. The health condition can be cancer, a cardiovascular disease, or a neurological disease. The cancer can be lung cancer.

[0015] In another aspect, the disclosure provides a non-transitory computer readable medium comprising executable instructions that, when executed by one or more processors, cause the one or more processors to perform a method of determining a genomic age of a subject, the method comprising determining a microsatellite minor allele signature in a first sample from the subject, processing the microsatellite minor allele signature with a reference, and determining the genomic age of the subject based on the processing step.

[0016] The processing step can include comparing the microsatellite minor allele signature to a reference. The minor allele signature can be a number of minor alleles at the locus. The number of minor alleles can be supported by at least three next generation sequencing sequence reads. The minor allele signature can be a total number of minor allele reads normalized to a total number of major allele reads at the locus. The method can further include performing next generation sequencing of a first sample from the subject to generate sequence reads of the microsatellites of the subject. The first sample can include blood, saliva, or tumor. The method can further include determining a minor allele signature in a second sample from the subject after determining the first genomic age. The method can include evaluating the minor allele signature in the first sample from the subject and the minor allele signature in the second sample from the subject, and determining a rate of genomic aging of the subject based on the evaluating step.

[0017] In another aspect, the disclosure provides a non-transitory computer readable medium comprising executable instructions that, when executed by one or more processors, cause the one or more processors to perform a method comprising: determining a plurality of classifiers for a sample from the subject using microsatellites in the sample from the subject; processing the plurality of classifiers with a plurality of reference classifiers for a plurality of states; and determining at least one state for the subject from among the plurality of states based on the processing.

[0018] The processing step can include comparing the plurality of classifiers to a plurality of reference classifiers for the plurality of conditions. At least one condition of the plurality of conditions can include the presence or absence of at least one health condition from among the plurality of health conditions of the subject. At least one condition of the plurality of conditions can include an increased or decreased likelihood of developing at least one health condition from among the plurality of health conditions of the subject. At least one condition of the plurality of conditions can include an increased or decreased likelihood that the subject will benefit from treatment of at least one health condition from among the plurality of health conditions of the subject. At least one condition of the plurality of conditions can include an increased or decreased likelihood that the subject will have an increased risk of an adverse event from treatment of at least one health condition from among the plurality of health conditions of the subject. At least one condition of the plurality of conditions can include the responsiveness of the subject to treatment for at least one health condition from among the plurality of health conditions of the subject. The plurality of health conditions can include a plurality of cancers, and the plurality of cancers can include ovarian cancer, breast cancer, low-grade glioma, glioblastoma, lung cancer, prostate cancer, or melanoma. The plurality of health conditions can include a plurality of neurological diseases or a plurality of cardiovascular diseases.

[0019] Another aspect of the present disclosure provides a non-transitory computer-readable medium containing machine-executable code that, when executed by one or more computer processors, performs any of the methods described above or elsewhere herein.

[0020] Another aspect of the present disclosure provides a system including one or more computer processors and a computer memory coupled thereto, the computer memory including machine-executable code that, when executed by the one or more computer processors, performs any of the methods described above or elsewhere herein.

[0021] Additional aspects and advantages of the present disclosure will become readily apparent to those skilled in the art from the following detailed description, in which merely illustrative embodiments of the present disclosure are shown and described. As will be understood, the present disclosure is capable of other and different embodiments, and its several details are capable of modification in various obvious respects, all without departing from the present disclosure. Accordingly, the drawings and description should be regarded as illustrative in nature, and not restrictive.

[0022] Incorporation by Reference All publications, patents, and patent applications mentioned in this specification are incorporated herein 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. To the extent that the publications and patents or patent applications incorporated by reference conflict with a disclosure contained in this specification, the specification is intended to take precedence and / or have the upper hand over any such conflicting material.

[0023] The novel features of the invention are set forth with particularity in the appended claims. A better understanding of the features and advantages of the present invention 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 that are described below. [Brief description of the drawings]

[0024] [Figure 1] FIG. 1 illustrates an example workflow of a computer-implemented method for generating a microsatellite classifier.

[0025] [Diagram 2] FIG. 2 illustrates an example of the development process using a computer-implemented method for identifying informative microsatellite loci and creating classifiers for conditions.

[0026] [Diagram 3]Figure 3 illustrates an example of the validation process for a lung cancer assay.

[0027] [Figure 4] FIG. 4 illustrates an example of validating the pan-cancer assay.

[0028] [Diagram 5] FIG. 5 illustrates an example of a workflow for the analysis of patient samples.

[0029] [Figure 6] Figure 6 illustrates a schematic of the approach used to identify and validate medulloblastoma (MB)-associated MS. The approach involves three stages: computational identification of informative MS loci using a training set, validation of microsatellite markers in an independent validation cohort, and downstream analysis of genes associated with these MS. The first stage includes filters to eliminate MS variants with age, ethnicity, and sequencing technology.

[0030] [Figure 7A] 7A-7D illustrate examples of validation and training data. FIG. 7A illustrates the distribution of metric scores in the training cohort. FIG. 7C illustrates the distribution of metric scores in the validation cohort. ROC analysis was performed on the training (120 MB subjects and 425 control subjects) (FIG. 7B) and validation (102 MB subjects and 428 control subjects) (FIG. 7D) cohorts. [Figure 7B] 7A-7D illustrate examples of validation and training data. FIG. 7A illustrates the distribution of metric scores in the training cohort. FIG. 7C illustrates the distribution of metric scores in the validation cohort. ROC analysis was performed on the training (120 MB subjects and 425 control subjects) (FIG. 7B) and validation (102 MB subjects and 428 control subjects) (FIG. 7D) cohorts. [Figure 7C]7A-7D illustrate examples of validation and training data. FIG. 7A illustrates the distribution of metric scores in the training cohort. FIG. 7C illustrates the distribution of metric scores in the validation cohort. ROC analysis was performed on the training (120 MB subjects and 425 control subjects) (FIG. 7B) and validation (102 MB subjects and 428 control subjects) (FIG. 7D) cohorts. [Figure 7D] 7A-7D illustrate examples of validation and training data. FIG. 7A illustrates the distribution of metric scores in the training cohort. FIG. 7C illustrates the distribution of metric scores in the validation cohort. ROC analysis was performed on the training (120 MB subjects and 425 control subjects) (FIG. 7B) and validation (102 MB subjects and 428 control subjects) (FIG. 7D) cohorts.

[0031] [Figure 8A] Figure 8A illustrates a pie chart displaying the genomic location of 139 MS informative loci relative to MB. Figure 8B illustrates a gene ontology analysis of informative medulloblastoma MS loci. Figure 8C illustrates a protein-protein interaction (PPI) network of 124 genes associated with informative MS loci. The PPI contains 129 nodes and 49 edges resulting in a network with an enrichment p-value of 0.0007. [Figure 8B] Figure 8A illustrates a pie chart displaying the genomic location of 139 MS informative loci relative to MB. Figure 8B illustrates a gene ontology analysis of informative medulloblastoma MS loci. Figure 8C illustrates a protein-protein interaction (PPI) network of 124 genes associated with informative MS loci. The PPI contains 129 nodes and 49 edges resulting in a network with an enrichment p-value of 0.0007. [Figure 8C]Figure 8A illustrates a pie chart displaying the genomic location of 139 MS informative loci relative to MB. Figure 8B illustrates a gene ontology analysis of informative medulloblastoma MS loci. Figure 8C illustrates a protein-protein interaction (PPI) network of 124 genes associated with informative MS loci. The PPI contains 129 nodes and 49 edges resulting in a network with an enrichment p-value of 0.0007.

[0032] [Figure 9] Figure 9 illustrates an example of the genotype distribution and contingency table used in the studies described herein. Distribution of genotypes for microsatellite marker 242626 on chromosome 1, base pairs 153645035. The p-value for this example is 3.5e-4. The table on the right is the contingency table for the same microsatellite marker.

[0033] [Figure 10] FIG. 10 outlines the workflow used to identify age-sensitive MS.

[0034] [Figure 11] FIG. 11 outlines the workflow used to identify sensitive MS to sequencing techniques.

[0035] [Figure 12] FIG. 12 outlines the workflow used to identify MS that are sensitive to ethnicity.

[0036] [Figure 13]Figure 13 illustrates an example of a metric used to assign scores to samples. Consider a hypothetical sample with genotypes 22|22, 12|12 and 13|13 for the above markers, respectively. To apply the metric to this sample, sum the differences in the frequencies of each genotype in the MB and healthy groups: the result is a score of 0.95. In other words, for each genotype, subtract its frequency in the normal group from its frequency in the MB group; then sum the differences. As a result, healthy control individuals have mostly negative scores, while diseased individuals have positive scores.

[0037] [Figure 14] Figure 14 illustrates the Youden index for determining the criterion for distinguishing between MB and healthy samples. The Youden index was used to determine the cutoff of the ROC curve in the training set. The optimal criterion for the list of 43 markers is 0.155. The same criterion was used to calculate the specificity and sensitivity of the validation cohort.

[0038] [Figure 15] FIG. 15 illustrates a circos plot indicating the chromosomal locations of 43 informative loci for MB.

[0039] [Figure 16] Figure 16 illustrates the distribution of genotypes for microsatellite markers 166663 (an exonic microsatellite located in the RAI gene) and 164048 (an exonic microsatellite located in the BLC6B gene). The addition of one CAG triplet can alter the protein structure and impair its function, similar to a missense mutation.

[0040] [Figure 17] FIG. 17 illustrates an example of a report output of microsatellite analysis by a computer-implemented method for assessing a subject's risk of developing cancer.

[0041] [Figure 18] FIG. 18 illustrates a computer system that is programmed or otherwise configured to perform the methods provided herein.

[0042] [Figure 19-1] FIG. 19 illustrates a list of 139 informative germline MS associated with MB. [Figure 19-2] FIG. 19 illustrates a list of 139 informative germline MS associated with MB. [Figure 19-3] FIG. 19 illustrates a list of 139 informative germline MS associated with MB. [Figure 19-4] FIG. 19 illustrates a list of 139 informative germline MS associated with MB. [Figure 19-5] FIG. 19 illustrates a list of 139 informative germline MS associated with MB. [Figure 19-6] FIG. 19 illustrates a list of 139 informative germline MS associated with MB. [Figure 19-7] FIG. 19 illustrates a list of 139 informative germline MS associated with MB.

[0043] [Figure 20-1] FIG. 20 illustrates a list of the 43 microsatellite loci in the MB signature set. [Figure 20-2] FIG. 20 illustrates a list of the 43 microsatellite loci in the MB signature set.

[0044] [Figure 21] FIG. 21 illustrates the Ingenuity Pathway analysis of the informative MB MS locus.

[0045] [Figure 22] FIG. 22 illustrates mutations in informative MB MS locus-associated genes in the cBioportal MB cohort.

[0046] [Figure 23-1] FIG. 23 illustrates an analysis of the cBioportal MB cancer study that revealed that mutations in 135 gene pairs tended to significantly co-occur within the MB cancer risk classifier. [Figure 23-2] FIG. 23 illustrates an analysis of the cBioportal MB cancer study that revealed that mutations in 135 gene pairs tended to significantly co-occur within the MB cancer risk classifier. [Figure 23-3] FIG. 23 illustrates an analysis of the cBioportal MB cancer study that revealed that mutations in 135 gene pairs tended to significantly co-occur within the MB cancer risk classifier.

[0047] [Figure 24] Figure 24 illustrates a threshold with a one standard deviation confidence interval. Classifiers outside this interval indicate subjects who either have the condition (above 0.5) or do not have the condition (below 0.1). Classifier values ​​further away from the threshold carry a stronger indication. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0048] I. Overview The disclosure provides a computer-implemented method for creating a classifier for a condition, for example, using microsatellites. FIG. 1 illustrates an example workflow of how the computer-implemented method is performed to create a classifier. Deoxyribonucleic acid (DNA) sequences are obtained from a database of sequence information from samples of subjects with the condition (101) and sequence information from reference subjects without the condition (102). Microsatellite loci from 101 and 102 are identified (genotyped) and compared to each other to reveal a population of microsatellites that are only associated or correlated with the condition (103). The population of microsatellite loci are then further analyzed and weighted to arrive at an initial set of microsatellite loci for classifier optimization (105) (104). The optimization iteratively ranks the degree to which the microsatellites are associated or correlated with the condition. The optimization can be repeated with additional sets of microsatellites for additional cycles of optimization. In some cases, the set of microsatellites is randomly split and recombined to generate a new initial set of microsatellites for additional cycles of optimization (106). Upon completing optimization, the computer-implemented method identifies the set of microsatellites that are likely to be most informative for creating a classifier (107). Additional validation or optimization steps (108) can be obtained by analyzing additional samples (e.g., from a database) of subjects with known presence or absence of a condition. After 108, the computer-implemented method can be used to create a final classifier (109).

[0049] In one aspect, the disclosure provides an improved computer-implemented method for identifying a set of microsatellites as markers (classifiers) for a condition. The method can further include comparing microsatellite loci from a first set of samples from subjects with the condition and microsatellite loci from a second set of samples from subjects without the condition, thereby identifying an initial population of microsatellite loci (informative loci).

[0050] In some cases, informative loci can be used directly as classifiers. In some cases, classifiers that include informative loci can indicate the presence or absence of a condition in a subject. In some cases, classifiers that include informative loci can indicate an increased or decreased likelihood of developing a condition in a subject. In some instances, classifiers that include informative loci can indicate an increased or decreased likelihood that a subject will benefit from a treatment, or an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment. In some cases, classifiers that include informative loci can indicate responsiveness to a treatment for a condition in a subject. In some instances, classifiers of informative loci can indicate a prognosis of a condition in a subject.

[0051] In some embodiments, the initial population loci of microsatellites (informative loci) are for use in a genetic algorithm performed by a computer-implemented method. The method can include iteratively ranking subsets of the initial population of microsatellites by comparing subsets of microsatellites in samples from subjects with the condition and from samples from subjects without the condition. The method can include initialization, where an initial subset of subsets is randomly selected from the initial population loci of microsatellites. In some instances, about 100 subsets of the initial population loci of microsatellites are used through the genetic algorithm (optimization cycles), with a minimum number of microsatellites in a subset of subsets being 8 and a maximum number of microsatellites in a subset of subsets being 64. In some instances, the iterative ranking step includes multiple optimization cycles, where the multiple optimization cycles include adding 10 new subsets of the initial population of microsatellites to the subset from a previous cycle of optimization. Seven of the 10 new subsets can be created by randomly splitting and recombining two randomly chosen subsets from a previous cycle of optimization, and three of the 10 new subsets are created by randomly selecting microsatellites from the initial population of microsatellites. In some cases, the method includes ranking the subsets in an optimization cycle, and the 10 subsets with the lowest rankings in the optimization cycle are discarded, thus maintaining 100 subsets of the population of microsatellites throughout the optimization cycle. The genetic algorithm can include iteratively ranking every combination of microsatellites to identify the most informative microsatellite loci. The genetic algorithm can improve sensitivity and specificity by removing less informative microsatellite loci and selecting for or weighting more informative microsatellite loci.In some cases, the condition identified by the cycle-optimized microsatellite loci may indicate the presence or absence of a health condition in a subject, an increased or decreased likelihood of developing a health condition in a subject, an increased or decreased likelihood that a subject will benefit from a treatment for a health condition, an increased or decreased likelihood that a subject has an increased risk of an adverse event as a result of a treatment for a health condition, the responsiveness of a subject to a treatment for a health condition, a prognosis of a subject's health condition, or a combination thereof.

[0052] In another aspect, the disclosure provides an improved computer-implemented method, comprising determining a classifier for a condition from a sample from a subject using a plurality of parameters, each parameter of the plurality of parameters being a statistical measure of correlation of each of a plurality of microsatellites from samples from subjects with the condition and / or from samples from subjects without the condition. In some cases, the plurality of parameters includes optimal weights, such as those determined by standard regression analysis and the use of genetic algorithms. In some cases, the classifier is determined by using minor allele frequency data. In some cases, the condition can indicate the presence or absence of a health condition in a subject, an increased or decreased likelihood of developing a health condition in a subject, an increased or decreased likelihood that a subject will benefit from a treatment for a health condition, an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment for a health condition, a responsiveness of the subject to a treatment for a health condition, a prognosis of a health condition in a subject, or a combination thereof. In some cases, the health condition is cancer, a neurological disease, or a cardiovascular disease.

[0053] In another aspect, the present disclosure provides a method of using a computer system to determine minor allele characteristics in a first sample from a subject, compare the minor allele characteristics with a reference, and determine the genomic age for the subject based on the comparison.The minor allele characteristics can be the number of minor alleles at a locus, and the number of alleles is supported by at least 1, at least 2, at least 3, or more than 3 next-generation sequencing sequence reads.In some cases, the minor allele characteristics is the total number of minor allele reads normalized to the total number of major allele reads at a locus.The minor allele characteristics from a first sample from a subject can be compared with a second minor allele characteristics in a second sample from the same subject to determine the rate of genomic aging.

[0054] The present disclosure provides a pan-condition assay based on a classifier generated using microsatellite loci and, optionally, minor allele information. In some cases, the pan-condition assay is a pan-cancer assay.

[0055] The terms "about" or "approximately" can mean within an acceptable error range for a particular value as determined by one of ordinary skill in the art, which will depend in part on how the value is measured or determined, e.g., the limitations of the measurement system. For example, "about" can mean within 1 or more than 1 standard deviation, according to convention, for a given value. About can mean + / - 10%, + / - 5%, + / - 2% or + / - 1% of the value. As used in this specification and claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise. For example, the term "a nucleic acid" includes multiple nucleic acids (including mixtures thereof).

[0056] II. Methods for determining microsatellite classifiers of a condition The present disclosure provides methods, e.g., computer-implemented methods (see, e.g., FIG. 2), and systems for identifying microsatellite classifiers for a condition. The condition can be the presence or absence of a health condition in a subject, an increased or decreased likelihood of development of a health condition in a subject, an increased or decreased likelihood that a subject will benefit from a treatment for a health condition, an increased or decreased likelihood that a subject has an increased risk of an adverse event as a result of a treatment for a health condition, a subject's responsiveness to a treatment for a health condition, a prognosis of a health condition in a subject, or a combination thereof. The method can include identifying (genotyping) microsatellite loci in samples from subjects with and without the condition. The method can include identifying statistically informative microsatellite loci for the condition. The method can include using the statistically informative microsatellite loci to develop a classification signature for the condition. The classification signature can be validated and used to test samples from the subject.

[0057] A. Genotyping of Microsatellite Loci Methods for identifying microsatellite classifiers can include genotyping microsatellite loci in samples from subjects with and without the condition. In some cases, the genotyping step includes analyzing sequence information in a database. In some cases, the genotyping step includes obtaining a sample and analyzing nucleic acid molecules in the sample, for example, by next generation sequencing.

[0058] 1. Sequence information database In some cases, the method of identifying (e.g., genotyping) microsatellite loci can include analyzing sequence information from one or more databases. The one or more databases can include sequence information (e.g., sequence reads) of a nucleic acid sample from a subject having a condition, such as a subject having cancer or a cancer cell line. The one or more databases can include a reference sequence (e.g., a human genome or a portion thereof). The one or more databases can include sequences of variations or polymorphisms in a population(s) of subjects.

[0059] The one or more databases can include sequence information generated by high-throughput or next-generation sequencing.The one or more databases can include sequence data (e.g., sequence read data) generated by whole exome sequencing (WES), whole genome sequencing (WGS) or a combination thereof of a sample from a subject.In certain examples, the one or more databases include sequence information (e.g., sequence read information) generated from targeted sequencing.Targeted sequencing can include enrichment of target sequences from a sample from a subject.

[0060] The database can include sequence information from The Cancer Genome Atlas (TCGA), e.g., exome data, e.g., lung cancer exome data. The database can be derived from the 1000 Genomes Project.

[0061] 2. Sample A sample can be a biological sample obtained or derived from one or more subjects.A sample can be processed or fractionated to produce other samples, such as other biological samples.A sample as described in this disclosure can include any material from which a nucleic acid molecule can be obtained.

[0062] The sample can be obtained from a subject having a condition. The sample can be obtained from a subject having symptoms of the condition. The sample can be obtained from a subject having a condition, but where the subject does not have symptoms of the condition. The sample can be obtained from a subject not having the condition. The sample can be obtained from a subject having cancer, a subject suspected of having cancer, or a subject that does not have or is not suspected of having cancer.

[0063] The sample can be obtained or derived from a human subject. The sample can be stored in various storage conditions before processing, such as at different temperatures (e.g., at room temperature, under refrigerated or freezer conditions, at 25°C, 4°C, -18°C, -20°C or -80°C) or in different suspensions (e.g., EDTA collection tubes, or cell-free DNA or RNA collection tubes).

