Transcription factor analysis

Abstract

The methods and systems disclosed herein can improve the ability to analyze genomic material. The methods provided herein can examine transcription factor binding site accessibility to diagnose disease or monitor the progression of disease in a subject.

Description

[0001] Cross-references

[0002] This application claims the benefit of U.S. Provisional Patent Application No. 62 / 742,854 filed on October 8, 2018, U.S. Provisional Patent Application No. 62 / 752,270 filed on October 29, 2018, and U.S. Provisional Patent Application No. 62 / 849,097 filed on May 16, 2019, each of which is incorporated herein by reference in its entirety. Background Art

[0003] Transcription factors (TFs) regulate the expression of their target genes and can play key roles in development and differentiation. Genomic alterations can lead to the activation or inactivation of TFs, and the resulting disturbances in gene regulation can lead to physiological conditions (e.g., aging) or underlying diseases (e.g., cancer). To bind regulatory deoxyribonucleic acid (DNA), TFs often have to interact with nucleosomes, which can affect their occupancy and positioning.

[0004] Alterations in transcription factors can be important drivers of tumorigenesis in cancer, and TF-nucleosome interactions remain largely unmapped. However, non-invasive assays for assessing transcription factor activity are lacking.

[0005] Given the role of TFs in regulating chromatin accessibility and transcription, understanding the impact of genetic variation on TF binding can provide insights into the noncoding genetic components of development and disease. Chromatin accessibility ("open chromatin"), nucleosome positioning, and transcription factor (TF) occupancy can be measured separately using high-throughput, genome-wide approaches to gain major insights into the epigenetic information encoded within chromatin nucleoprotein architecture.

[0006] Dysregulation of transcription factors (TFs) may be an important driver of tumorigenesis. In order for TFs to bind DNA, the binding regions may need to be accessible. Therefore, TFs and chromatin remodeling complexes move and position nucleosomes to enhance accessibility. Therefore, a method is needed to dissect transcription factor binding sites to infer nucleosome position and chromatin accessibility. There is also a need for methods to use transcription factor binding site profiling and transcription factor binding site features to infer disease state, disease progression, and treatment responsiveness. Summary of the invention

[0007] The present disclosure provides methods and systems for assessing (e.g., modeling) transcription factor (TF) binding sites (TFBS) and using TFBS information to detect, assess, diagnose, and analyze disease states and identify treatment responsiveness.

[0008] Next generation sequencing based genome-wide assays can be used to provide TF binding patterns and associated chromatin architecture. Since nucleosomes and sequence-specific TFs bind to regulatory deoxyribonucleic acid (DNA) regions in a mutually exclusive manner, TFs compete or interact with nucleosomes, which affects their occupancy and positioning. In a given cell population, nucleosome occupancy refers to the average number of nucleosomes measured within a specified genomic region, while nucleosome positioning represents the probability of a reference point on a nucleosome (usually a dyad, such as the midpoint of a typical nucleosome) being present at a specific genomic coordinate.

[0009] Cell-free circulating nucleic acids, such as cell-free DNA (cfDNA), can provide an easily accessible source of nucleic acids for TFBS analysis. Such cfDNA can be the product of a digestion process that preferentially degrades DNA that is not protected by proteins (e.g., histone complexes). Cell-free DNA coverage patterns can reflect nucleosome positioning and occupancy caused by active binding of transcription factors to the genome. These nucleosome occupancy patterns measured by cfDNA can then be used to infer the activity of TFs in normal and tumor genomes.

[0010] TFs may preferentially bind in open chromatin, which may affect nucleosome positioning. Circulating cell-free DNA from plasma may represent mononucleosomal DNA, and nucleosomal plasma footprints may be informative about TFBSs.

[0011] The present disclosure provides methods and systems for mapping nucleosome positions from cfDNA to provide information about TF for applications related to disease identification, prediction, staging and / or identification of treatment responsiveness. Methods and systems for using transcription factor information determined from nucleosome footprints in nucleic acid molecules (e.g., cfDNA) are described herein. Information from nucleosome footprints in nucleic acid molecules can be used to evaluate, assess, detect and diagnose diseases, such as cancer. In some examples, this information can be characterized and used as input to machine learning models useful in many of these applications (e.g., disease identification, prediction, staging and identification of treatment responsiveness).

[0012] In one aspect, the present disclosure provides a computer-implemented method for determining a transcription factor binding profile in a nucleic acid sample from a subject, the method comprising: (a) providing a set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject; (b) using the set of sequence reads to generate a coverage pattern for the transcription factor; (c) processing the coverage pattern to provide a signal; and (d) processing the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby determining the transcription factor binding profile in the sample.

[0013] In some examples, the DNA is cell-free DNA.

[0014] In some examples, (c) includes using a low pass filter. In some examples, (c) includes using a Savitzky-Golay filter.

[0015] In some examples, the subject is a human.

[0016] In another aspect, the present disclosure provides a computer-implemented method for detecting the presence or absence of a disease in a subject, the method comprising: (a) providing a set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject; (b) using the set of sequence reads to generate a coverage pattern of a transcription factor; (c) processing the coverage pattern to provide a signal; and (d) processing the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby detecting the presence or absence of the disease in the subject.

[0017] In some examples, the DNA is cell-free DNA.

[0018] In some examples, the disease is cancer.

[0019] In some examples, (b) includes aligning the sequence read set to a reference sequence to provide an aligned sequence pattern, selecting a region of the aligned sequence pattern corresponding to a binding site for a transcription factor, and normalizing the aligned sequence