[0064] Samples can be taken before and / or after treatment of a subject with cancer. Samples can be taken from a subject during treatment or treatment regimen. Multiple samples can be taken from a subject to monitor the effectiveness of treatment over time. Samples can be taken from subjects known or suspected to have cancer where a definitive positive or negative diagnosis is not available by clinical testing. Samples can be taken from subjects suspected of having cancer. Samples can be taken from subjects experiencing unexplained symptoms such as fatigue, nausea, weight loss, aches and pains, weakness, or bleeding. Samples can be taken from subjects with explainable symptoms. Samples can be taken from subjects at risk of developing cancer due to factors such as family history, age, hypertension or prehypertension, diabetes or prediabetes, overweight or obesity, environmental exposures, lifestyle risk factors (e.g., smoking, alcohol consumption, or drug use), or the presence of other risk factors.

[0065] The sample can be a biological sample from a subject. The sample can be whole blood, peripheral blood, plasma, serum, saliva, mucus, urine, semen, lymph, amniotic fluid, fecal extract, buccal swab, cells, or other bodily fluids or tissues, including tissues obtained from surgical biopsy or surgical resection. In some cases, the sample can be a cell line from a primary subject (e.g., a patient) or an archival subject (e.g., a patient) sample, such as an archived sample, such as a formalin-fixed paraffin-embedded (FFPE) sample or a fresh frozen sample. The sample, e.g., biological sample, can be obtained or derived from a subject using an ethylenediaminetetraacetic acid (EDTA) collection tube, a DNA or RNA collection tube, or a cell-free DNA or cell-free RNA collection tube. The sample, e.g., biological sample, can be derived from a whole blood sample by fractionation. The sample, e.g., biological sample, or a derivative thereof, can include cells. The sample, e.g., biological sample, can be a blood sample or a derivative thereof (e.g., blood collected from a collection tube, or a blood drop).

[0066] The sample can contain one or more analytes that can be assayed. The sample can include one or more nucleic acid molecules. The one or more nucleic acid molecules (or any of the nucleic acid molecules disclosed herein, including primers and probes) can be any length of polymeric form of nucleotides, such as deoxyribonucleotides (dNTPs) or ribonucleotides (rNTPs) or any of their analogs. Analogs can include non-naturally occurring bases, nucleotides, including bases that participate in linkages with other nucleotides other than naturally occurring phosphodiester bonds or are attached by linkages other than phosphodiester bonds. Nucleotide analogs include, for example, phosphorothioates, phosphorodithioates, phosphorotriesters, phosphoramidates, boranophosphates, methyl phosphonates, chiral-methyl phosphonates, 2-O-methyl ribonucleotides, peptide nucleic acids (PNAs), and others. The nucleic acid molecule can be deoxyribonucleic acid (DNA). The DNA can be genomic DNA, viral DNA, mitochondrial DNA, plasmid DNA, amplified DNA, circular DNA, circulating DNA, cell-free DNA, or exosomal DNA. In some instances, the DNA is single-stranded DNA (ssDNA), double-stranded DNA, denatured double-stranded DNA, synthetic DNA, and combinations thereof. The circular DNA can be cleaved or fragmented. The DNA can include coding or non-coding regions of a gene or gene fragment of interest, a locus or loci defined from linkage analysis, exons, or introns. The DNA can be a complementary DNA (cDNA). The nucleic acid molecule can be a recombinant nucleic acid, a branched nucleic acid, a plasmid, a vector, or an isolated DNA. The nucleic acid molecule can include one or more modified nucleotides, such as methylated nucleotides or nucleotide analogs. Modifications to the nucleotide structure can be made before or after assembly of the nucleic acid molecule. The sequence of nucleotides of the nucleic acid molecule can be interrupted by non-nucleotide components. The nucleic acid molecule can be further modified after polymerization, such as by conjugation or binding to a reporter agent.

[0067] A nucleic acid molecule can include a locus, gene locus, or genomic region that can be identified by its location in a genome or chromosome. In some cases, a locus can be referred to by gene name and can include coding and non-coding regions associated with that physical region of nucleic acid. A gene can include coding regions (exons), non-coding regions (introns), transcriptional or other regulatory regions, and promoters. In another example, a genomic region can incorporate introns or exons or intron / exon boundaries within a named gene.

[0068] In some instances, the nucleic acid molecule comprises ribonucleic acid (RNA). The RNA can be fragmented RNA. The RNA can be degraded RNA. The RNA can be a microRNA or a portion thereof. The RNA can be an RNA molecule or a fragmented RNA molecule (RNA fragment) selected from microRNA (miRNA), pre-miRNA, pri-miRNA, messenger RNA (mRNA), pre-mRNA, small interfering RNA (siRNA), small hairpin RNA (shRNA), viral RNA, viroid RNA, virusoid RNA, circular RNA (circRNA), ribosomal RNA (rRNA), transfer RNA (tRNA), pre-tRNA, long non-coding RNA (lncRNA), small nuclear RNA (snRNA), circular RNA, cell-free RNA, exosomal RNA, RNA expressed in a vector, RNA transcript, synthetic RNA, ribozyme, cell-free RNA, and combinations thereof.

[0069] In some cases, the sample comprises cell-free nucleic acid molecules. Cell-free nucleic acid molecules can include any unencapsulated nucleic acid molecules, for example, from subject-derived bodily fluids. Cell-free nucleic acid (cfNA) molecules can be nucleic acids in biological samples that are not contained in cells (e.g., cell-free RNA (cfRNA) molecules or cell-free DNA (cfDNA) molecules). cfDNA molecules can circulate freely in bodily fluids, such as in bloodstream. Cell-free DNA molecules can be circulating tumor DNA, for example, cfDNA originating from tumors.

[0070] The sample can be an acellular sample. The acellular sample can be a biological sample that is substantially devoid of intact cells. The acellular sample can be a biological sample that is itself substantially devoid of cells or can be derived from a sample from which cells have been removed. Examples of acellular samples include samples derived from blood, such as serum or plasma; urine; or samples derived from other sources, such as semen, sputum, feces, ductal exudates, lymphatic fluid, or collected lavage fluid.

[0071] The sample can include germline nucleic acid molecules (e.g., nucleic acids from non-diseased cells or tissue, e.g., a tumor). The sample can include nucleic acid molecules from a tumor. In some cases, the sample can include germline nucleic acid molecules (e.g., from non-diseased tissue) and nucleic acid molecules from diseased tissue (e.g., a tumor).

[0072] A sample can contain a target nucleic acid molecule. A target nucleic acid molecule can be any nucleic acid molecule having a nucleotide sequence whose presence, amount and / or sequence or one or more changes therein are desired to be determined.

[0073] Nucleic acid molecules (e.g., RNA or DNA) can be extracted from a sample using, for example, the Qiagen QIAmp DNA Blood Mini kit, the FastDNA kit protocol from MP Biomedicals, or the cell-free biological DNA isolation kit protocol from Norgen Biotek. The extraction method can extract any RNA or DNA molecule from a sample. The extraction method can selectively extract a portion of an RNA or DNA molecule from a sample. The RNA molecules extracted from a sample can be converted into DNA molecules by reverse transcription (RT). Reverse transcription can result in the creation of deoxyribonucleic acid (DNA) from a ribonucleic acid (RNA) template by the action of the enzyme reverse transcriptase.

[0074] The quality of the extracted nucleic acid can be analyzed, for example, using the BIOANALYZER or NANODROP systems.

[0075] The subject can be a person or individual. The subject can be a patient. The subject can be a person who has or is suspected of having cancer. The subject can display symptoms indicative of a health or physiological state or condition. The subject can be asymptomatic with respect to a health or physiological state or condition. The subject described herein can include any member of the class Mammalia: mammals, including humans and non-human primates, such as chimpanzees and other ape and monkey species; domestic animals, such as cows, horses, sheep, goats, pigs; domestic animals, such as rabbits, dogs, and cats; laboratory animals, including rodents, e.g., rats, mice, and guinea pigs, and others. In one embodiment, the mammal is a human.

[0076] Processing a sample obtained from a subject can include subjecting the sample to conditions sufficient to isolate, enrich, or extract a plurality of nucleic acid molecules, and assaying the plurality of nucleic acid molecules to generate a dataset.

[0077] The subject's sample can be analyzed to genotype one or more microsatellites. A microsatellite, microsatellite locus, or microsatellite region as described herein can refer to a tandem repeat of 1-6 nucleotides in a nucleotide sequence. In some cases, the microsatellite comprises a tandem repeat of more than 6 nucleotides. The one or more microsatellites can be found upstream of an exon, downstream of an exon, in an exon, in an intergenic sequence, in an intron, in a region spanning an exon and an intron, in a 3' untranslated region (UTR), in a 5'UTR, or in any other region in the genome. In some instances, the microsatellite pattern of the sample differs from the microsatellite pattern of the reference. The difference in the microsatellite pattern can include single nucleotide polymorphisms (SNPs), percentage of SNPs, indels (insertions, deletions, ratios of insertions and deletions, and combinations thereof), or ratios of indels to SNPs. In some instances, the pattern of microsatellite differences includes, for example, haplotyping the percentage of homozygosity, heterozygosity, or minor alleles at a given locus. When the pattern of microsatellite differences is located in an exon region, the differences can include nonsynonymous SNPs, synonymous SNPs, frameshift indels, nonframeshift indels, stopgains, and stoplosses. Samples can be matched, for example, for age, sex, or ethnicity (e.g., Caucasian, African American, Hispanic American). In some cases, samples are unmatched. In some cases, samples can be accompanied by additional clinical metadata, including, for example, health condition, cancer, cardiac or neurological condition, treatment condition or response, or disease stage. Clinical metadata can be correlated with the microsatellites to determine whether the microsatellites are informative with respect to the clinical metadata.

[0078] The identity (e.g., genotype) of one or more microsatellites can be obtained by any available method or technique, including next-generation sequencing, high-throughput sequencing, sequencing-by-synthesis, pyrosequencing, classical Sanger sequencing methods, sequencing-by-ligation, sequencing-by-synthesis, sequencing-by-hybridization, RNA-Seq (Illumina), ILLUMINA sequencing (using reversible terminating nucleotides), paired-end sequencing, digital gene expression (Helicos), single molecule sequencing, e.g., single molecule sequencing-by-synthesis (SMSS) (Helicos), Ion Torrent (semiconductor) sequencing (Life Technologies / Thermo-Fisher), massively parallel sequencing, clonal single molecule arrays (Solexa), nanopore sequencing, Pacific Biosciences SMRT sequencing, shotgun sequencing, Maxim-Gilbert sequencing, primer walking, and any other sequencing method.

[0079] Next generation sequencing can include sample multiplexing. Sample multiplexing can be at least or at most about 12 samples, 24 samples, 48 ​​samples, 96 samples, 192 samples, 384 samples, 768 samples, or 1536 samples. Sequencing depth can be about 1× to about 10×, about 10× to about 100×, about 100× to about 500×, or about 500× to about 1000×.

[0080] Sequencing depth can be at least, at most, or about 1×, 5×, 10×, 50×, 100×, 200×, 250×, 300×, 400×, or 500×. Base calling consensus accuracy can be at least 95%, 96%, 97%, 98%, 99%, or greater than about 99%. The quality score can be at least Q10 (e.g., less than 1:10 error rate, greater than 90% inferred base calling accuracy), greater than Q20 (e.g., less than 1:100 error rate, greater than 99% inferred base calling accuracy), greater than Q30 (e.g., less than 1:1000 error rate, greater than 99.9% inferred base calling accuracy), greater than Q40 (e.g., less than 1:10,000 error rate, greater than 99.99% inferred base calling accuracy), or greater than Q50 (e.g., less than 1:100,000 error rate, greater than 99.999% inferred base calling accuracy). The assembly method can produce at least 95%, 96%, 97%, 98% or 99% accuracy for calling microsatellite genotypes in next generation sequencing datasets.

[0081] After nucleic acid molecule is subjected to sequencing, suitable bioinformatics processing can be carried out on sequence read.For example, sequence read can be aligned to one or more reference genomes (e.g., one or more species genomes, such as human genome).Aligned sequence read can be quantified at one or more loci (e.g., one or more microsatellite loci).

[0082] In some embodiments, identification (e.g., genotyping) of one or more microsatellites includes amplifying the nucleotide sequence of one or more microsatellite loci, e.g., by performing a polymerase chain reaction (PCR) using primers, e.g., specific primers that flank one or more microsatellite loci, and evaluating the amplified fragments, e.g., by capillary electrophoresis or sequencing. PCR can be quantitative PCR (qPCR), digital PCR, or reverse transcription PCR. Amplifying or amplification can increase the size or amount of a nucleic acid molecule. Amplified nucleic acid molecules can be single-stranded or double-stranded. Amplification can include creating one or more copies of an amplified product of a nucleic acid molecule. Amplification can be performed, for example, by extension (e.g., primer extension) or ligation. Amplification can include performing a primer extension reaction to create a strand complementary to a single-stranded nucleic acid molecule, and in some cases, creating one or more copies of a strand and / or a single-stranded nucleic acid molecule.

[0083] Amplification of a nucleic acid molecule, e.g., a nucleic acid molecule comprising one or more microsatellite loci, can be performed using any nucleic acid amplification method, such as loop-mediated isothermal amplification (LAMP), nucleic acid sequence-based amplification (NASBA), self-sustained sequence replication (3SR), rolling circle amplification (RCA), recombinase polymerase amplification (RPA), multiple displacement amplification (MDA), helicase-dependent amplification (HDA), strand displacement amplification (SDA), nicking enzyme amplification reaction (NEAR), exponential amplification reaction (EXPAR), polymerase spiral reaction (PSR), isothermal multiple displacement amplification (IMDA), The amplification can be performed by ramification amplification method (RAM), single primer isothermal amplification (SPIA), signal-mediated amplification technology of RNA (SMART), beacon-assisted detection amplification (BADAMP), hinge-initiated primer-dependent amplification of nucleic acids (HIP), smart amplification process (SmartAmp), hybridization chain reaction (HCR), some toehold-mediated strand displacement (TMSD), ligase chain reaction, digital PCR (dPCR), droplet digital PCR (ddPCR) or transcription-mediated amplification. Amplification can involve multiplex amplification, for example, using AMPLISEQ. In some cases, RNA is converted to cDNA by reverse transcription prior to amplification. The assay readout can include quantitative PCR (qPCR) values, digital PCR (dPCR) values, digital droplet PCR (ddPCR) values, fluorescence values, etc., or normalized values ​​thereof. Other assays that can be used in the methods provided herein include immunoassays, electrochemical assays, surface-enhanced Raman spectroscopy (SERS), quantum dot (QD)-based assays, molecular inversion probes, CRISPR / Cas-based detection (e.g., CRISPR typing PCR (ctPCR), specific highly sensitive enzyme reporter un-locking (SHERLOCK), DNA endonuclease targeted CRISPR transreporter (DETECTR), CRISPR-mediated analog multi-event recorder (CAMERA)), and laser transmission spectroscopy (LTS).

[0084] Multiplex amplification can include amplifying about 10 to about 50 targets, about 50 to about 100 targets, about 100 to about 500 targets, or about 500 to about 1000 targets. Adapters can be added (e.g., ligated) to the nucleic acid molecules to facilitate amplification and / or sequencing, such as in the ILLUMINA sequencing platform, including universal adapters. Universal primers can bind to universal adapters for amplification.

[0085] Multiple samples can be analyzed, and each multiplexed sample can be barcoded. The RNA or DNA molecules isolated or extracted from the sample can be tagged with, for example, identifiable tags, allowing multiple samples to be multiplexed. Any number of RNA or DNA samples can be multiplexed. For example, a multiplexed reaction can contain at least about 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100 or more than 100 RNA or DNA from initial samples. For example, multiple samples can be tagged with sample barcodes so that each DNA molecule can be traced back to the sample (and subject) from which it originates. Such tags can be attached to RNA or DNA molecules by ligation or by PCR amplification with primers.

[0086] In some cases, a bait set (e.g., hybridization probes, e.g., SURESELECT or SEQCAP) is used to obtain a target, e.g., a target nucleic acid molecule. The target can include RNA and / or DNA. The hybridization probe can be at least 15, 25, 50, 75, 100, 120, or 150 bases in length. The hybridization probe can be 15-50 bases, 50-100 bases, or 100-150 bases in length. The probe can be a nucleic acid molecule (e.g., RNA or DNA) having sequence complementarity to a nucleic acid sequence (e.g., RNA or DNA) of one or more loci (e.g., one or more microsatellites). Assaying a sample with a probe selective for one or more loci (e.g., one or more microsatellites) can include the use of array hybridization (e.g., microarray-based), polymerase chain reaction (PCR), or nucleic acid sequencing (e.g., RNA sequencing or DNA sequencing).

[0087] In some aspects, the analysis of nucleic acid molecules comprises performing next-generation sequencing. In some cases, the sequencing of microsatellites can be performed directly, for example, without amplification. Next-generation sequencing methods can include whole genome, whole exome and partial genome or exome. Next-generation sequencing methods can be used in targeted sequences, enriched sequences or combinations thereof.

[0088] In some instances, enrichment is performed by an enrichment kit prior to sequencing and downstream analysis. In some cases, enrichment is performed by an enrichment kit to enrich microsatellite loci that are subjected to genetic algorithm validation. The use of an enrichment kit can increase the number of callable alleles or genotypes in the reads and can increase the ability to analyze a greater percentage of informative loci for a given sample. The enrichment kit can include an enrichment array or probe that hybridizes to target sequences of microsatellites and flanking sequences on one or both sides of the microsatellite. In some cases, the use of enrichment increases the number of callable genotypes by at least 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100% or more compared to the number of callable genotypes that can be obtained without the use of an enrichment kit. In some instances, the use of the enrichment kit increases the number of callable genotypes by at least 2, 3, 4, 5, 6, 7, 8, 9, 10 or more times compared to the number of callable genotypes without the use of the enrichment kit. In some embodiments, the enrichment kits disclosed herein comprise compositions that can be used to perform the methods described herein.

[0089] 3. Algorithm for genotyping Microsatellites can be genotyped using algorithms. The algorithm can be, for example, Bayesian model selection guided by empirically derived error models, or discretized Gaussian mixtures (e.g., GenoTan). The algorithm can be, for example, Repeatseq. Microsatellites can be genotyped using dynamic programming-based approaches or heuristic methods. Other tools for microsatellite genotyping include PHOBOS, MISA, Tandem Repeats Finder, FullSSR or bMSISEA.

[0090] B. Identification of informative microsatellites Identifying informative microsatellites can include identifying a first set of microsatellite loci from a sample of a subject having a condition and a second set of microsatellite loci from a sample of a subject not having the condition. In some cases, the second set of microsatellite loci can be obtained from a database of reference sequences.

[0091] 1. Statistics Differences between the first and second sets of microsatellite loci can be statistically detected and compared by one or more statistical tests, such as t-tests, Z-tests, ANOVA, regression analysis, Mann-Whitney-Wilcoxon, chi-squared tests, correlation, Fisher exact test, Bonferroni correction, and Benjamini-Hochberg test. In some cases, statistical differences are quantified using a generalized Fisher exact test. In some cases, a Benjamini-Hochberg multiple testing correction is applied to control for false discovery rates.

[0092] 2. Microsatellite filtering For example, if samples from subjects with and without the condition do not match the factors, the microsatellites can be filtered to control for any number of factors, such as age, ethnicity, sex, sequencing protocol (e.g., WSG, WES, or targeted sequencing). Microsatellites with potential bias can be excluded from subsequent analysis. Additional filters for filtering microsatellites can include the length of the microsatellite repeat motif, the total length of the microsatellite (e.g., number of copies of the motif), the sequence of the motif (e.g., only use those with high GC content), the purity of the microsatellite, such as whether it has any bases that may interrupt the perfect set of copies of the motif. In some instances, the microsatellites can be filtered by their position in the genome, such as exome, intron, intergenic region, or untranslated region. Filtering can include filtering by genes or functional elements in close proximity to the microsatellite.

[0093] 3. Scoring the Samples The statistical test can generate a receiver operating characteristic (ROC) curve, and the area under the ROC curve is referred to as the area under the curve (AUC). The AUC can be determined to assess the accuracy of the comparison of the sets of microsatellite loci. A larger AUC can indicate a higher accuracy of the association or correlation of the condition to the difference between the first and second sets of microsatellite loci. The ROC curve can determine the sensitivity (e.g., true positive) and specificity (e.g., true negative) rates for the association or correlation of the condition to the difference between the first and second sets of microsatellite loci. Sensitivity, also referred to as true positive rate, recall, or detection probability, can measure the proportion of actual positives that are correctly identified for the presence or absence of a condition. Sensitivity can quantify the avoidance of false negatives by calculating the number of true positives divided by the sum of the number of true positives and the number of false negatives. Specificity, also referred to as true negative rate, can measure the proportion of actual negatives that are correctly identified for the presence or absence of a condition. Specificity can be quantified by calculating the number of true negatives divided by the sum of the number of true negatives and the number of false positives.

[0094] In some instances, a statistically significant association or correlation of a condition to a first set of microsatellite loci that differs from a second set of microsatellite loci has a statistical accuracy of at least 70%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98% or 99%. In some cases, a statistically significant association or correlation of a condition to a first set of microsatellite loci that differs from a second set of microsatellite loci has a statistical specificity of at least 0.70, 0.80, 0.85, 0.90, 0.91, 0.92, 0.93, 0.94, 0.95, 0.96, 0.97, 0.98, or 0.99 and a statistical sensitivity of at least 0.70, 0.80, 0.85, 0.90, 0.91, 0.92, 0.93, 0.94, 0.95, 0.96, 0.97, 0.98, or 0.99.

[0095] In some instances, identifying informative microsatellites includes identifying a first set of microsatellite loci from a database that includes nucleic acid sequences obtained from a subject having a condition, such as sequences of cancer types from the Cancer Genome Atlas Program (TCGA), and a second set of microsatellite loci from a reference database (e.g., hg19 or the 1000 Genomes Project). A cancer type, such as breast cancer, can be subtyped based on, for example, stage, morphology, histology, gene expression, receptor profile, mutation profile, invasiveness, prognosis, malignant characteristics, and the like. Cancer types and cancer subtypes can be applied at a finer level to distinguish a certain histological type of cancer or a cancer subtype, for example, defined according to mutation profile or gene expression. Cancer stage can refer to a classification of a cancer type based on histological and pathological features related to disease progression. In some instances, the set of microsatellite loci is obtained from a database that includes nucleic acid sequences that include nucleotide variants or polymorphisms. In some cases, a first set of microsatellite loci is obtained from a sample having a condition and compared to a second set of microsatellite loci obtained from a database.

[0096] 4. Status In some cases, the condition associated with or correlated to a difference in the set of microsatellite loci may indicate the presence or absence of a health condition in a subject, an increased or decreased likelihood of development of a health condition in a subject, an increased or decreased likelihood that a subject will benefit from a treatment for a health condition, an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment for a health condition, the subject's responsiveness to a treatment for a health condition, a prognosis of a health condition in a subject, or a combination thereof. In some cases, the health condition is cancer. In some cases, the cancer is a solid or hematological malignancy. In certain cases, the cancer is metastatic, relapsing, or refractory. Cancers that may be associated with or correlated to a difference in the set of microsatellite loci include acute myeloid leukemia (LAML or AML), acute lymphoblastic leukemia (ALL), adrenocortical carcinoma (ACC), bladder urothelial carcinoma (BLCA), brain stem glioma, brain low grade malignant tumors (LDMTs), and gliomas with low grade ... grade)glioma (LGG), brain tumor, breast cancer (BRCA), bronchial tumor, Burkitt's lymphoma, carcinoma of unknown primary site, carcinoid tumor, carcinoma of unknown primary site, atypical teratoid / rhabdoid tumor of the central nervous system, central nervous system embryonal tumor, squamous cell carcinoma of the cervix, endocervical adenocarcinoma (CESC) cancer, childhood cancer, cholangiocarcinoma (CHOL), chordoma, chronic lymphocytic leukemia, chronic myeloid leukemia, chronic myeloproliferative disorder, colon (adenocarcinoma) cancer (COA D), colorectal cancer, craniopharyngioma, cutaneous T-cell lymphoma, endocrine pancreatic islet cell tumor, endometrial cancer, ependymoblastoma, ependymoma, esophageal cancer (ESCA), esthesioneuroblastoma, Ewing's sarcoma, extracranial germ cell tumor, extragonadal germ cell tumor, extrahepatic bile duct cancer, gallbladder cancer, gastric / stomach cancer, gastrointestinal carcinoid tumor, gastrointestinal stromal cell tumor, gastrointestinal stromal tumor (GIST), gestational trophoblastic tumor, glioblastoma multiformemultiforme) glioma GBM), hairy cell leukemia, head and neck cancer (HNSD), cardiac cancer, Hodgkin's lymphoma, hypopharyngeal cancer, intraocular melanoma, islet cell tumor, Kaposi's sarcoma, kidney cancer, Langerhans cell histiocytosis, laryngeal cancer, lip cancer, liver cancer, lymphoid neoplasms diffuse large B-cell lymphoma (DLBCL), malignant fibrous histiocytoma bone cancer, medulloblastoma, medulloepithelioma, melanoma, Merkel cell Cancer, Merkel cell skin cancer, mesothelioma (MESO), metastatic squamous cell neck cancer of unknown primary, oral cavity cancer, multiple endocrine neoplasia syndrome, multiple myeloma, multiple myeloma / plasma cell neoplasm, mycosis fungoides, myelodysplastic syndrome, myeloproliferative neoplasm, nasal cavity cancer, nasopharyngeal cancer, neuroblastoma, non-Hodgkin's lymphoma, non-melanoma skin cancer, non-small cell lung cancer, oral cancer, oralpulmonary cavity cancer, oropharyngeal cancer, osteosarcoma, other brain and spinal cord tumors, ovarian cancer, ovarian epithelial cancer, ovarian germ cell tumors, ovarian low malignant potential tumors, pancreatic cancer, papilloma, paranasal sinus cancer, parathyroid cancer, pelvic cancer, penile cancer, pharyngeal cancer, pheochromocytoma and paraganglioma (PCPG), moderately differentiated pineal parenchymal tumors, pineoblastoma, pituitary tumors, plasma cell neoplasms / multiple myeloma, pleuropulmonary blastoma, primary central nervous system (CNS) lymphoma, primary hepatocellular liver cancer, prostate cancer including prostate adenocarcinoma (PRAD), rectal cancer, renal cancer, renal cell (kidney) cancer, renal cell carcinoma, airway cancer, retinoblastoma, rhabdomyosarcoma, salivary gland cancer, sarcoma (SARC), Sézary syndrome, cutaneous melanoma (SKCM), small cell lung cancer, small intestine cancer, soft tissue sarcoma, squamous cell carcinoma, squamous cell cervical cancer, stomach / gastric cancer, supratentorial primitive neuroectodermal tumor, T-cell lymphoma, testicular cancer, testicular germ cell tumor (TGCT), pharyngeal cancer, thymic carcinoma, thymoma (THYM), thyroid cancer (THCA), transitional cell carcinoma, transitional cell carcinoma of the renal pelvis and ureter, trophoblastic tumor, ureteral cancer, urethral cancer, uterine cancer, uterine cancer, uveal melanoma (UVM), vaginal cancer, vulvar cancer, Waldenstrom's hypergammaglobulinemia, or Wilms' tumor. In some aspects, the cancer type comprises acute lymphocytic leukemia, acute myeloid leukemia, bladder cancer, breast cancer, brain cancer, cervical cancer, cholangiocarcinoma, colon cancer, colorectal cancer, endometrial cancer, esophageal cancer, gastrointestinal cancer, glioma, glioblastoma, head and neck cancer, kidney cancer, liver cancer, lung cancer, lymphoid neoplasms, melanoma, myeloid neoplasms, ovarian cancer, pancreatic cancer, pheochromocytoma and paraganglioma, prostate cancer, rectal cancer, squamous cell carcinoma, testicular cancer, gastric cancer, or thyroid cancer.

[0097] In some cases, the health condition is lung cancer or a subtype of lung cancer. Lung cancers that may be associated or correlated with different sets of microsatellite loci include non-small cell lung cancer (NSCLC) (e.g., lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and large cell carcinoma), small cell lung cancer (SCLC), and lung carcinoid tumors.

[0098] In some cases, the health condition is a neurological disease. Examples of neurological diseases that may be associated or correlated with differences in a set of microsatellite loci include myotonic dystrophy, fragile X-associated tremor / ataxia syndrome, spinocerebellar ataxia, Kennedy disease, Huntington's disease, spinal-bulbar muscular atrophy, progressive myoclonus epilepsy 1 (Unverricht-Lundborg disease), fragile X syndrome, fragile XE syndrome, dentatorubral-pallidoluysian atrophy, Friedreich ataxia, oculopharyngeal muscular dystrophy, fragile X-associated primary ovarian failure, Huntington's disease-like 2, C9ORF72-associated frontotemporal dementia, and amyotrophic lateral sclerosis. The health condition can be autism.

[0099] In some cases, the condition is inflammatory bowel disease (IBD), which can include gastrointestinal disorders of the gastrointestinal tract. Non-limiting examples of IBD include Crohn's disease (CD), ulcerative colitis (UC), indeterminate colitis (IC), microscopic colitis, diversion colitis, Behcet's disease, and other indeterminate forms of IBD. In some instances, IBD includes fibrosis, fibrostenosis, stricture and / or perforation disease, obstructive disease, or refractory disease (e.g., mrUC, refractory CD), perianal CD, or other complex forms of IBD.

[0100] In some instances, the condition is a cardiovascular disease, which can include coronary heart disease (CAD), rheumatic heart disease, congenital heart disease, cardiomyopathy, heart tumors, vascular tumors, heart valve disease, disorders of the heart lining, stroke, aortic aneurysm, peripheral artery disease, deep vein thrombosis (DVT), or pulmonary embolism.

[0101] In some cases, the condition is a metabolic disease or disorder, which may include acid-base disorders, metabolic brain diseases, disorders of calcium metabolism, DNA repair deficiency disorders, disorders of glucose metabolism, hyperlactatemia, disorders of iron metabolism, disorders of lipid metabolism, malabsorption syndromes, metabolic syndrome X, inborn errors of metabolism, mitochondrial diseases, disorders of phosphorus metabolism, porphyrias, proteostasis deficiencies, metabolic skin diseases, wasting syndromes, or fluid-electrolyte imbalances.

[0102] In some cases, the health conditions include Achalasia, Addison's Disease, Adult Still's Disease, Agammaglobulinemia, Alopecia Areata, Amyloidosis, Ankylosing Spondylitis, Anti-GBM / Anti-TBM Nephritis, Antiphospholipid Syndrome, Autoimmune Angioedema, Autoimmune Autonomic Neuropathy, Autoimmune Encephalomyelitis, Autoimmune Hepatitis, Autoimmune Inner Ear Disease (AIED), Autoimmune Myocarditis, Autoimmune Oophoritis, Autoimmune Orchitis, Autoimmune Pancreatitis, Autoimmune Retinopathy, Autoimmune Urticaria, Axonal & Neuronal Neuropathy (AMAN), Balo's Disease, Behçet's Disease, Benign Mucous Membrane Pemphigoid, Bullous Pemphigoid, Castleman's Disease (CD), Celiac Disease, Chagas Disease, Chronic Inflammatory Demyelinating Polyneuropathy (CIDP), Chronic Recurrent Multifocal Osteomyelitis (CRMO), Churg-Strauss Syndrome (CHS). SS) or eosinophilic granulomatosis (EGPA), cicatricial pemphigoid, Cogan's syndrome, cold agglutinin disease, congenital heart block, Coxsackie myocarditis, CREST syndrome, Crohn's disease, dermatitis herpetiformis, dermatomyositis, Devic's disease (neuromyelitis optica), discoid lupus, Dressler's syndrome, endometriosis, eosinophilic esophagitis (EoE), eosinophilic fasciitis, erythema nodosum, essential mixed cryoglobulinemia , Evans syndrome, fibromyalgia, fibrosing alveolitis, giant cell arteritis (temporal arteritis), giant cell myocarditis, glomerulonephritis, Goodpasture's syndrome, granulomatosis with polyangiitis, Graves' disease, Guillain-Barré syndrome, Hashimoto's thyroiditis, hemolytic anemia, Henoch-Schönlein purpura (HSP), herpes gestationis or pemphigoid of gestationis (PG), hidradenitis suppurativa (HS) (acne inversaInversa), hypogammaglobulinemia, IgA nephropathy, IgG4-related sclerosing disease, immune thrombocytopenic purpura (ITP), inclusion body myositis (IBM), interstitial cystitis (IC), juvenile arthritis, juvenile diabetes mellitus (type 1 diabetes), juvenile myositis (JM), Kawasaki disease, Lambert-Eaton syndrome, leukocytoclastic vasculitis, lichen planus, lichen sclerosus, woody conjunctivitis, linear immunoglobulin A disease (LAD), lupus, Lyme disease, Meniere's disease, microscopic polyangiitis (MPA), mixed connective tissue disease (MCTD), Mooren's ulcer, Much-Habermann disease, multifocal motor neuropathy (MMN) or MMNCB, multiple sclerosis, myasthenia gravis, myositis, narcolepsy, neonatal lupus, neuromyelitis optica, neutropenia, ocular cicatricial pemphigoid, pptic neuritis, relapsing rheumatoid arthritis (PR), PANDAS, paraneoplastic cerebellar degeneration (PCD), paroxysmal nocturnal hemoglobinuria (PNH), Parry-Romberg syndrome (PNH), Romberg syndrome, pars planitis (peripheral uveitis), Parsonage-Turner syndrome, pemphigus, peripheral neuropathy, perivenous encephalomyelitis, pernicious anemia (PA), POEMS syndrome, polyarteritis nodosa, polyglandular syndrome types I, II, and III, polymyalgia rheumatica, polymyositis, post-myocardial infarction syndrome, post-pericardiotomy syndrome, primary biliary cirrhosis, primary sclerosing cholangitis, progesterone dermatitis, psoriasis, psoriatic arthritis, pure red cell aplasia (PRCA), pyoderma gangrenosum, Raynaud's phenomenon, reactive arthritis, reflex sympathetic dystrophy, relapsing polychondritis, restless legs syndrome (RLS) ), retroperitoneal fibrosis, rheumatic fever, rheumatoid arthritis, sarcoidosis, Schmidt's syndrome, scleritis, scleroderma, Sjogren's syndrome, sperm & testicular autoimmunity, stiff-body syndrome (SPS), subacute bacterial endocarditis (SBE), Susac's syndrome, sympathetic ophthalmia (SO), Takayasu's arteritis, temporal arteritis / giant cell arteritis, thrombocytopenic purpura (TTP), Tolosa-Hunt syndrome (THS), transverse myelitis, type 1 diabetes mellitus, ulcerative colitis (UC), undifferentiated connective tissue disease (UCTD), uveitis, vasculitis, vitiligo, or Vogt-Koyanagi-Harada disease.

[0103] C. Development of classification signatures The present disclosure provides a computer-implemented method for creating a classifier for a condition from a sample from a subject (see, for example, FIG. 2 and FIG. 3). A list of informative microsatellite loci can be created by statistically analyzing samples obtained or derived from a first group of subjects with a condition and / or samples obtained or derived from a second group of subjects without the condition (e.g., cancer, such as lung cancer). DNA from both groups of samples can be sequenced in a multiplex platform. In some cases, targeted sequencing is performed by enrichment of certain targets. Sequencing results can then be analyzed for quality and mapped to reveal differences between the cancer samples and the control or reference. This difference can then be analyzed using a computer-implemented method to create a classifier. The classifier can be further optimized and validated with additional samples obtained or derived from subjects with the condition and / or samples obtained or derived from subjects without the condition. In some embodiments, a list of informative genetic markers other than microsatellites can be created by these methods for developing classification signatures.

[0104] The condition can indicate the presence or absence of a health condition in a subject. In some cases, the condition can indicate an increased or decreased likelihood of development of the health condition in a subject. In some instances, the condition can indicate an increased or decreased likelihood that a subject will benefit from a treatment, or an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment (a classifier for a condition can serve as a companion diagnostic for a therapeutic agent). In some cases, the condition can indicate responsiveness to a treatment for a health condition in a subject. In some instances, the condition can indicate a prognosis of a health condition in a subject. In some cases, the classifier can be a value, e.g., a number. For example, the value can indicate an increased or decreased likelihood (e.g., a probability value between 0 and 1). The value of the classifier, e.g., a number, can be compared to a threshold, e.g., a number. In some instances, the distance of the classifier value from the threshold can indicate an increased confidence or probability of having or not having the condition being true. In some cases, a call is made if the classifier value is about 0.5, 1, 1.5, 2, 2.5, 3, or more than 3 standard deviations from the threshold value (Figure 24).

[0105] The computer-implemented method for creating a classifier can perform processing, combining, statistical evaluation or further result analysis, or any combination thereof. The computer-implemented method can include supervised or unsupervised learning methods, including support vector machines (SVM), neural networks, random forests, clustering algorithms (or software modules), gradient boosting, linear regression, logistic regression, and / or decision trees. A supervised learning algorithm can be an algorithm that relies on the use of a set of labeled and paired training data examples to infer relationships between input data and output data. An unsupervised learning algorithm can be an algorithm that is used to draw inferences from a training data set to output data. An unsupervised learning algorithm can include cluster analysis, which can be used for exploratory data analysis to find hidden patterns or groupings in processed data. An example of an unsupervised learning method is principal component analysis. Principal component analysis can include a reduction in the dimensionality of a set of one or more variables. The dimensionality of a given set of variables can be at least 1, 5, 10, 50, 100, 200, 300, 400, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, or more than 1800. The dimensionality of a given set of variables can be at most 1800, 1600, 1500, 1400, 1300, 1200, 1100, 1000, 900, 800, 700, 600, 500, 400, 300, 200, 100, 50, 10, or less than 10.

[0106] The computer-implemented method can include performing statistical techniques, which in some instances can include linear regression, classification, resampling methods, subset selection, shrinkage, dimensionality reduction, non-linear models, tree-based methods, support vector machines, unsupervised learning, or any combination thereof.

[0107] Linear regression can be a method for predicting a target variable by fitting the best linear relationship between dependent and independent variables. The best fit can correspond to a least squares approach, such that the sum of all distances between the shape and the actual observations at each point is minimized. Linear regression can include simple linear regression and multiple linear regression. Simple linear regression can predict a dependent variable using a single independent variable. Multiple linear regression can predict a dependent variable using two or more independent variables by fitting the best linear relationship.

[0108] Classification can be a data mining technique that assigns categories to a collection of data to achieve accurate predictions and analysis. Classification techniques can include logistic regression and discriminant analysis. Logistic regression can be used when the dependent variable is dichotomous (binary). Logistic regression can be used to discover and describe relationships between one dependent binary variable and one or more nominal, ordinal, interval or ratio level independent variables. Resampling can be a method that includes deriving replicate samples from the original data sample. In some cases, resampling may not involve the use of general distribution tables to compute approximate probability values. Resampling can create a unique sampling distribution based on the actual data. In some cases, resampling may use empirical rather than analytical methods to create a unique sampling distribution. Resampling techniques can include bootstrapping and cross-validation. Bootstrapping can be done by sampling with replacement from the original data, with the "unselected" data points being test cases. Cross-validation can be done by splitting the training data into multiple portions.

[0109] Subset selection can identify a subset of predictors for the response. Subset selection can include best subset selection, forward stepwise selection, backward stepwise selection, hybrid methods, or any combination thereof. In some examples, shrinkage fits a model with all predictors involved, but the estimated coefficients are shrunk towards zero compared to least squares estimation. This shrinkage can reduce the variance. Shrinkage can include ridge regression and lasso. Dimension reduction can reduce the problem of estimating n + 1 coefficients to a simpler problem of estimating m + 1 coefficients, where m < n. This can be achieved by computationally processing n different linear combinations or projections of the variables. Next, such n projections can be used as predictors to fit a linear regression model, for example, by least squares. Dimension reduction can include principal component regression and partial least squares. Principal component regression can be used to obtain a low-dimensional set of features from a large set of variables. The principal components used in principal component regression can then capture most of the variance in the data using linear combinations of the data in the directions of orthogonality. Partial least squares can be used as a supervised alternative to principal component regression because it can utilize the response variable to identify new features.

[0110] Nonlinear regression can be a form of regression analysis in which the observed data are modeled by a function that is a nonlinear combination of model parameters and depends on one or more independent variables. Nonlinear regression can include step functions, piecewise functions, splines, generalized additive models, or any combination thereof.

[0111] Tree-based methods can be used for both regression and classification problems. Regression and classification problems can involve stratification or partitioning of the predictor space into multiple simple regions. Tree-based methods can include bagging, boosting, random forests, or any combination of these. Bagging can reduce the variance of predictions by creating additional data for training from the original data set using a combination with iterations to produce multiple stages of the same carnality / size as the original data. Boosting can calculate the output using several different models and then average the results using a weighted average approach. Random forest algorithms can lead to random bootstrap samples of the training set. Support vector machines can be used for classification techniques. Support vector machines can include finding a hyperplane that best separates two classes of points with the maximum margin. Support vector machines can constrain the optimization problem such that the margin is maximized, subject to the constraint of perfectly classifying the data.

[0112] An unsupervised method can be a method for drawing inferences from a dataset that includes input data without labeled responses. Unsupervised methods can include clustering, principal component analysis, k-means clustering, hierarchical clustering, or any combination thereof.

[0113] 1. Genetic Algorithm In some embodiments, the computer-implemented method for creating a classifier includes the use of a genetic algorithm. The method can include creating an initial population of a subset of microsatellite loci (informative loci) associated or correlated with a condition by identifying microsatellite loci from samples with the condition that are different from microsatellite loci from samples without the condition. A genetic algorithm can be used to determine a classification signature based on the informative loci. The genetic algorithm can select the most informative subset of microsatellite loci to be included in the final classifier. The genetic algorithm can assign a weight to each subset. Weighting can be combined with other weighting schemes, such as proportionality to the relative risk of each microsatellite locus. Each subset of microsatellites can be iteratively ranked based on the association or correlation of the subset with the condition. The subset of the initial population of microsatellite loci can then be optimized by comparing the initial population with additional samples obtained or derived from subjects with the condition and / or subjects without the condition. In some cases, an initial population of about 100 subsets is used in the optimization. In some cases, an initial population of at least 100, 200, 300, 400, or 500 subsets are used in the optimization. In some instances, the optimization includes at least one cycle of comparing about 100 subsets to additional samples. In some instances, the optimization includes multiple cycles of comparing about 100 subsets to additional samples. Each subset can include at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, or 100 microsatellites.

[0114] Iterative ranking can be performed after each cycle is completed. In some cases, iterative ranking includes performing statistical analysis of the subsets for receiver operating characteristic (ROC) analysis of the accuracy, sensitivity and specificity in determining the presence or absence of a condition in additional samples. A predetermined number (e.g., 10) of the worst performing or lowest ranked subsets in terms of indicating the presence or absence of a condition can be identified and discarded. New subsets can be added to the population of subsets to maintain a constant number of subsets before the start of each cycle of optimization. In some cases, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 or more than 10 new subsets are created by randomly splitting and recombining two randomly selected subsets from a previous cycle of optimization. In some instances, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 or more than 10 new subsets are randomly selected from a previous cycle of optimization. In some instances where 10 new subsets are added, 3 are created from randomly splitting and recombining 2 randomly selected subsets from the previous cycle of optimization, and 7 are randomly selected from the subsets from the previous cycle of optimization. In some instances where 10 new subsets are added, 4 are created from randomly splitting and recombining 2 randomly selected subsets from the previous cycle of optimization, and 6 are randomly selected from the subsets from the previous cycle of optimization. In some instances where 10 new subsets are added, 5 are created from randomly splitting and recombining 2 randomly selected subsets from the previous cycle of optimization, and 5 are randomly selected from the subsets from the previous cycle of optimization. In some instances where 10 new subsets are added, 6 are created from randomly splitting and recombining 2 randomly selected subsets from the previous cycle of optimization, and 4 are randomly selected from the subsets from the previous cycle of optimization.In some instances where 10 new subsets are added, 6 are created from randomly splitting and recombining 2 randomly selected subsets from a previous cycle of optimization, and 4 are randomly selected from subsets from a previous cycle of optimization. In some instances where 10 new subsets are added, 7 are created from randomly splitting and recombining 2 randomly selected subsets from a previous cycle of optimization, and 3 are randomly selected from subsets from a previous cycle of optimization. Overlap of the new subsets may be included in the cycle of optimization. In some instances, overlap of the new subsets is not included in the cycle of optimization.

[0115] In some cases, the number of subsets discarded at the end of each cycle of optimization is the same as the number of subsets added to the subsets prior to each cycle of optimization. In some cases, the 5 lowest ranked subsets are discarded at the end of each cycle of optimization while 5 new subsets are added prior to each cycle of optimization. In some cases, the 10 lowest ranked subsets are discarded at the end of each cycle of optimization while 10 new subsets are added prior to each cycle of optimization. In some cases, the 20 lowest ranked subsets are discarded at the end of each cycle of optimization while 20 new subsets are added prior to each cycle of optimization. In some cases, the 50 lowest ranked subsets are discarded at the end of each cycle of optimization while 50 new subsets are added prior to each cycle of optimization.

[0116] In some aspects, the computer-implemented method for creating a classifier includes determining a statistically unweighted subset of microsatellites. In some aspects, the computer-implemented method for creating a classifier includes determining a statistically weighted subset of microsatellites. In some cases, the weighted subset is weighted by a relative risk, risk ratio, or odds ratio. The classifier may be unweighted or weighted. In some cases, the classifier created by the computer-implemented method described above can be based on genetic markers other than microsatellites. In some cases, the classifier can be based on other genomic information, such as single nucleotide polymorphisms (SNPs) or genetic abnormalities, such as copy number abnormalities, indels, etc. In some cases, the classifier can be based on the identity of the gene in which the microsatellite is located.

[0117] After completing the cycles of optimization, the computer-implemented method can include determining microsatellites associated or correlated with the condition with optimized accuracy, sensitivity, and specificity. In some embodiments, the computer-implemented method can be validated with additional sets of samples including samples with the condition, samples without the condition, or a combination thereof (see, e.g., FIG. 3). Validation can include using at least 10, 20, 30, 50, 100, or 1000 samples from subjects with the condition, e.g., cancer (the samples can be non-tumor (germline) samples or tumor samples), and at least 10, 20, 30, 50, 100, or 1000 samples from subjects without the condition, e.g., cancer, e.g., lung cancer.

[0118] The optimized and validated computer-implemented method can generate a classifier for a condition when analyzing a sample from a subject. The condition can indicate the presence or absence of a health condition in a subject. In some cases, the condition can indicate an increased or decreased likelihood of the development of a health condition in a subject. In some instances, the condition can indicate an increased or decreased likelihood that a subject will benefit from a treatment, or an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment. In some cases, the condition can indicate the responsiveness of a subject to a treatment for a health condition. In some instances, the condition can indicate the prognosis of a health condition in a subject.

[0119] The status can indicate the presence or absence of cancer. In some cases, the status indicates an increased or decreased likelihood of developing cancer. In some instances, the status indicates an increased or decreased likelihood that a subject will benefit from a treatment, or an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment (the classifier can be a companion diagnostic for cancer treatment). In some cases, the status can indicate responsiveness to a treatment for cancer. The treatment can be surgery, chemotherapy, radiation, targeted treatment with a drug (e.g., afatinib, gefitinib, bevacizumab, crizotinib, or ceritinib), or immunotherapy (e.g., treatment with a monoclonal antibody, checkpoint inhibitor, therapeutic vaccine, or adoptive T cell transfer). In some instances, the status indicates a prognosis of cancer. In some cases, the cancer is lung cancer, including non-small cell lung cancer (e.g., lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and large cell carcinoma), small cell lung cancer (SCLC), or lung carcinoid tumor.

[0120] The classifier can include microsatellite loci from any chromosome, e.g., chromosome 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, X, or Y. In some cases, the classifier does not contain microsatellite loci from chromosome X and / or Y.

[0121] III. Creating a Weighted Classifier for the States The present disclosure provides a method of weighting microsatellite loci identified to be associated or correlated with a condition. The present disclosure also provides a method of weighting genetic markers other than microsatellite loci identified to be associated or correlated with a condition. Weight or weighting can refer to the relative importance or prevalence of each individual microsatellite locus that statistically contributes to the association or correlation to the condition. For example, high weights can be assigned to both microsatellite loci that appear only in samples obtained from subjects with the condition and those that appear there at a higher frequency. In some cases, weights are assigned based on risk ratios, odds ratios, or relative risks. Examples of numerical components that are part of the weight determination include sensitivity, specificity, negative predictive value, positive predictive value, odds ratios, hazard ratios, or any combination thereof. In some cases, a cutoff (e.g., threshold) is imposed on the numerical components used to calculate the weight. Samples with numerical classifiers below the cutoff can be excluded from the weight calculation. Weights can be calculated based on a combination of linear, non-linear, algebraic, trigonometric, statistical learning, Bayesian, regression or correlational means of calculation. A weighting scheme using values ​​(e.g., relative risk) associated with a microsatellite or set of microsatellites, or a regression approach can be used to create a classifier. Weighted classifiers can be evaluated to determine whether adding weights improves the sensitivity or specificity of the classifier. To maximize sensitivity and specificity (e.g., the sum of sensitivity and specificity), optimal weights can be computed for each locus using regression analysis (e.g., standard regression analysis).

[0122] In some cases, the weight assigned to each microsatellite is a predetermined value, which indicates the sample size or the strength of the association or correlation between the condition and the microsatellite locus. In certain instances, the weight assigned to each microsatellite comprises a relative risk, a risk ratio, or an odds ratio. In some instances, the predetermined value of the weight determines the numerical range of the sensitivity, specificity, or a combination thereof (e.g., the sum). In some instances, the calculation and assignment of weights includes decision-making models performed by a computer via models such as Support Vector Machines, Decision Trees, Random Forests, Neural Networks or Deep Learning Neural Networks (e.g., Artificial Neural Networks, Recurrent Neural Networks, Convolutional Neural Networks, Cognitive, Feedforward, Radial Basis Networks, Deep Feedforward, Recurrent Neural Networks, Long / Short-Term Memory, Gated Recurrent Units, Autoencoders (AE), Deformed Form AE, Denoising AE, Sparse AE, Markov Chains, Hopfield Networks, Boltzmann Machines, Restricted BM, Deep Belief Networks, Deep Convolutional Networks, Deconvolutional Networks, Deep Convolutional Inverse Graphics Networks, Generative Adversarial Networks, Liquid State Machines, Extreme Learning Machines, Each State Networks, Deep Residual Networks, Kohonen Networks, Support Vector Machines, and Neural Turing Machines).

[0123] In some instances, the weights assigned to the microsatellite loci are used as part of the calculation for the classifiers described herein. In such instances, microsatellite loci with larger weights can contribute more to the value of the classifier than microsatellite loci with smaller weights. In some cases, the calculation of the classifier includes using only optimal weights. Optimal weights can include weights that are at least or exceed a predetermined threshold.

[0124] The status determined by the weighted classifier can indicate the presence or absence of a health condition in a subject. In some cases, the status determined by the weighted classifier indicates an increased or decreased likelihood of the development of a health condition in a subject. In some instances, the status determined by the weighted classifier indicates an increased or decreased likelihood that a subject will benefit from a treatment, or an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment. In some instances, the status determined by the weighted classifier indicates the responsiveness to a treatment for a health condition in a subject. In other instances, the status determined by the weighted classifier can indicate a prognosis of a health condition in a subject. In some cases, the health condition is cancer. In some cases, the cancer is lung cancer, such as non-small cell lung cancer (e.g., lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC) and large cell carcinoma), small cell lung cancer (NSLC) or lung carcinoid tumor.

[0125] Classifiers can also be determined based on minor allele distribution, for example, of microsatellites. In some cases, classifiers can be determined by calculating a weighted combination of informative microsatellite loci and minor allele distribution. Minor allele frequency can be an additional weighted parameter for the classifier. Minor allele frequency can be an indicator of overall genome stability. Classifiers based on minor allele frequency can be statistically evaluated (e.g., by regression analysis) to determine whether adding minor allele frequency to the classifier improves the classifier.

[0126] IV. Pan-Condition (e.g., Cancer) Risk Assays The present disclosure provides a computer-implemented method for creating a pan-state (e.g., cancer) classifier (see, e.g., FIG. 2 and FIG. 4). An informative microsatellite locus list can be created from statistically analyzing samples of various state (e.g., cancer) types and healthy reference sequences. DNA from both groups of samples can be sequenced in a multiplex platform. In some cases, sequencing is targeted with additional enrichment, for example, using a bait set. Sequencing results are then analyzed for quality and mapped to reveal differences between state (e.g., cancer) samples and the reference. The differences can be analyzed to create a pan-state (e.g., cancer) classifier. The pan-state (e.g., cancer) classifier can be further optimized and validated with additional samples of various types of state, e.g., cancer.

[0127] A pan-condition (e.g., pan-cancer) classifier for a condition or conditions can indicate the presence or absence of at least one of the plurality of health conditions in a subject, an increased or decreased likelihood of development of at least one of the plurality of health conditions in a subject, an increased or decreased likelihood that a subject will benefit from a treatment for at least one of the plurality of health conditions, an increased or decreased likelihood that a subject will have an increased risk of an adverse event from a treatment for at least one of the plurality of health conditions, a subject's responsiveness to a treatment for at least one of the plurality of health conditions, or a combination thereof. The plurality of health conditions can be any combination of health conditions disclosed herein.

[0128] In some cases, the pan-cancer status can indicate the presence or absence of multiple types of cancer in a subject. In some instances, the pan-cancer status can indicate an increased or decreased likelihood of developing multiple types of cancer in a subject. In certain cases, the multiple types of cancer are cancers that occur frequently together in the same subject. In alternative cases, the multiple types of cancer are cancers that occur independently. In some instances, the pan-cancer status can indicate that a subject is more likely or less likely to benefit from a treatment, or that a subject is more likely or less likely to have an increased risk of adverse events as a result of a treatment (the pan-cancer classifier can be a companion diagnostic for a therapeutic product). In some instances, the pan-cancer status can indicate responsiveness to a treatment for cancer in a subject. In other instances, the pan-cancer status can indicate a prognosis of cancer in a subject. The subjects described herein can be either symptomatic or asymptomatic for cancer. In some cases, additional tests (e.g., physical examination, analysis of circulating or acellular cancer biomarkers, imaging (e.g., computed tomography (CT), bone scan, magnetic resonance imaging (MRI), positron emission tomography (PET), ultrasound and x-ray), biopsy, genetic screening, gene or protein expression levels, etc.) can be used based on the pan-cancer classifier for the subject.

[0129] The computer-implemented method for creating a pan-status (e.g., pan-cancer) classifier can include processing, combining, statistical evaluation, or further result analysis, or any combination thereof. In some aspects, the computer-implemented method for creating a pan-status (e.g., cancer) classifier includes first creating a population of subsets of microsatellite loci that are associated or correlated with a multi-type condition (e.g., cancer) by identifying microsatellite loci from samples obtained or derived from subjects with a multi-type condition (e.g., cancer) that are different from microsatellite loci from samples obtained or derived from subjects without the multi-type condition (e.g., cancer). The sequences of the microsatellites can first be obtained by any sequencing method.

[0130] Microsatellite loci associated or correlated with multiple conditions (e.g., cancer) can be identified using one or more statistical tests, such as t-tests, Z-tests, ANOVA, regression analysis, Mann-Whitney-Wilcoxon, chi-squared tests, correlation, Fisher exact tests, Bonferroni corrections, and Benjamini-Hochberg tests.

[0131] The statistical test can generate a receiver operating characteristic (ROC) curve, and the area under the ROC curve is referred to as the area under the curve (AUC). The AUC can determine the accuracy of identifying microsatellite loci that are associated or correlated with multiple types of conditions (e.g., cancer). A larger AUC can indicate a higher accuracy of association or correlation. The ROC curve can determine the sensitivity (e.g., true positive) and specificity (e.g., true negative) rates for the association or correlation of microsatellite loci with multiple types of conditions (e.g., cancer). A statistically significant association or correlation of a microsatellite locus with multiple types of conditions (e.g., cancer) can have a statistical accuracy of at least about 70%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98% or 99%. In some cases, a statistically significant association or correlation of a microsatellite locus to a multitype condition (e.g., cancer) has a statistical specificity of at least 0.70, 0.80, 0.85, 0.90, 0.91, 0.92, 0.93, 0.94, 0.95, 0.96, 0.97, 0.98, or 0.99, and a statistical sensitivity of at least 0.70, 0.80, 0.85, 0.90, 0.91, 0.92, 0.93, 0.94, 0.95, 0.96, 0.97, 0.98, or 0.99.

[0132] In some instances, identifying microsatellite loci associated or correlated with multiple types of conditions (e.g., cancer) includes identifying a first set of microsatellite loci from a database comprising nucleic acid sequences of multiple types of conditions (e.g., cancer) and a second set of microsatellite loci from a reference database (e.g., hg19). In some cases, some of the microsatellites are identified as associated or correlated with multiple types of conditions (e.g., cancer). In some cases, some of the microsatellites are identified as associated or correlated with one type of condition (e.g., cancer).

[0133] The multiple types of cancer can include solid or hematological malignant types of cancer. In some cases, the multiple types of cancer can be metastatic, relapsing, or refractory. The multiple types of cancer associated or correlated with the identified microsatellite loci can include any number of cancers disclosed herein (e.g., about 4 to about 10, about 10 to about 15, about 15 to about 20, or about 4, about 10, about 15, about 20, about 25, about 30, or about 50).

[0134] The pan-cancer assay may assay or test for at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 or 16 of the following cancers: breast cancer, ovarian cancer, prostate cancer, lung cancer, glioblastoma multiforme, endometrial cancer of the uterine corpus, colon adenocarcinoma, bladder cancer, urothelial carcinoma, squamous cell carcinoma of the head and neck, squamous cell carcinoma and endocervical adenocarcinoma of the cervix, gastric adenocarcinoma, thyroid cancer, brain low-grade glioma, papillary renal cell carcinoma of the kidney and hepatocellular carcinoma of the liver.

[0135] In some cases, the types of cancer that are associated or correlated with a different set of microsatellite loci include lung cancer. Lung cancers that may be associated or correlated with a different set of microsatellite loci include non-small cell lung cancer (e.g., lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and large cell carcinoma), small cell lung cancer (SCLC), and lung carcinoid tumors.

[0136] A population of subsets that includes microsatellite loci associated or correlated with multiple types of conditions (e.g., cancer) can include at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, or 100 microsatellite loci per subset. In some embodiments, the population of subsets is iteratively ranked based on the association or correlation of the subsets with multiple types of conditions (e.g., cancer).

[0137] The subset of the population of microsatellite loci can then be optimized by comparing the population of subsets to additional samples obtained or derived from subjects with multiple types of conditions (e.g., cancer) and / or subjects without multiple types of conditions (e.g., cancer). In some cases, a population of about 100 subsets is used in the optimization. In some cases, a population of at least 100, 200, 300, 400, 500, 1000, 2000, 3000, or 5000 subsets is used in the optimization. In some instances, the optimization includes at least one cycle of comparing about 100 identified subsets to additional samples. In some instances, the optimization includes multiple cycles of comparing about 100 identified subsets to additional samples.

[0138] The iterative ranking can be performed after each cycle is completed. In some cases, the iterative ranking includes performing statistical analysis of the subsets for receiver operating characteristic (ROC) analysis for accuracy, sensitivity and specificity in determining the presence or absence of a multi-type condition (e.g., cancer) in additional samples. One or more of the worst performing or lowest ranked subsets in indicating the presence or absence of a multi-type condition (e.g., cancer) can be identified and discarded. To maintain a constant number of subsets before the start of each cycle of optimization, new subsets can be added to the population of subsets. In some cases, the new subsets are created from randomly splitting and recombining two randomly selected subsets from a previous cycle of optimization. In some instances, the new subsets are randomly selected from a previous cycle of optimization. In some cases, the number of subsets discarded at the end of each cycle of optimization is the same number of subsets added to the subsets prior to each cycle of optimization.

[0139] A computer-implemented method for creating a pan-status (e.g., pan-cancer) classifier can include determining a statistically unweighted subset of microsatellite loci. In some aspects, a computer-implemented method for creating a pan-status (e.g., pan-cancer) classifier includes determining a statistically weighted subset of microsatellite loci. A pan-status (e.g., pan-cancer) classifier can be unweighted or weighted.

[0140] After completing the optimization cycles, the computer-implemented method for generating a pan-status (e.g., pan-cancer) classifier includes microsatellite loci associated or correlated with the condition with optimized accuracy, sensitivity, and specificity. In some embodiments, the computer-implemented method can be validated with additional sets of samples including samples obtained or derived from subjects with multiple types of conditions (e.g., cancer), samples obtained or derived from subjects without multiple types of conditions (e.g., cancer), or combinations thereof. The optimized and validated computer-implemented method can generate a pan-status (e.g., pan-cancer classifier) ​​upon analyzing samples from a subject. The pan-status (e.g., pan-cancer) can indicate the presence or absence of a type of health condition (e.g., cancer) in a subject. In certain cases, the pan-status (e.g., pan-cancer) indicates an increased or decreased likelihood of developing a type of health condition (e.g., cancer) in a subject. In some cases, the pan-status (e.g., pan-cancer) can indicate an increased or decreased likelihood that a subject will benefit from a treatment, or an increased or decreased likelihood that a subject will have an increased risk of an adverse event as a result of a treatment (the pan-status, e.g., pan-cancer classifier, can be a companion diagnostic for a therapeutic product). In some instances, the pan-status (e.g., pan-cancer) indicates the responsiveness of a subject to a treatment for a certain type of health condition (e.g., cancer). In other instances, the pan-status (e.g., pan-cancer) indicates the prognosis of a certain type of health condition (e.g., cancer) in a subject.

[0141] Classifiers (e.g., sets of microsatellites) can be developed for each state (e.g., cancer) in a pan-state (e.g., pan-cancer) assay. In some cases, individual microsatellite loci can be pan-state (e.g., pan-cancer) microsatellite loci.

[0142] V. Evaluation of Subject-Derived Samples Classifiers created as described herein can be used to analyze subject (e.g., patient) samples. Samples from subjects can be analyzed, for example, in a Clinical Laboratory Improvements Amendments (CLIA) certified laboratory. In some cases, kits are prepared and samples from subjects are analyzed outside of the CLIA laboratory. FIG. 5 illustrates an example of a workflow (500) for a subject (e.g., patient) sample analysis pipeline, for example, in a CLIA certified laboratory; the workflow can be used to process samples for multiplexed pan-cancer assays. Samples, for example, blood, urine, cerebrospinal fluid, semen, saliva, sputum, feces, lymph, tissue (e.g., thyroid, skin, heart, lung, kidney, breast, pancreas, liver, muscle, smooth muscle, bladder, gallbladder, colon, intestine, brain, esophagus, or prostate, or any combination thereof, are obtained from multiple subjects (501). Nucleic acid molecules, for example, genomic DNA, are extracted from the samples. Targets, e.g., microsatellite targets, are enriched by multiplexing (e.g., using baits, e.g., hybridization probes); the enriched targets can be barcoded and amplified (503). Next-generation sequencing assays are performed on the target-enriched samples in batches of, e.g., about 4, 8, 12, 24, 96, 128, 384, or 1536 (505). The sequencing data can be de-multiplexed (e.g., using unique sequence tags (e.g., barcodes) attached to each individual sample), quality control filters can be applied to the raw sequence reads (e.g., Phred quality above Q30), genotyped (e.g., for each locus, reads are aligned to a reference sequence using flanking sequences and then the two major alleles (genotypes) are computed), and the minor allele distribution (e.g., the number of minor alleles or portions of minor alleles compared to the major genotype) is determined for each microsatellite locus for each sample (507) (minor alleles can be supported by at least 1, at least 2, at least 3, or more than 3 sequence reads).A risk classifier (e.g., based on at least 5, 10, 25, 50 or 100 microsatellite loci) for each sample per cancer is calculated (509) (e.g., genotypes are determined to be modal or non-modal with respect to the most prominent genotype in a healthy population (e.g., GRCh38) genotypes are summed across all loci, and the sample can be classified as at risk or not at risk of the condition depending on where the locus is located with respect to the cutoff for the fraction of loci that have cancer or normal genotypes). Risk can be expressed on a quantitative scale or indicated by a categorical assessment. A clinical laboratory report including the risk classifier is generated (511) and provided to the health care provider, subject, or insurance company.

[0143] Figure 17 illustrates an example of a clinical laboratory report. The clinical laboratory report can include patient information, specimen information, test summary, test results, comments, and result details. The result details can include the number of microsatellite loci genotyped, one or more risk classifiers for the condition, one or more thresholds, and the relative risk of having or acquiring a condition, e.g., lung cancer (e.g., low risk, high risk, "at risk," "not at risk").

[0144] The report can include the number of loci in the subject's sample that have a non-modal (mostly cancer) genotype. The sensitivity and specificity for detection of the presence of a health condition determined to be high risk can be greater than 90%, and absence in these control sample germlines is determined to be "low risk" for lung cancer. The accuracy of the assay can be greater than 99%, as measured by highly conserved loci in the reference control.

[0145] In some instances, the condition can be verified or further tested by additional tests, such as a physical examination, analysis of circulating or acellular cancer biomarkers, imaging (e.g., computed tomography, bone scan, magnetic resonance imaging, positron emission tomography, ultrasound and x-ray), biopsy, genetic screening, gene expression or protein expression, etc.

[0146] VI. Minor alleles in microsatellites The present disclosure provides a computer-implemented method for determining a subject's genomic age and rate of genomic aging. Genomic age can be expressed in a number calibrated to years. For example, if the genomic age is approximately equal to the subject's numerical age, the overall genomic stability can be normal for the genomic age. In some instances, the genomic age can be younger, the same, or older than the subject's actual age. A genomic age that is older than the subject's actual age, or a faster rate of genomic aging can indicate genomic instability and susceptibility to developing health conditions (e.g., diseases) associated with aging, such as cancer, cardiovascular disease, neurological disease, and the like. Genomic age and rate of genomic aging can vary between samples obtained from different tissues (e.g., skin or blood) of the same subject. In some cases, genomic age and rate of genomic aging can indicate the subject's lifestyle (e.g., nutrition, physical or mental stress) or medical condition. Based on the subject's genomic age, lifestyle changes (e.g., smoking cessation, dietary changes, and exercise) may be recommended to the subject.

[0147] A computer-implemented method for determining genomic age and rate of genomic aging can include determining a minor allele signature in a first sample from a subject, and comparing the minor allele signature of the first sample with the minor allele signature of a reference to obtain a first difference in the minor allele signature. The reference can include a distribution of minor allele content across a large population to determine an average genomic age as a function of numerical age, ethnicity, gender, etc. The first difference in the minor allele signature between the first sample and the reference can be determined by the computer-implemented method to be the genomic age of the subject. In some embodiments, a second sample from the subject is compared to the reference at a time point after the comparison of the reference to the first sample to obtain a second difference in the minor allele signature. The change between the first and second differences can be determined by the computer-implemented method to be the rate of genomic aging of the subject. In some cases, an additional rate of genomic aging can be determined by obtaining the minor allele signature and comparing the later minor allele signature with the earlier minor allele signature.

[0148] The minor allele characteristic described herein can be the number of minor alleles at at least one locus.In some embodiments, the minor allele characteristic comprises the percentage of SNPs, the percentage of expansion, the percentage of contraction, the ratio of expansion and contraction to SNPs, the percentage of heterozygous loci, the percentage of homozygous loci, and the percentage of loci with minor alleles.In some cases, the minor allele characteristic comprises the combination of SNPs and indel variants, microsatellite variants, synonymous SNPs, non-synonymous SNPs, stop-gain SNPs, stop-loss SNPs, splicing variants (e.g., 2bp in splicing junction), frameshift indels, and non-frameshift indels at at least one locus.In some cases, the minor allele characteristic is determined in the same subject over multiple time points.

[0149] The minor allele characteristic determined from the sample from the subject can require at least one sequence read from any sequencing method.In some cases, the minor allele characteristic can be identified in at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 20, 30, 50 or 100 sequence reads from any next-generation sequencing method.The minor allele characteristic determined from the sample from the subject can require at least 1, at least 2, at least 3 or more than 3 sequence reads from any sequencing method.

[0150] In some examples, the minor allele characteristics determined from the sequence of the sample from subject are compared with reference sequence.Comparing can produce the difference of minor allele characteristics from reference sequence, including different number of SNPs and indel variants at at least one locus, microsatellite variants, synonymous SNPs, non-synonymous SNPs, stop-gain SNPs, stop-loss SNPs, splicing variants (e.g., 2bp in splicing junction), frameshift indels and non-frameshift indels in combination.The difference of minor allele characteristics between sample and reference can be determined by computer-implemented method to obtain genomic age.

[0151] In some cases, a first sequence of a first sample from a subject is compared with a reference sequence to obtain a first minor allele signature and a first genomic age. In some examples, a second sequence of a second sample from the same subject is compared with the same reference sequence to obtain a second minor allele signature and a second genomic age. The comparison between the first minor allele signature and the second minor allele signature can determine the rate of genomic aging. In certain examples, multiple minor allele signatures can be obtained from samples from the same subject at later time points for comparison, to obtain multiple rates of genomic aging at different ages of the subject.

[0152] The present disclosure provides a computer-implemented method for determining the genomic age of a subject by determining a microsatellite minor allele feature in a first sample from the subject.The microsatellite minor allele feature can be a minor allele that includes a microsatellite with a different SNP percentage, expansion percentage, contraction percentage, expansion and contraction ratio to SNP, heterozygous locus percentage, or homozygous locus percentage compared to a reference sequence.In some cases, the microsatellite minor allele feature includes a minor allele that includes a microsatellite with a different combination of SNP and indel variants, microsatellite variants, synonymous SNPs, non-synonymous SNPs, stop-gain SNPs, stop-loss SNPs, splicing variants (e.g., 2 bp in a splicing junction), frameshift indels, or non-frameshift indels at at least one locus compared to a reference sequence.In some cases, the microsatellite minor allele feature is determined across multiple time points in the same subject.

[0153] VI. Computer Systems, Processors and Memory The present disclosure provides a computer system configured to perform the method described in the present disclosure. In some examples, a system is disclosed herein that includes a computer processing device, optionally connected to a computer network, and a software module executed by the computer processing device. In some examples, the system includes a central processing unit (CPU), a memory (e.g., random access memory, flash memory), an electronic storage device, a computer program, a communication interface for communicating with one or more other systems, and any combination thereof. In some examples, the system is coupled to a computer network, such as the Internet, an intranet and / or an extranet in communication with the Internet, a telecommunication, or a data network. In some aspects, the system includes a storage device for storing data and information related to any aspect of the method described in the present disclosure. Various aspects of the system are products or articles of manufacture.

[0154] One feature of a computer program includes sequences of instructions written to perform specified tasks and executable by a CPU of a digital processing device. In some aspects, computer readable instructions are implemented as program modules, such as functions, features, application programming interfaces (APIs), data structures, etc., that perform particular tasks or implement particular abstract data types. In various embodiments, computer programs can be written in various languages ​​and different versions.

[0155] The functionality of the computer-readable instructions may be combined or distributed as desired in various environments. In some instances, a computer program includes one instruction sequence or multiple instruction sequences. A computer program may be provided from one location. A computer program may be provided from multiple locations. In some aspects, a computer program includes one or more software modules. In some aspects, a computer program includes, in part or in whole, one or more web applications, one or more mobile applications, one or more standalone applications, one or more web browser plug-ins, extensions, add-ins or add-ons, or combinations thereof.

[0156] Computer Systems The present disclosure provides a computer system programmed to carry out the methods of the present disclosure. Figure 18 shows a computer system (1801) that can be programmed or otherwise configured to carry out the methods described herein. The computer system (1801) can accommodate various aspects of the present disclosure, including inputting nucleic acid position information, transferring attribution information to a dataset, and creating an algorithm trained with the dataset. The computer system (1801) can be a user electronic device or a remote computer system. The electronic device can be a mobile electronic device.

[0157] The computer system (1801) includes a central processing unit (CPU, also referred to herein as "processor" and "computer processor") (1805), which can be a single-core or multi-core processor with either sequential or parallel processing. The computer system (1801) also includes a memory device or device (1810) (e.g., random access memory, read-only memory, flash memory), a storage device (1815) (e.g., hard disk), a communication interface (1820) (e.g., network adapter) for communicating with one or more other systems, and peripheral devices (1825), either external or internal, or both, such as a printer, a monitor, a USB drive, and / or a CD-ROM drive. The memory (1810), the storage device (1815), the interface (1820), and the peripheral devices (1825) are in communication with the CPU (1805) via a communication bus (solid line), such as a motherboard. The storage device (1815) can be a data storage device (or data repository) for storing data. The computer system (1801) may be operatively coupled to a computer network ("network") (1830) utilizing the communication interface (1820). The network (1830) may be the Internet, an Internet and / or an extranet, or an intranet and / or an extranet in communication with the Internet. The network (1830) may, in some cases, be a telecommunications and / or data network. The network (1830) may include one or more computer servers that may enable a peer-to-peer network supporting distributed computer processing. The network (1830) may, in some cases, utilize the computer system (1801) to implement a client-server architecture that may enable devices coupled to the computer system (1801) to act as clients or servers.

[0158] The CPU (1805) can execute sequences of machine-readable instructions, which may be embodied in a program or software. The instructions may be stored in memory (1810). The instructions may be directed to the CPU (1805), which may then program or otherwise configure the CPU (1805) to perform methods of the present disclosure. Examples of operations performed by the CPU (1805) may include fetch, decode, execute, and writeback.

[0159] The CPU (1805) can be part of a circuit, such as an integrated circuit. One or more other components of the system (1801) can be included in the circuit. In some embodiments, the circuit is an application specific integrated circuit (ASIC).

[0160] The storage device (1815) can store files such as drivers, libraries, and saved programs. The storage device (1815) can store user data, such as user settings and user programs. The computer system (1801) can include one or more additional data storage devices that are, in some cases, external to the computer system (1801), such as located on a remote server in communication with the computer system (1801) via an intranet or the Internet.

[0161] The computer system (1801) can communicate with one or more remote computer systems via a network (1830). For example, the computer system (1801) can communicate with a remote computer system or user. Examples of remote computer systems include personal computers (e.g., portable PCs), slate or tablet PCs (e.g., Apple® iPad®, Samsung® Galaxy Tab), phones, smartphones (e.g., Apple® iPhone®, Android-enabled devices, Blackberry®), or personal digital assistants. A user can access the computer system (1801) via the network (1830).

[0162] The methods described herein can be implemented as machine (e.g., computer processor) executable code stored in an electronic storage location of a computer system (1801), for example, in memory (1810) or data storage (1815). The machine executable or machine readable code can be provided in the form of software. In use, the code can be executed by the processor (1805). In some cases, the code can be retrieved from storage (1815) and stored in memory (1810) for immediate access by the processor (1805). In some aspects, storage (1815) can be excluded and the machine executable instructions are stored in memory (1810).

[0163] The code may be pre-compiled and configured for use by a machine having a processor adapted to execute the code, or may be compiled during run-time. The code may be provided in a programming language that may be selected to allow the code to execute in a pre-compiled or as-compiled manner.

[0164] Aspects of the systems and methods provided herein, such as the computer system (1801), may be incorporated into programming. Various aspects of the technology may be considered as "products" or "articles of manufacture," typically in the form of machine (or processor) executable code and / or associated data carried or embodied in some type of machine-readable medium. The machine-executable code may be stored in a storage device such as a hard disk or memory (e.g., read-only memory, random access memory, flash memory). A "storage" type medium may include any or all of the tangible memory of a computer, processor, or other or their associated modules, including various semiconductor memories, tape drives, disk drives, and the like, which may provide non-transitory storage at any time for software programming. The entire or portions of the software may sometimes be communicated over the Internet or various other telecommunications networks. Such communication may enable, for example, loading of the software from one computer or processor to another, for example, from a management server or host computer to the computer platform of an application server. Thus, another type of medium that may bear software elements includes light waves, radio waves, and electromagnetic waves, such as those used across physical interfaces between local devices, via wired and optical landline networks, and through various air-links. The physical elements that carry such waves, such as wired or wireless links, optical links, and the like, may also be considered as media bearing the software. As used herein, unless limited to non-transitory, tangible "storage" media, terms such as computer or machine "readable medium" refer to any medium that participates in providing instructions to a processor for execution.

[0165] A. Electronic Devices In some aspects, the platforms, media, methods and applications described herein include electronic devices, processors, or uses thereof (also referred to as digital processing devices). In further aspects, the electronic device includes one or more hardware central processing units (CPUs) that execute the functions of the device. In still further aspects, the electronic device further includes an operating system configured to execute executable instructions. In some aspects, the electronic device is optionally connected to a computer network. In further aspects, the electronic device is optionally connected to the Internet to access the World Wide Web. In still further aspects, the electronic device is optionally connected to a cloud computing infrastructure. In some aspects, the electronic device is optionally connected to an intranet. In some aspects, the electronic device is optionally connected to a data storage device. In accordance with the description herein, suitable electronic devices include, by way of non-limiting examples, server computers, desktop computers, laptop computers, notebook computers, sub-notebook computers, netbook computers, net pad computers, set-top computers, handheld computers, Internet appliances, mobile smartphones, tablet computers, personal digital assistants, video game consoles, and vehicles. In various embodiments, many smart phones are suitable for use in the systems described herein. In various embodiments, select televisions, video players and digital music players with optional computer network connectivity are suitable for use in the systems described herein. Suitable tablet computers include those having booklet, slate and convertible configurations.

[0166] In some aspects, the electronic device includes an operating system configured to execute executable instructions. An operating system is software including, for example, programs and data that manages the device's hardware and provides services for the execution of applications. In various embodiments, suitable server operating systems include, by way of non-limiting examples, FreeBSD, OpenBSD, NetBSD®, Linux®, Ubuntu Linux®, Apple® Mac OS X Server®, Oracle® Solaris®, Windows Server®, and Novell® NetWare®. In various embodiments, suitable personal computer operating systems include, by way of non-limiting examples, UNIX-like operating systems such as Microsoft® Windows®, Apple® Mac OS X®, UNIX®, and GNU / Linux®. In some aspects, the operating system is provided by cloud computing. In various embodiments, suitable mobile smartphone operating systems include, by way of non-limiting examples, Nokia® Symbian® OS, Apple® iOS®, Research In Motion® BlackBerry OS®, Google® Android®, Microsoft® Windows Phone® OS, Microsoft® Windows Mobile® OS, Linux®, and Palm® WebOS®.

[0167] In some embodiments, the device includes a storage and / or memory device. A storage and / or memory device is one or more physical devices used to store data or programs, either transiently or permanently. In some embodiments, the device is a volatile memory, requiring a power supply to maintain the stored information. In some embodiments, the device is a non-volatile memory, retaining the stored information when the electronic device is not powered. In further embodiments, the non-volatile memory includes a flash memory. In some embodiments, the non-volatile memory includes a dynamic random access memory (DRAM). In some embodiments, the non-volatile memory includes a ferroelectric random access memory (FRAM®). In some embodiments, the non-volatile memory includes a phase change random access memory (PRAM). In some embodiments, the non-volatile memory includes a magnetoresistive random access memory (MRAM). In some embodiments, the device is a storage device, including, as non-limiting examples, CD-ROM, DVD, flash memory device, magnetic disk drive, magnetic tape drive, optical disk drive, and cloud computing based storage. In further embodiments, the storage and / or memory device is a combination of devices, such as those disclosed herein.

[0168] In some embodiments, the electronic device includes a display for sending visual information to a subject. In some embodiments, the display is a cathode ray tube (CRT). In some embodiments, the display is a liquid crystal display (LCD). In further embodiments, the display is a thin film transistor liquid crystal display (TFT-LCD). In some embodiments, the display is an organic light emitting diode (OLED) display. In various further embodiments, the OLED display is a passive matrix OLED (PMOLED) or an active matrix OLED (AMOLED) display. In some embodiments, the display is a plasma display. In some embodiments, the display is E-paper or E-ink. In some embodiments, the display is a video projector. In still further embodiments, the display is a combination of devices, such as those disclosed herein.

[0169] In some aspects, the electronic device includes an input device for receiving information from a subject. In some aspects, the input device is a keyboard. In some aspects, the input device is a pointing device, including, by way of non-limiting example, a mouse, a trackball, a trackpad, a joystick, a game controller, or a stylus. In some aspects, the input device is a touch screen or a multi-touch screen. In some aspects, the input device is a microphone for capturing voice or other audio input. In some aspects, the input device is a video camera or other sensor for capturing motion or visual input. In further aspects, the input device is a Kinect, Leap Motion, or the like. In still further aspects, the input device is a combination of devices, such as those disclosed herein.

[0170] B. Non-Transitory Computer-Readable Storage Medium In some aspects, the platforms, media, methods and applications described herein include one or more non-transitory computer-readable storage media encoded with a program including instructions executable by an operating system of a digital processing device, optionally networked. In further aspects, the computer-readable storage medium is a tangible component of an electronic device. In yet further aspects, the computer-readable storage medium is optionally removable from the electronic device. In some aspects, the computer-readable storage medium includes, as non-limiting examples, CD-ROMs, DVDs, flash memory devices, solid-state memories, magnetic disk drives, magnetic tape drives, optical disk drives, cloud computing systems and services, and others. In some cases, the programs and instructions are encoded in the medium permanently, substantially permanently, semi-permanently, or non-transitoryly.

[0171] C. Computer Programs In some aspects, the platform, medium, method and application described herein includes at least one computer program, or the use thereof. A computer program includes a sequence of instructions written to perform a specified task and executable by a CPU of an electronic device. The computer readable instructions may be implemented as program modules, such as functions, objects, application programming interfaces (APIs), data structures, etc., that perform a particular task or implement a particular abstract data type. In various embodiments, the computer program may be written in various versions of various languages.

[0172] The functionality of the computer-readable instructions may be combined or distributed as desired in various environments. In some embodiments, the computer program includes one instruction sequence. In some embodiments, the computer program includes multiple instruction sequences. In some embodiments, the computer program is provided from one location. In some embodiments, the computer program is provided from multiple locations. In various embodiments, the computer program includes one or more software modules. In various embodiments, the computer program includes, in part or in whole, one or more web applications, one or more mobile applications, one or more standalone applications, one or more web browser plug-ins, extensions, add-ins or add-ons, or combinations thereof.

[0173] D. Web Applications In some aspects, the computer program comprises a web application. In various embodiments, the web application, in various aspects, utilizes one or more software frameworks and one or more database systems. In some aspects, the web application is created on a software framework such as Microsoft® .NET or Ruby on Rails (RoR). In some aspects, the web application utilizes one or more database systems, including, by way of non-limiting examples, relational, non-relational, object-oriented, associative, and XML database systems. In further aspects, suitable relational database systems include, by way of non-limiting examples, Microsoft® SQL Server, mySQL™, and Oracle®. In various embodiments, the web application, in various aspects, is written in one or more versions of one or more languages. The web application may be written in one or more markup languages, presentation definition languages, client-side scripting languages, server-side coding languages, database query languages, or combinations thereof. In some aspects, a web application is written to some extent in a markup language such as HyperText Markup Language (HTML), Extensible HyperText Markup Language (XHTML), or Extensible Markup Language (XML). In some aspects, a web application is written to some extent in a presentation definition language such as Cascading Style Sheets (CSS). In some aspects, a web application is written to some extent in a client-side scripting language such as Asynchronous JavaScript (Javascript) and XML (AJAX), Flash® ActionScript (Actionscript), JavaScript, or Silverlight®.In some aspects, the web application is written to some extent in a server-side coding language, such as Active Server Pages (ASP), ColdFusion, Perl, Java, JavaServer Pages (JSP), Hypertext Preprocessor (PHP), Python, Ruby, Tcl, Smalltalk, WebDNA, or Groovy. In some aspects, the web application is written to some extent in a database query language, such as Structured Query Language (SQL). In some aspects, the web application integrates with enterprise server products, such as IBM Lotus Domino. In some aspects, the web application includes a media player element. In various further aspects, the media player element utilizes one or more of many suitable multimedia technologies, including, by way of non-limiting examples, Adobe Flash, HTML 5, Apple QuickTime, Microsoft Silverlight, Java, and Unity.

[0174] E. Mobile Applications In some aspects, the computer program comprises a mobile application provided to the mobile electronic device. In some aspects, the mobile application is provided to the mobile electronic device at the time of manufacture. In some aspects, the mobile application is provided to the mobile electronic device over a computer network as described herein.

[0175] In various embodiments, the mobile application is created by various techniques using hardware, languages ​​and development environments. In various embodiments, the mobile application is written in several languages. Suitable programming languages ​​include, by way of non-limiting examples, C, C++, C#, Objective-C, Java, JavaScript, Pascal, Object Pascal, Python, Ruby, VB.NET, WML and XHTML / HTML with or without CSS, or combinations thereof.

[0176] Suitable mobile application development environments are available from several sources. Commercially available development environments include, by way of non-limiting example, Airplay SDK, alcheMo, Appcelerator®, Celsius, Bedrock, Flash Lite, .NET Compact Framework, Rhomobile, and WorkLight Mobile platforms. Other development environments are available at no cost, and by way of non-limiting example, Lazarus, MobiFlex, MoSync, and Phonegap. Mobile device manufacturers also distribute software development kits, including, by way of non-limiting example, the iPhone® and iPad® (iOS) SDK, Android™ SDK, BlackBerry® SDK, BREW SDK, Palm® OS SDK, Symbian SDK, webOS SDK, and Windows® Mobile SDK.

[0177] In various embodiments, several commercial forums are available for the distribution of mobile applications including, as non-limiting examples, the Apple® App Store, Android™ Market, BlackBerry® App World, App Store for Palm devices, App Catalog for webOS, Windows® Marketplace for Mobile, Ovi Store for Nokia® devices, Samsung® Apps, and Nintendo® DSi Shop.

[0178] F. Standalone Applications In some aspects, a computer program includes a standalone application, which is a program that is run as an independent computer process, rather than an add-on to an existing process, e.g., a plug-in. In various embodiments, a standalone application is often compiled. A compiler is a computer program or programs that convert source code written in a programming language into binary object code, such as assembly language or machine code. Suitable compiled programming languages ​​include, by way of non-limiting examples, C, C++, Objective-C, COBOL, Delphi, Eiffel, Java, Lisp, Python, Visual Basic, and VB.NET, or combinations thereof. Compilation is often performed, at least in part, to create an executable program. In some aspects, a computer program includes one or more executable compiled applications.

[0179] G. Software Modules In some aspects, the platforms, media, methods and applications described herein include software, server and / or database modules, or the use thereof. In various embodiments, the software modules are created by various techniques using machines, software and languages. The software modules disclosed herein can be implemented in numerous ways. In various aspects, the software modules include files, code sections, programming objects, programming structures, or combinations thereof. In further various aspects, the software modules include multiple files, multiple code sections, multiple programming objects, multiple programming structures, or combinations thereof. In various aspects, the one or more software modules include, as non-limiting examples, web applications, mobile applications, and standalone applications. In some aspects, the software modules are present in one computer program or application. In some aspects, the software modules are present in more than one computer program or application. In some aspects, the software modules are hosted on one machine. In some aspects, the software modules are hosted on more than one machine. In further aspects, the software modules are hosted on a cloud computing platform. In some aspects, a software module is hosted on one or more machines in one location. In some aspects, a software module is hosted on one or more machines in two or more locations.

[0180] H. Database In some aspects, the platforms, systems, media and methods disclosed herein include one or more databases, or the use thereof. In various embodiments, many databases are suitable for storing and retrieving barcode, route, parcel, object or network information. In various aspects, suitable databases include, as non-limiting examples, relational databases, non-relational databases, object-oriented databases, object databases, entity-relationship model databases, associative databases and XML databases. In some aspects, the database is based on the Internet. In further aspects, the database is based on the Web. In still further aspects, the database is based on cloud computing. In some aspects, the database is based on one or more local computer storage devices.

[0181] I. Data Transmission The subject matter described herein, including the methods and systems provided herein, can be configured to be performed in one or more facilities in one or more locations. The location of the facility is not limited by country, but includes any country or territory. In some instances, one or more steps are performed in a country different from the country in which another step of the method is performed. In some instances, one or more steps for obtaining a sample are performed in a country different from the country in which one or more steps for detecting the presence or absence of a condition from a sample are performed. In some embodiments, one or more method steps involving a computer system are performed in a country different from the country in which another step of the method provided herein is performed. In some embodiments, data processing and analysis are performed in a country or location different from the country or location in which one or more steps of the method described herein are performed. In some embodiments, one or more items, products, or data are transferred from one or more of the facilities to one or more different facilities for analysis or further analysis. The items include, but are not limited to, one or more components obtained from a subject, such as processed cellular material. The processed cellular material includes, but is not limited to, cDNA reverse transcribed from RNA, amplified RNA, amplified cDNA, sequenced DNA, isolated and / or purified RNA, isolated and / or purified DNA, and isolated and / or purified polypeptides. The data includes, but is not limited to, information regarding subject stratification, and any data produced by the methods disclosed herein. In some aspects of the methods and systems described herein, an analysis is performed and a subsequent data transmission step will communicate or transmit the results of the analysis.

[0182] J. Web Browser Plug-ins In some aspects, the computer program includes a web browser plug-in. In computer processing, a plug-in is one or more software components that add specific functionality to a larger software application. Makers of software applications support plug-ins to allow third party developers to create the ability to extend the application, to support easy addition of new features, and to reduce the size of the application. When supported, plug-ins allow customization of the functionality of the software application. For example, plug-ins are commonly used in web browsers to play videos, provide interactivity, scan for viruses, and display specific file types. In various embodiments, several web browser plug-ins can be used, including Adobe® Flash® player, Microsoft® Silverlight®, and Apple® QuickTime®. In some aspects, the toolbar includes one or more web browser extensions, add-ins, or add-ons. In some aspects, the toolbar includes one or more explorer bars, tool bands, or desk bands.

[0183] In various embodiments, several plug-in frameworks are available that allow for development of plug-ins in a variety of programming languages, including, by way of non-limiting examples, C++, Delphi, Java, PHP, Python™, and VB.NET, or combinations thereof.

[0184] A web browser (also called an Internet browser) is a software application designed for use by network-connected electronic devices to search, present and traverse information resources in the World Wide Web. Suitable web browsers include, by way of non-limiting example, Microsoft® Internet Explorer®, Mozilla® Firefox®, Google® Chrome, Apple® Safari®, Opera Software® Opera® and KDE Konqueror. In some embodiments, the web browser is a mobile web browser. Mobile web browsers (also called microbrowsers, minibrowsers and wireless browsers) are designed for use in mobile electronic devices, including, by way of non-limiting example, handheld computers, tablet computers, netbook computers, sub-notebook computers, smartphones, music players, personal digital assistants (PDAs) and handheld video game systems. Suitable mobile web browsers include, by way of non-limiting example, Google® Android® browser, RIM BlackBerry® browser, Apple® Safari®, Palm® Blazer, Palm® WebOS® browser, Mozilla® Firefox® for mobile, Microsoft® Internet Explorer® mobile, Amazon® Kindle® Basic Web, Nokia® browser, Opera Software® Opera® mobile and Sony® PSP™ browser.

[0185] K. Computer-Aided Business Methods The methods described herein can utilize one or more computers. The computer can be used to manage customer and sample information, such as sample or customer tracking, database management, analysis of molecular profiling data, analysis of cytological data, data storage, billing, marketing, reporting of results, storage of results, or combinations thereof. The computer can include a monitor or other graphical interface for displaying data, results, billing information, marketing information (e.g., demographics), customer information, or sample information. The computer can also include a means for data or information input. The computer can include a processing device and fixed or removable media, or combinations thereof. The computer can be accessed by a user in physical proximity to the computer, for example, via a keyboard and / or mouse, or by a user who does not necessarily have access to the physical computer, via a communication medium, such as a modem, Internet connection, telephone connection, or wired or wireless communication signal carrier. In some cases, the computer can be connected to a server or other communication device to relay information from the user to the computer or from the computer to the user. In some cases, the user can store data or information obtained from the computer via a communication medium on a medium, such as a removable medium. It is envisioned that data relating to the method can be transmitted across such networks or connections for receipt and / or review by an interested party. The receiving party can be, but is not limited to, an individual, a health care provider, or a health care administrator. In one example, the computer-readable medium includes a medium suitable for transmitting results of an analysis of a biological sample. The medium can include the results of interest, such results being obtained using the methods described herein.

[0186] An entity obtaining sample information may enter it into a database for one or more of the following purposes: inventory tracking, assay result tracking, order tracking, customer management, customer service, billing, and sales. Sample information may include, but is not limited to, customer name, unique customer identification, customer associated medical professional, indicated assay(s), assay results, validation modality, indicated validation test, individual medical history, preliminary diagnosis, suspected diagnosis, sample history, insurance company, healthcare provider, third party testing center, or any information suitable for storage in a database. Sample history may include, but is not limited to, sample age, sample type, method of acquisition, storage, or transportation.

[0187] The database can be accessible by customers, medical professionals, insurance companies, or other third parties. Database access can take the form of electronic communication, such as computer or telephone. The database can be accessed through an intermediary, such as a customer service representative, business representative, consultant, independent testing center, or medical professional. The availability or extent of database access or sample information, such as assay results, can vary depending on the payment of a fee for products and services provided or to be provided. The extent of database access or sample information can be limited to comply with generally accepted or legal requirements for patient or customer confidentiality. EXAMPLES

[0188] The examples provided below are for illustrative purposes only and are not intended to limit the scope of the claims provided herein.

[0189] Example 1 Germline microsatellite genotypes distinguish children with medulloblastoma (MB) preface Medulloblastoma (MB) is a common malignant childhood brain tumor. Because children with MB have not yet experienced a lifetime of environmental exposures and stresses, MB may be primarily due to inherited or spontaneous mutations. Large-scale genomic characterization has divided MB tumors into at least four consensus molecular subgroups: WNT, SHH, group 3, and group 4, each with distinct transcriptional profiles, copy number alterations, somatic mutations, and clinical outcomes. Pediatric brain cancers in general and specifically MB have 5-10 times fewer mutations than typically observed in adult solid tumors. The most prominent tumor-initiating genetic mutations, such as p53, PTEN, RB, and EGFR, are remarkably rare. In addition, the incidence of known genetic tumor-predisposing mutations can be relatively low. Several known genetic abnormalities, such as PTCH, SMO and CTNNB1 mutations and MYC and MYCN amplifications, may individually be insufficient to efficiently cause MB in animal models and require an enhancing background, usually p53 inactivation, that can be found in less than 5% of human tumors. Many genome-wide association studies (GWAS) in MB have been able to focus on single-base variants while ignoring non-coding regions and repetitive DNA. However, linkage can be shown between germline microsatellite (MS) insertions and deletions (indels) and a number of neurological disorders, such as Huntington's disease and Friedreich's ataxia, the former caused by microsatellite variants in coding sequences and the latter caused by microsatellite variants in non-coding intronic sequences. Furthermore, microsatellite variants can contribute to the genetic background of some cancers. In addition, many cancer-related genes contain MS loci (e.g., PTEN and NF1), and in some cases, somatic MS indels have been inevitably linked to cancer. Based on these findings, a permissive constitutional genetic environment may be created by the cooperation of DNA microsatellite repetitive elements that influence the transcriptional and translational landscape of an individual, rendering it susceptible to tumorigenesis through modulation of underlying cellular processes.

[0190] MS can contain 1-6 base pair units that are repeated in tandem to form an array. Over 600,000 unique MSs exist in the human genome, and they can be embedded in gene introns, exons, and regulatory regions. The length of microsatellite loci can change frequently and vary between alleles and individuals due to strand slippage replication and heterozygote instability. Such changes can affect gene expression by inducing Z-DNA and H-DNA folding; altering nucleosome positioning; and changing the spacing of DNA binding sites. Non-coding variants can alter the DNA secondary structure and protein / RNA binding of genes in close proximity to their location, leading to altered transcriptional and translational activity and alternative splicing. For these reasons, MSs have been called the "tuning knob" of gene expression. Microsatellite loci containing 3 or 6 base pair repeat elements within exons can cause gain or loss of amino acids by maintaining in-frame with codon triplets; other non-modulo-3 lengths can cause frameshift mutations. Genes with MS can contribute disproportionately to nervous system disorders. This particular vulnerability to tandem repeats, especially the expansion of CAG motifs, can point to their importance in neurogenesis. Indeed, repetitive elements can play a role in neurological diseases; polyglutamic acid repeats in particular can play a role in Huntington's disease, spinocerebellar ataxia, and spinal-bulbar muscular atrophy. Similarly, bioinformatics studies indicate that many genes hosting tandem repeats may have neurological functions.

[0191] The development of microsatellite genotyping algorithms and advances in genome sequencing have allowed the identification of germline microsatellite genotypes that can distinguish healthy individuals from diseased individuals with different types of cancer (breast, colon, glioma, etc.). In this example, a set of microsatellite genotypes that can distinguish children with MB from healthy individuals based on their germline DNA is described. method

[0192] Patient samples Germline DNA WES and WGS from medulloblastoma (MB) patients were downloaded from the following datasets: phs000504, phs000409, EGAD00001000122, EGAD00001000275, EGAD00001000816, and Waszak, SM, et. Al (Spectrum and prevalence of genetic predisposition in medulloblastoma: a retrospective genetic study and prospective validation in a clinical trial cohort. The Lancet Oncology, Volume 19, Issue 6, 785-798, the entire contents of which are incorporated herein by reference). In addition, WES from blood DNA of six MB patients was newly generated using TruSeq Exome Target Enrichment Kit and Illumina sequencer HiSeq 2500. Germline DNA WES and WGS from healthy controls were downloaded from 1000 Genomes. Germline DNA WES from 100 healthy children was provided by the Hopp Children's Cancer Center at NCT Heidelberg, Heidelberg, Germany.

[0193] Sequence mapping and coverage WES and WGS reads were mapped to the human GRCh38 / hg38 reference genome using Bowtie2. Overall, coverage for the 120 MB germline samples was 31× (31.0±18.2). Coverage for the control samples was 13× (13.4±7.8).

[0194] Microsatellite list creation A list of microsatellites in the human reference genome version GRCh38 / hg38 was generated by a custom perl script "searchTandemRepets.pl" using default parameters. This script can be used in microsatellite studies and is freely available online. Briefly, the "searchTandemRepets.pl" script first searched for pure repetitive stretches: no impurities were allowed. Then, imperfect and complex repeats were handled using the "mergeGap" parameter with a default value of 10 base pairs. Essentially, impurities that interrupted stretches of pure repetitive sequences were allowed as long as they did not exceed 10 base pairs. Similarly, repeats closer than 10 base pairs were considered complex. As a result, the repeats in the CAGm database were highly pure, and the components of the complex repeats were also highly pure. The initial list generated by this script contained 1,671,121 microsatellites. To reduce the likelihood of incorrect read mapping between microsatellites, we removed the entire subset of microsatellites that harbored the same repeat motif between their 5 base pair long 3' and 5' flanking regions. For example, the microsatellite "GCTGC(A) 34 CTTAG" and "GCTGC(A) 15 "CTTAG" was preemptively removed from the initial list of microsatellites. Microsatellites can be embedded in larger repetitive motifs. The filtered list contained 625,195 unique microsatellites in the human genome.

[0195] Microsatellite genotyping The program Repeatseq was used to genotype microsatellites in next generation sequencing reads. Repeatseq uses Bayesian model selection guided by an empirically derived error model. The error model incorporated sequence and read characteristics: units, length and base quality. Repeatseq operated on three input files: a reference genome, a file containing reads aligned to the human reference genome (.bam file), and a list of known microsatellites (according to the methods and systems disclosed herein). The output was a variant call format (.vcf) file listing the genotype for each microsatellite locus consisting of the two alleles with the most supportive reads. The advantage of Repeatseq over other microsatellite genotyping programs was that it realigned each read to the reference genome prior to array length detection. Repeatseq can be used in microsatellite studies and is freely available.

[0196] The capabilities of Repeatseq were extended for the detection of somatic microsatellite variability: for example, minor alleles. A minor allele can be distinct from the major allele of a genotype; it can be somatically acquired in normal tissues as a certain age. A minor allele was used as an indication of microsatellite variability. Briefly, the detection of minor alleles was made possible by two steps based on the Repeatseq output. First, the output of realigned reads was made possible in the call to Repeatseq. Second, the realigned reads were stripped of all major alleles of the genotype. Among the remaining reads, the length of the array supported by at least three reads was counted as a minor allele. However, when comparing minor alleles in different samples, an alternative approach was used. Specifically, the length of the array supported by at least 20% of the total read depth was counted as a minor allele.

[0197] statistics A power calculation was performed based on previous observations of microsatellite genotype distributions for other cancers and controls to select the size of the training set while ensuring there were sufficient samples in the test set for validation. A conservative type I error probability associated with testing this null hypothesis of 0.01 was chosen as part of the validation. Responses within each subject group can be shown to be normally distributed with a standard deviation of 1. For a true difference between experimental and control means of 2, the null hypothesis that the population means of the experimental and control groups were equal was rejected with a probability (power) of greater than 0.99 for a study with 120 experimental subjects and 426 control subjects. Thus, the training set was expected to be adequately driven by the number of samples available.

[0198] For each microsatellite, the distribution of genotypes differed in germline DNA from two groups of samples in the training dataset: 120 MB and 425 healthy controls. In each case, the generalized Fisher exact test was used to quantify statistical differences. Briefly, for each microsatellite, a contingency table was populated with the genotype counts of the two groups: MB and normal (Figure 9). The p-value for each contingency table was then calculated using the Fisher test function in R. The Benjamini-Hochberg multiple testing correction (n=43,457 tested microsatellites) was applied to control for false discovery rate.

[0199] Microsatellite filtering to control for age, ethnicity and sequencing protocol This study was designed to identify germline microsatellite variants specific to MB; specifically, statistically significant microsatellites were identified in 120 MB samples and 425 healthy controls. However, these samples were not matched for age or sequencing protocol; furthermore, they were only partially matched for ethnicity. Thus, this approach may run the risk of identifying microsatellites due to age, sequencing, and ethnic bias, rather than disease mode alone. To mitigate this risk, microsatellites were identified with potential bias (for age, sequencing, or ethnicity) and excluded from subsequent analysis.

[0200] Controlling for age: To identify microsatellites whose genotypes vary non-randomly with age, we compared 100 healthy European children and 501 European adults from the 1,000 Genomes Project. Fisher's exact test identified 738 (out of 29,061) statistically significant microsatellites: Benjamini-Hochberg correction (p-value <0.05) (Figure 10).

[0201] Controlling for sequencing protocol: To identify microsatellites that vary based on DNA sequencing protocol (WGS vs. WES), genotypes from paired WGS and WES experiments in 16 individuals in the 1,000 Genomes Project were compared. The distribution of genotypes for 37,511 microsatellites was tested for statistical differences (Fisher exact test); 157 were found to be different using Benjamini-Hochberg false discovery correction (p-value < 0.05) (Figure 11). This result was likely due to the fact that microsatellites are prone to read mapping errors, especially if they have large insertions or deletions. Thus, the 157 identified microsatellites may be particularly prone to mapping errors or to being in highly variable regions of the genome; these were excluded from subsequent analyses. In addition, 37,775 identified microsatellite calls were absent in 134 WGS samples. Consequently, 37,775 of these could not be used in risk, diagnostic or prognostic microsatellite-based assays; they were excluded from subsequent analyses (Figure 11).

[0202] Controlling for ethnicity: To identify DNA microsatellites that vary according to ethnicity, we compared and analyzed the distribution of genotypes in 352 American and 502 European samples, all derived from the 1,000 Genomes Project. In total, 184,981 statistical tests were performed and 1,037 microsatellites were found to be significantly different using the Benjamini-Hochberg false discovery correction (p-value < 0.05). Furthermore, we tested the distribution of microsatellite genotypes in a group of 59 predominantly European and 55 predominantly American MB samples. There, 13,899 tests were performed and 478 microsatellites were found to be different after the Benjamini-Hochberg false discovery correction (p-value < 0.05). 71 microsatellites present in both lists were identified and were excluded from further analysis (Figure 12).

[0203] The three steps described above yielded 38,653 unique microsatellites; all were removed from further analysis.

[0204] Metrics and ROC analysis for scoring samples Metrics for scoring samples: The metrics for scoring samples were designed based on their unique microsatellite genotype distribution. Essentially, the metric was a weighted sum of the genotypes belonging to each sample: the weights came from the difference in frequency per genotype in the MB and healthy groups. A visual summary of the metrics is shown in Figure 13.

[0205] ROC analysis: Receiver operating characteristic (ROC) analysis was used to design a classification scheme that could distinguish samples with MB from healthy controls. Briefly, the area under the ROC curve (AUC) was used as a measure of how well the scores in the two groups distinguished between the two groups. A cutoff was then selected for any future classification. The cutoff was then a single score that minimized sensitivity and simultaneously maximized specificity; this was identified using the Youden index. ROC analysis, AUC calculation and Youden index optimization were performed using the freely available R package: ROCR.

[0206] Microsatellite subset (genetic algorithm) Genetic algorithms can be a class of biologically inspired algorithms. Briefly, a genetic algorithm was used to identify the most informative subset of markers from the set of 139 using a two-step iterative process. First, the algorithm was initialized with a random subset of the 139 microsatellite markers; then, the top performing subsets were successively recombined, reevaluated, and reranked. Three hyperparameters (e.g., the parameter set before the iterative algorithm began) were used to control the maximum population size, the size of each subset, the performance of each subset, and the diversity of subsets in the population. Details of each step and hyperparameters are provided below.

[0207] Initialization: Each subset in the initial population consisted of markers randomly chosen from a total of 139. Hyperparameters controlled the initial population size and the size of each subset. Once entered, the initial subsets were ranked based on the performance metrics described below.

[0208] Optimization: Each optimization cycle started by placing 10 new subsets into the population; of these, 7 were created by recombining 2 members (randomly chosen) of the existing population, and 3 were created randomly. To recombine the 2 subsets, each was split; then, the 2 pieces (one from each subset) were rejoined. The split point and the pieces were chosen randomly. Three random subsets were created at initialization, which helps maintain the diversity of the population. Once the new subsets were created, the population was re-ranked based on the performance metric. Finally, the 10 worst performing subsets were discarded to maintain the population size.

[0209] Hyperparameters: A population size of 100 subsets was initialized and used throughout the algorithm. The minimum and maximum subset sizes were set to 8 and 64 markers, respectively. No overlapping markers were allowed in the subsets. The performance of each subset was determined by ROC analysis using 120 MB samples and 425 healthy controls, i.e., the same training samples used throughout the study. The sum of sensitivity and specificity, which indicates the performance of each subset, was used to rank the populations at each generation of the genetic algorithm.

[0210] Robustness: The parameters of the genetic algorithm were chosen for computational feasibility. However, the performance of the genetic algorithm was insensitive to the choice of hyperparameters. In addition, the details of the optimization cycle (e.g., the number of new subsets in each cycle) did not affect the results of the genetic algorithm.

[0211] verification Samples used: 102 experimental subjects and 428 control subjects were chosen in the validation study to ensure that the study was overpowered. Using the subject (MB) and control distributions found when analyzing the training set (Figure 7A), the responses within each subject group were normally distributed with a standard deviation of 1.1. For a true difference between the experimental and control means of 4.4, rejection was based on the null hypothesis that the population means of the experimental and control groups were equal with a probability (power) of more than 0.99 for a type I error probability of 0.01 with a sample and control validation set of this size. All control samples used in training and validation were subjected to whole exome sequencing. For MB, the collection included both whole exome and whole genome samples. Whole genome sequencing samples were used exclusively for validation.

[0212] Procedure: Each validation sample was scored with the same metric used for the training samples. A cutoff (identified in training) was used to predict which of the 530 validation samples had MB and which were healthy controls. MB was predicted for validation samples above the cutoff. Predictions were compared to the known identity of 102 MB samples and 428 healthy controls. The sensitivity and specificity of these predictions were comparable to training.

[0213] Microsatellite variability To test whether individuals with MB were more prone to microsatellite variants, the total number of genotyped alleles per microsatellite (allele load) was used as a measure of its variability, and this metric was compared across disease and control cohorts. Alleles were defined such that the resulting counts were robust to two sources of error: (a) the potential effect of PCR artifacts was mitigated by requiring that each allele be supported by at least two reads; (b) to normalize for differences in read coverage across samples, each allele was required to be supported by at least 20% of the total number of reads mapped to the microsatellite. Alleles were only counted for microsatellites with mapped reads present in at least 20 percent of the samples. Fisher's exact test was then performed to establish statistical significance between MB patients and healthy individuals. This process was repeated 50 times with an average p-value of 0.077.

[0214] Two additional lines of evidence were used to also assess the integrity of the mismatch repair machinery in the medulloblastoma germline: (a) homozygous and heterozygous genotypes matched across all (71,192 total) microsatellites in MB and control samples; and (b) comparison of median microsatellite array length across all (71,192 total) microsatellites in MB and control samples. For the former analysis, aberrant mismatch repair can be expected to increase the enumeration of heterozygous genotypes; however, the difference between case and control samples was not statistically significant. The medulloblastoma samples together had 299,802 heterozygous genotypes and 2,596,324 homozygous genotypes; the control samples had 283,037 heterozygous genotypes and 2,449,046 homozygous genotypes. For the latter analysis, aberrant mismatch repair can be expected to result in the accumulation of longer or shorter median microsatellite array lengths in medulloblastoma samples compared to controls; again, the results were not statistically significant. Medulloblastoma samples had shorter median array lengths for 1,031 microsatellites and longer median array lengths for 907 microsatellites; the remaining 69,254 microsatellites did not differ in median array length.

[0215] Downstream analysis Genes associated with 139 microsatellite loci whose genotypes were significantly different between MB subjects and controls were used for functional analysis. In total, 124 genes were included in the analysis, excluding microsatellites located in intergenic regions. Pathway analysis was performed using Ingenuity Pathway Analysis (QIAGEN Inc.). Mutations and co-occurrence events were analyzed using PedcBioPortal. Protein-protein interaction (PPI) network construction was performed with STRING (high confidence) with a minimum interaction score of 0.7 and no more than 5 molecules in the first shell. This setting created a hub with 129 nodes and 49 edges, resulting in a network with a PPI enrichment p-value of 0.0007.

[0216] result Identification of informative loci in medulloblastoma microsatellites Single base mutations can be characterized in MB genome-wide analysis. Therefore, we studied the influence of microsatellite variants on medulloblastoma predisposition. To this end, a computational workflow was developed to identify germline microsatellites whose genotypes differ between children with medulloblastoma and control subjects, correcting for variations with age, ethnicity, and DNA sequencing protocol (Figure 6). A metric was also developed to score each sample based on its unique collection of microsatellite genotypes. This approach was applied to germline DNA sequencing data from 222 children with medulloblastoma and 853 healthy control subjects. The data were divided into two groups, both containing diseased and healthy subjects, the first group for training containing 120 medulloblastoma patients and 425 control individuals, and the second group for validation with 102 medulloblastoma patients and 428 control individuals. In the first stage of the analysis, the training set was used to genotype 43,457 different microsatellites present in both the 120 medulloblastoma samples and the 425 healthy controls. For each of these microsatellites, a generalized Fisher exact test was used to assess statistical differences in the genotype distribution between the two groups for each microsatellite. 2,094 microsatellites were identified with a p-value <0.05. After Benjamini-Hochberg multiple testing correction (α=.05), 422 passed false discovery. Three additional steps were taken to remove microsatellites that varied with age, ethnicity, and DNA sequencing protocol (Figures 6, 10, 11, and 12). In total, 283 microsatellites were removed from the 422 list, resulting in a reduced list of 139 (Figure 19). Collectively, this approach identified 139 microsatellites from germline DNA whose genotypes differed significantly between medulloblastoma subjects and healthy controls.

[0217] Medulloblastoma microsatellite classifier set To identify the subset of microsatellites with the best performance in distinguishing medulloblastoma samples and healthy controls, a set of 139 microsatellites was used to train a medulloblastoma classifier. First, a metric was designed to score each medulloblastoma and control sample based on the genotype of the 139 microsatellites (see Methods and Figure 13 for details). Then, a receiver operating characteristic (ROC) was created and used to determine the ability of the sample scores to act as a binary classifier for medulloblastoma. A subset optimization strategy based on a genetic algorithm method was used to identify the best subset of discriminatory markers using a two-step iterative process. First, subsets were randomly created from the complete list and ranked by their F-measure. Second, the top-performing subsets were successively mixed, re-evaluated, and re-ranked. The algorithm converged in 87 cycles to reveal a subset of 43 microsatellites with an F-measure of 0.90 and an area under the curve (AUC) of 0.962 (Figures 7, 20). The Youden index was determined indicating that the optimal cut-off score for distinguishing medulloblastoma samples from healthy controls was 0.155 (Figure 14). With a specificity of 0.92, the sensitivity when applied to the training set was 0.88 (Figure 7B). The chromosomal locations of these 43 markers in the human genome are shown in Figure 15. Thus, a set of 43 microsatellites was identified whose genotype distributions were able to distinguish medulloblastoma patients from healthy controls with 88% sensitivity and 92% specificity.

[0218] An independent cohort of germline DNA from medulloblastoma patients and healthy controls was used to validate the previous results. For the validation study, 102 experimental subjects and 428 control subjects were included, using the subject (medulloblastoma) and control distribution found when analyzing the training set (Figure 7) to ensure that the study was overpowered. In the training set, the responses within each subject group were normally distributed with a standard deviation of 1.1. It was found that for a true difference between the experimental and control means of 4.4, rejection of the null hypothesis that the population means of the experimental and control groups were equal could be made with a probability (power) of more than 0.99 for a type I error probability of 0.01 with a sample and control set of this size. When the optimal cutoff (0.155) was applied to the independent validation sample set, it was found that the classifier was able to distinguish cases from controls with a sensitivity of 0.95 and a specificity of 0.90 (Figures 7C and 7D). Taken together, a set of 43 MSs was obtained that were identified and validated, whose genotype distributions could distinguish MB patients from healthy controls, using germline DNA with high sensitivity and specificity.

[0219] Mutability of informative microsatellite loci in medulloblastoma In the germline, the rate of indels in MS is significantly higher than the rate of single base substitutions elsewhere in the genome, at 10 per locus per generation, respectively. -8 Compared to 10 -4 ~10 -3However, mutation rates also vary for different MSs based on the length of the repeat, its repetitive motif, and its effect on DNA folding. It was hypothesized that the differences found in the 139 MSs whose genotypes were non-randomly associated with MB (Figure 20) could be the result of an increase in microsatellite genotype variants specific to individuals with MB. To test whether individuals with MB were more prone to microsatellite variants, the total number of alleles genotyped per microsatellite (allele load) was used as a measure of their variability, and this metric was compared across disease and control cohorts. There was no significant difference in the number of alleles genotyped between healthy and MB individuals, supporting the conclusion that there was no generalized microsatellite instability in MB patients. By ranking all of the MSs by allele load, the predictive ability of informative microsatellites themselves for trait association was investigated to determine whether the 139 markers were present among the most mutable loci analyzed. It was found that these were present among the more mutable MSs, but did not include the most mutable sites. Moreover, the number of homozygous and heterozygous genotypes and the length of the microsatellite array were compared as potential sources of variability in MB. In both cases, there were no statistically significant differences between MB and control germline DNA. These results and data indicate that the association of MB with these 139 MS was a consequence of the genotypes of these individual microsatellites, rather than simply the result of constitutional hypermutability.

[0220] Informative roles of MST-related genes Of the 139 MS loci whose genotypes differed between MB and control samples, 114 were located in intronic regions, 15 in intergenic regions, 6 in 3'UTR, 3 in exonic regions, and 1 in 5'UTR (Figure 8A). To understand the potential mechanistic role of these genes, Ingenuity Pathway Analysis® was performed to analyze 124 genes associated with informative MS loci (excluding MS located in intergenic regions). The analysis revealed statistically significant associations with cancer and molecular cellular functions such as cell cycle, DNA replication, recombination and repair, and cell growth and proliferation, pointing to a relevance to cancer biology (Figure 8B and Figure 21). The occurrence of mutations in these 124 genes associated with informative MS was examined in four MB cohorts available in cBioportal. Despite the known low mutation rate in MB tumors, on average 17% of MB cancer samples contained mutations in at least one of these 124 genes compared to 4.5% of neuroblastoma tumors (Figure 22). Analysis of mutational co-occurrence using the Sick Kids 2016 dataset in cBioportal revealed that all possible (9,591 = 139 * It was shown that 135 of the (139-1) / 2) microsatellite pairs were found to be significantly co-occurring (p-value < 0.05). Two patients were found to have co-occurring mutations at 20 and 10 MB informative MS loci, respectively (Figure 23).

[0221] A protein-protein interaction (PPI) network composed of 124 genes associated with informative MS loci (Figure 8C) was found to contain 129 nodes and 49 edges, resulting in a network with a PPI enrichment p-value of 0.0007. Despite the small number of proteins used as input, a significant hub was related to mTOR, a prominent pathway (PI3K / AKT / mTOR) in MB tumors.

[0222] Three informative microsatellite loci were located in protein coding sequences (Figure 8A); all of these were trinucleotide repeats (RAI1, BCL6B, TNS1). Variants of trinucleotide repeats were recognized as causes of neurological and neuromuscular diseases such as Huntington's disease, spinocerebellar ataxia, and fragile X syndrome. Two of these genes (RAI1, BCL6B) are transcription factors located on the short arm of chromosome 17, whose deletions were the recurrent alterations in the most common subgroup of MB tumors. The BCL6B gene has been implicated in colon, gastric, and hepatic cancers, with the major genotype in MB tumors being 33 / 33, whereas in controls it is 30 / 33 (Figure 16); in this reading frame, the codon CAG was translated to serine. RAI1 (retinoid acid-inducible protein) encodes a nuclear protein of unknown function, whose haploinsufficiency causes Smith-Magenis syndrome. The two major genotypes of RAI1 in MB tumors were 38 / 41 and 41 / 41, whereas in controls they were 38 / 38 and 38 / 41 (Figure 16). Apart from inducing changes in RAI1 protein structure, short polyglutamine expansions also appeared to modulate transcription factor activity. RAI1 protein is highly expressed in the cerebellum, the region where MB tumors arise.

[0223] In this study, a set of 139 MS were identified that harbored distinct genotypes between MB patients and healthy controls. A subset of 43 MS could distinguish MB individuals from controls based on their germline DNA with a sensitivity and specificity of 0.95 and 0.90, respectively.

[0224] This study identified three sets of microsatellites: (a) 43 microsatellites that discriminated medulloblastoma samples and healthy controls together; (b) 139 microsatellites that had statistically different genotypes between medulloblastoma samples and healthy controls; and (c) 422 microsatellites identified in the initial screening. The microsatellites in all three sets passed false discovery. The set of microsatellites identified in the initial screening (c) contained 283 that were sensitive to age, ethnicity, and / or DNA sequencing; consequently, none were used in the subsequent analysis. Some of the microsatellites with ethnic bias may also have a role in medulloblastoma. The prevalence of many diseases, including medulloblastoma, can show ethnic differences. Thus, retesting of the 283 microsatellites may once again be feasible, and the genetic mechanisms that cause medulloblastoma are known.

[0225] Furthermore, the relationship between the group of 139 microsatellites (b) and its subset of 43 microsatellites (a) was investigated: the latter distinguished medulloblastoma samples from healthy controls, but the former did not. Mutations in the set of 43 microsatellites may have a greater effect on gene expression; or genes carrying these microsatellites may have a greater effect on disease initiation. This may be supported by the presence of two coding microsatellites in the set of 43; in both cases, mutations may have a direct effect on protein primary structure and a potential effect on secondary structure and function. In addition, the set of 43 microsatellites had a greater proportion embedded in the 5' and 3' UTR regions; it is possible that MS in these regions more strongly influenced gene expression / translation. These indications can be determined by expression studies of those genes carrying informative microsatellites in tumor tissues.

[0226] These results indicate that polyglutamine microsatellites embedded in the BCL6B and RAI1 genes may play a role in medulloblastoma. Only 181 polyglutamine microsatellites (out of 627,174) were present in the complete list of microsatellites screened. Thus, chance alone cannot explain the presence of two in the 43 informative microsatellites in the final list; using computer simulations, this chance of random occurrence was estimated to be approximately 1 in 1,000,000. Secondly, polyglutamine microsatellites may play a role in diseases such as spinal and bulbar muscular atrophy, Huntington's disease and various spinocerebellar ataxias. Furthermore, both BCL6B and RAI1 genes can be associated with disease; the former with lymphoma and the latter with Smith-Magenis syndrome. Polyglutamine diseases can be characterized by insoluble protein aggregates: these are not seen in some cancers. On the other hand, polyglutamine expansions can confer both gain and loss of function depending on the protein affected.

[0227] This study demonstrated two overall conclusions. First, the identified microsatellites (specifically, a set of 139 and a subset of 43) may play a role in medulloblastoma etiology. The effects of microsatellite array length variants included effects on DNA secondary structure, nucleosome positioning and DNA binding sites. Three of the identified microsatellites affected protein primary sequence. Microsatellites can aid in the differentiation of individuals with medulloblastoma from healthy controls; the classification scheme demonstrated high sensitivity and specificity of 0.95 and 0.90, respectively.

[0228] Treatment for medulloblastoma may leave survivors with lifelong burdens, including hearing loss, cognitive impairment, endocrine disorders, and increased risk of seizures and secondary malignancies. Identification of populations at risk for developing medulloblastoma may allow for early detection strategies that allow for less invasive, more localized tumor control measures. However, an effective way to improve the lives of such children may be to prevent their tumor formation. Recent advances in immunotherapy, including cancer vaccines, create the potential to immunize individuals against tumor-specific antigens. Such strategies may require selection of individuals appropriate for such intervention.

[0229] Example 2 Identification of informative microsatellite markers Samples of nucleic acid sequences from both subjects with a condition (first group) and healthy controls (second group) are obtained from public domain databases. Microsatellite loci are identified in both groups. The microsatellites are compared to reveal differences in microsatellite loci found only in the first group that are specifically associated or correlated with the condition. Statistical analysis and modeling are applied to these different microsatellites for their association or correlation to the condition. In some instances, the microsatellites are statistically weighted. After a set of microsatellites are identified as being strongly associated with the condition, these microsatellites are assembled into a training algorithm to further optimize the accuracy, sensitivity and specificity of these microsatellite associations to the condition. The training microsatellites can be randomly recombined to create additional combinations of microsatellites. After training is completed, the algorithm can be validated with additional independent sets of samples.

[0230] For example, the nucleic acid sequences of cancer patients and corresponding healthy controls are downloaded from The Cancer Genome Atlas (TCGA) and the 1000 Genomes Project, respectively. Microsatellite loci are identified in both groups. Comparison of the microsatellites between the two groups reveals a cluster of microsatellite loci found only in the cancer patient group and specifically associated or correlated with the cancer type. These microsatellites associated with the cancer type are then subjected to a training algorithm to enhance the accuracy, sensitivity and specificity of these microsatellites in association with cancer. After completion of training, the algorithm is validated with an additional set of samples from cancer-bearing or healthy controls. After validation, the algorithm is ready for application with patient samples.

[0231] Example 3 Patient risk assessment A serum sample is isolated from a subject during a routine health check. DNA is extracted from the serum sample and sequenced. The sequencing data is processed and analyzed to obtain a set of microsatellites unique to the subject. This set of microsatellites is then analyzed using a computer-implemented method designed to determine risk of developing cancer based on a comparison between the subject's microsatellites and microsatellites from a pan-cancer database. Each of the identified informative microsatellites is assigned a weight ranging between 0 and 1. The weights are created based on the accuracy, sensitivity and specificity of the identified microsatellites. The sum of the weights is then determined and used to create a classifier for the likelihood of developing one type of cancer. A pan-cancer classifier is then compiled and the classifiers for the likelihood of developing multiple cancers are reported for risk assessment of the subject. The pan-cancer classifier provides a risk assessment of a subject's likelihood of developing cancer, e.g., breast cancer, lung cancer, prostate cancer, cervical cancer, glioblastoma multiforme, endometrial cancer of the uterine corpus, colon adenocarcinoma, bladder, urothelial carcinoma, squamous cell carcinoma of the head and neck, squamous cell carcinoma and endocervical adenocarcinoma of the cervix, gastric adenocarcinoma, thyroid cancer, brain low-grade glioma, papillary renal cell carcinoma of the kidney, and hepatocellular carcinoma of the liver.

[0232] The subject is informed of the risk assessment by the laboratory report (Figure 5 and Figure 17). Along with the test summary, patient, medical professional and serum sample information is included. The summary reveals that the subject is currently cancer-free, but there are several identified microsatellites in the subject's genome that increase the likelihood of the subject developing lung cancer. The likelihood classifier for developing lung cancer includes a numerical output and is compared to a threshold for the likelihood of developing lung cancer. The threshold for the likelihood of developing lung cancer is 0.3, with a one standard deviation range of 0.1 to 0.5 (Figure 24). The likelihood classifier for developing lung cancer for the subject is 2.3, which indicates that the subject is very likely to develop cancer in the future. Therefore, additional clinical attention is paid to the subject's lungs and respiratory system. It is recommended that more routine lung imaging be performed at regular intervals. The subject is also advised not to start smoking and to avoid prolonged exposure to certain environments with known aerosolized carcinogens. Additionally, the summary provides an overview regarding the parameters of the risk assessment, such as the type of statistical methods and thresholds utilized, and the number of microsatellite loci analyzed.

[0233] Example 4 Measuring genomic age using minor alleles A sample of DNA from primary skin fibroblasts is obtained from the subject at age 17 and again at age 30. A DNA-seq library is constructed and then sequenced by a next generation sequencing platform and mapped to hg19. Enrichment can be performed to enrich for hotspots where minor alleles tend to occur in the population. Minor alleles with a minimum of 5 reads are independently confirmed by Sanger sequencing. True positive minor alleles are analyzed and weighted. Examples of locations where minor alleles appear include upstream or downstream of genes, exonic regions, intergenic regions, intron and exon spanning regions, 3'UTR, and 5'UTR. Minor alleles can be nonsynonymous variants, synonymous variants, frameshift indels, nonframeshift indels, stop gains, stop losses, or combinations thereof.

[0234] The minor alleles obtained from the comparison between the sample obtained at age 17 and the hg19 reference sequence are analyzed by computer-implemented methods to reveal genomic age. The increase in the number of minor alleles or the loci of minor alleles may contribute to the genomic age of the subject more than the chronological age and physical strength. The samples obtained from the same subject at age 17 and age 30 can be compared with each other to reveal the additional accumulation or shift of the pattern of minor alleles in the same subject. The comparison of minor alleles between age 17 and age 30 reveals that the subject has a slight increase in the total number of minor alleles. This increase is analyzed by computer-implemented methods to reveal the accelerated genomic aging rate in the subject. Therefore, the subject is recommended to adopt a certain lifestyle that emphasizes nutritional balance and reduced mental stress.

[0235] While preferred embodiments of the present invention have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will recognize numerous variations, changes, and substitutions without departing from the present disclosure. It is understood that various alternatives to the embodiments of the present disclosure described herein can be employed in the practice of the present disclosure. It is intended that the following claims define the scope of the disclosure, and that methods and structures within the scope of such claims and their equivalents are covered thereby. In certain embodiments, for example, the following items are provided: (Item 1) 1. A computer-implemented method for constructing an optimized classifier for a condition, comprising: ranking a subset of a plurality of microsatellites as a classifier for the condition over a plurality of optimization cycles, wherein the subset of the plurality of microsatellites comprises microsatellites in an initial population of microsatellites that are correlated with the condition, thereby identifying an optimized subset of the subset of microsatellites as the optimized classifier for the condition. (Item 2) 2. The method of claim 1, further comprising comparing microsatellites in a first set of samples from subjects having the condition and microsatellites in a second set of samples from subjects not having the condition, thereby identifying said initial population of microsatellites. (Item 3) 2. The method of claim 1, wherein the ranking step comprises comparing the subset of microsatellites to microsatellites in a sample from a subject having the condition and to microsatellites from a sample from a subject not having the condition. (Item 4) 2. The method of claim 1, further comprising an initialization step of randomly selecting an initial subset of population of microsatellites from said initial population of microsatellites for use in ranking in an optimization cycle of said plurality of optimization cycles. (Item 5) 2. The method of claim 1, wherein a population of at least 100 subsets of the initial population of microsatellites is used in the multiple optimization cycles. (Item 6) 2. The method of claim 1, wherein the minimum number of microsatellites in said subset of microsatellites is eight. (Item 7) 2. The method of claim 1, wherein the maximum number of microsatellites in said subset of microsatellites is 64. (Item 8) 2. The method of claim 1, wherein overlapping microsatellites are not allowed in said subset of the subsets of microsatellites. (Item 9) 2. The method of claim 1, wherein the ranking step comprises performing a receiver operating characteristic (ROC) analysis using (i) the subset of microsatellites, (ii) the microsatellites in samples from subjects with the condition, and (iii) the microsatellites in samples from subjects without the condition. (Item 10) 10. The method of claim 9, wherein the ranking in an optimization cycle of the plurality of optimization cycles comprises determining the sum of the sensitivity and specificity of microsatellites in each subset of the subsets as the classifier for the condition. (Item 11) 11. The method of claim 10, wherein an optimization cycle of the plurality of optimization cycles comprises adding 10 new subsets of the initial population of microsatellites to a subset from a previous optimization cycle of the plurality of optimization cycles. (Item 12) 12. The method of claim 11, wherein seven of the ten new subsets are generated by randomly splitting and recombining two randomly chosen subsets from the previous optimization cycle, and three of the ten new subsets are generated by randomly selecting microsatellites from the initial population of microsatellites. (Item 13) Item 13. The method of item 12, further comprising discarding ten of the subsets in the optimization cycle based at least in part on having the lowest ranking in the optimization cycle. (Item 14) 2. The method of claim 1, wherein the condition comprises the presence or absence of a health condition in the subject. (Item 15) 2. The method of claim 1, wherein the condition comprises an increased or decreased likelihood of developing a health condition in the subject. (Item 16) 2. The method of claim 1, wherein the condition comprises an increase or decrease in the likelihood that the subject will benefit from treatment of a health condition. (Item 17) 2. The method of claim 1, wherein the condition comprises an increase or decrease in the likelihood that the subject has an increased risk of an adverse event from treatment for a health condition. (Item 18) 2. The method of claim 1, wherein the condition comprises the subject's responsiveness to a treatment for a condition. (Item 19) 2. The method of claim 1, wherein the condition comprises a prognosis of a health condition in the subject. (Item 20) 20. The method of any one of items 14 to 19, wherein the condition is cancer. (Item 21) 21. The method of claim 20, wherein the cancer is lung cancer. (Item 22) 20. The method of any one of items 14 to 19, wherein the condition is a neurological disease. (Item 23) 20. The method according to any one of items 14 to 19, wherein the condition is a cardiovascular disease. (Item 24) 1. A computer-implemented method comprising determining a classifier value for a condition from a sample from a subject using a plurality of parameters, wherein each parameter of the plurality of parameters is a statistical measure of correlation of a respective one of a plurality of microsatellites from a sample from a subject having the condition or from a sample from a subject not having the condition. (Item 25) 25. The method of claim 24, wherein the plurality of parameters comprises a plurality of weights. (Item 26) 26. The method of claim 25, wherein the plurality of weights comprises a plurality of optimal weights. (Item 27) 27. The method of claim 26, further comprising determining the plurality of optimal weights. (Item 28) 28. The method of claim 27, wherein the step of determining the plurality of optimal weights comprises the step of applying a standard regression analysis to the plurality of weights. (Item 29) 25. The method of claim 24, wherein the step of determining the plurality of optimal weights comprises the use of a genetic algorithm. (Item 30) 25. The method of claim 24, wherein the step of determining the value of the classifier comprises using minor allele frequency data. (Item 31) 25. The method of claim 24, wherein the plurality of microsatellites comprises at least 10 microsatellites. (Item 32) 25. The method of claim 24, wherein each of the plurality of microsatellites is correlated with the condition. (Item 33) 25. The method of claim 24, further comprising comparing the value of the classifier to a threshold value. (Item 34) 25. The method of claim 24, wherein the condition comprises the presence or absence of a health condition in the subject. (Item 35) 25. The method of claim 24, wherein the condition comprises an increased or decreased likelihood of developing a health condition in the subject. (Item 36) 25. The method of claim 24, wherein the condition comprises an increase or decrease in the likelihood that the subject will benefit from treatment of a health condition. (Item 37) 25. The method of item 24, wherein the condition comprises an increase or decrease in the likelihood that the subject has an increased risk of an adverse event from treatment for a health condition. (Item 38) 25. The method of claim 24, wherein the condition comprises the subject's responsiveness to a treatment for a condition. (Item 39) 39. The method according to any one of items 34 to 38, wherein the condition is cancer, a cardiovascular disease or a neurological disease. (Item 40) 40. The method of claim 39, wherein the cancer is lung cancer. (Item 41) 1. A computer-implemented method for determining the genomic age of a subject, comprising: a) determining a microsatellite minor allele signature in a first sample from a subject; b) processing the microsatellite minor allele signature with a reference; and c) determining the genomic age of the subject based on the processing. The method includes: (Item 42) 42. The method of claim 41, wherein the processing step comprises comparing the microsatellite minor allele signature to the reference. (Item 43) 42. The method of claim 41, wherein the minor allele characteristic is the number of minor alleles at a locus. (Item 44) 44. The method of claim 43, wherein the number of minor alleles is supported by at least three next generation sequencing sequence reads. (Item 45) 42. The method of claim 41, wherein the minor allele signature is the total number of minor allele reads normalized to the total number of major allele reads at a locus. (Item 46) 42. The method of claim 41, further comprising performing next-generation sequencing of the first sample from the subject to generate microsatellite sequence reads for the subject. (Item 47) 47. The method of claim 46, wherein the first sample comprises blood, saliva or a tumor. (Item 48) 46. ​​The method according to claim 45, further comprising, after operation c), determining a minor allele signature in a second sample from the subject. (Item 49) 48. The method of claim 47, further comprising assessing the minor allele signature in the first sample from the subject and the minor allele signature in the second sample from the subject, and determining a rate of genomic aging of the subject based on the assessing. (Item 50) 1. A computer-implemented method comprising: a) determining a plurality of classifiers for a sample from a subject using microsatellites in the sample; b) processing the plurality of classifiers with a plurality of reference classifiers for a plurality of states; c) determining at least one state for the object from among the plurality of states based on the processing step; The method includes: (Item 51) 51. The method of claim 50, wherein the processing step comprises comparing the plurality of classifiers to the plurality of reference classifiers for the plurality of conditions. (Item 52) 51. The method of claim 50, wherein the at least one condition of the plurality of conditions comprises the presence or absence of at least one health condition from among a plurality of health conditions in the subject. (Item 53) 51. The method of claim 50, wherein the at least one condition of the plurality of conditions comprises an increase or decrease in the likelihood of developing at least one health condition from a plurality of health conditions in the subject. (Item 54) 51. The method of claim 50, wherein the at least one condition of the plurality of conditions comprises an increase or decrease in the likelihood that the subject will benefit from treatment for at least one health condition from among a plurality of health conditions in the subject. (Item 55) 51. The method of claim 50, wherein the at least one condition of the plurality of conditions comprises an increase or decrease in the likelihood that the subject has an increased risk of an adverse event from treatment for at least one health condition from among the plurality of health conditions in the subject. (Item 56) 51. The method of claim 50, wherein the at least one condition of the plurality of conditions comprises responsiveness of the subject to a treatment for at least one health condition among a plurality of health conditions of the subject. (Item 57) 57. The method of any one of items 51 to 56, wherein the multiple health conditions comprise multiple cancers. (Item 58) 58. The method of claim 57, wherein the multiple cancers comprise ovarian cancer, breast cancer, low-grade glioma, glioblastoma, lung cancer, prostate cancer or melanoma. (Item 59) 51. The method of claim 50, wherein the multiple health conditions comprise multiple neurological diseases or multiple cardiovascular diseases. (Item 60) 60. A non-transitory computer-readable medium comprising executable instructions that, when executed by one or more processors, cause the one or more processors to perform the method of any one of items 1 to 59. (Item 61) 61. A computer system comprising a hardware processor configured to execute the instructions in the non-transitory computer-readable medium of item 60.

Claims

1. 1. A computer-implemented method for constructing an optimized classifier for a condition, comprising: ranking a subset of a plurality of microsatellites as a classifier for the condition over a plurality of optimization cycles, the subset of the plurality of microsatellites comprising microsatellites in an initial population of microsatellites that are correlated with the condition, thereby identifying an optimized subset of the subset of microsatellites as the optimized classifier for the condition; The step of ranking comprises: (i) first performing a receiver operating characteristic (ROC) analysis using (1) the subset of the plurality of microsatellites, (2) microsatellites obtained from a reference subject having the condition, and (3) microsatellites obtained from a reference subject not having the condition; and (ii) determining, for an optimization cycle of the plurality of optimization cycles, a sum of the sensitivity and specificity of the classifier for the condition for each subset of the plurality of microsatellites; 11. The method of claim 10, wherein an optimization cycle of the plurality of optimization cycles comprises adding a new subset of the initial population of microsatellites to a subset from a previous optimization cycle of the plurality of optimization cycles, the new subset of the initial population of microsatellites comprising a first portion and a second portion of the new subset, the first portion of the new subset being created by randomly splitting and randomly recombining a randomly selected subset from the previous optimization cycle, and the second portion of the new subset being created by randomly selecting microsatellites from the initial population of microsatellites.

2. 2. The method of claim 1, further comprising comparing microsatellites in a first set of samples from subjects having the condition and microsatellites in a second set of samples from subjects not having the condition, thereby identifying said initial population of microsatellites.

3. 2. The method of claim 1, wherein said ranking step further comprises comparing said subset of said plurality of microsatellites to microsatellites in a sample from a subject having said condition and to microsatellites from a sample from a subject not having said condition.

4. 2. The method of claim 1, further comprising initializing the ranking by randomly selecting an initial subset of populations of microsatellites from the initial population of microsatellites for use in ranking in one optimization cycle of the plurality of optimization cycles.

5. 2. The method of claim 1, wherein a population of at least 100 subsets of said initial population of microsatellites is used in said multiple optimization cycles.

6. The method of claim 1, wherein the minimum number of microsatellites in one of the subsets of the plurality of microsatellites is eight.

7. The method of claim 1, wherein the maximum number of microsatellites in any one of the subsets of the plurality of microsatellites is 64.

8. The method of claim 1, wherein one subset of the plurality of microsatellites does not contain two identical microsatellites.

9. The method of claim 1, wherein the new subsets of the initial population of microsatellites include 10 new subsets.

10. 10. The method of claim 9, wherein the first portion of the new subsets includes seven of the ten new subsets and the second portion of the new subsets includes three of the ten new subsets.

11. The method of claim 10 , further comprising discarding a subset of ten optimization cycles based at least in part on having the lowest ranking in the optimization cycles.

12. 10. The method of claim 1, wherein the condition comprises an increased or decreased likelihood of developing a health condition in a subject.

13. 1. A computer system configured to construct an optimized classifier for a condition, comprising: A memory containing computer-executable instructions; and One or more processors, collectively or individually programmed to execute the computer-executable instructions the computer executable instructions comprising: ranking a subset of a plurality of microsatellites as a classifier for the condition over a plurality of optimization cycles, the subset of the plurality of microsatellites comprising microsatellites in an initial population of microsatellites that are correlated with the condition, thereby identifying an optimized subset of the subset of microsatellites as the optimized classifier for the condition; The step of ranking comprises: (i) first performing a receiver operating characteristic (ROC) analysis using (1) the subset of the plurality of microsatellites, (2) microsatellites obtained from a reference subject having the condition, and (3) microsatellites obtained from a reference subject not having the condition; and (ii) determining, for an optimization cycle of the plurality of optimization cycles, a sum of the sensitivity and specificity of the classifier for the condition for each subset of the plurality of microsatellites; a computer system comprising: an optimization cycle of the plurality of optimization cycles comprising adding a new subset of the initial population of microsatellites to a subset from a previous optimization cycle of the plurality of optimization cycles; the new subset of the initial population of microsatellites comprising a first portion and a second portion of the new subset; the first portion of the new subset being created by randomly splitting and randomly recombining a randomly selected subset from the previous optimization cycle; and the second portion of the new subset being created by randomly selecting microsatellites from the initial population of microsatellites.

14. 14. The computer system of claim 13, wherein the computer executable instructions further comprise comparing microsatellites in a first set of samples from subjects having the condition and microsatellites in a second set of samples from subjects not having the condition, thereby identifying the initial population of microsatellites.

15. 14. The computer system of claim 13, wherein the computer executable instructions further comprise comparing the subset of the plurality of microsatellites to microsatellites in a sample from a subject having the condition and to microsatellites in a sample from a subject not having the condition.

16. 14. The computer system of claim 13, wherein the computer executable instructions further comprise initializing the ranking by randomly choosing an initial subset of population of microsatellites from the initial population of microsatellites for use in ranking in one optimization cycle of the plurality of optimization cycles.

17. 14. The computer system of claim 13, wherein a population of at least 100 subsets of the initial population of microsatellites is used in the multiple optimization cycles.

18. The computer system of claim 13, wherein a subset of the plurality of microsatellites does not include two identical microsatellites.

19. 1. A non-transitory computer readable medium comprising instructions that, when executed by a processor of a processing system, cause the processing system to perform a method of constructing an optimized classifier for a condition, the method comprising: ranking a subset of a plurality of microsatellites as a classifier for the condition over a plurality of optimization cycles, the subset of the plurality of microsatellites comprising microsatellites in an initial population of microsatellites that are correlated with the condition, thereby identifying an optimized subset of the subset of microsatellites as the optimized classifier for the condition; The step of ranking comprises: (i) first performing a receiver operating characteristic (ROC) analysis using (1) said subset of microsatellites, (2) microsatellites obtained from a reference subject having said condition, and (3) microsatellites obtained from a reference subject not having said condition; and (ii) determining, for an optimization cycle of the plurality of optimization cycles, a sum of the sensitivity and specificity of the classifier for the condition for each subset of the plurality of microsatellites; 20. The method of claim 1, wherein an optimization cycle of the plurality of optimization cycles comprises adding a new subset of the initial population of microsatellites to a subset from a previous optimization cycle of the plurality of optimization cycles, the new subset of the initial population of microsatellites comprising a first portion and a second portion of the new subset, the first portion of the new subset being created by randomly splitting and randomly recombining a randomly selected subset from the previous optimization cycle, and the second portion of the new subset being created by randomly selecting microsatellites from the initial population of microsatellites.

20. The non-transitory computer-readable medium of claim 19, wherein the method further comprises comparing microsatellites in a first set of samples from subjects having the condition with microsatellites in a second set of samples from subjects not having the condition, thereby identifying the initial population of microsatellites.

Citation Information

Patent Citations

  • Computer system and methods for constructing biological classifiers and uses thereof

    US20070269804A1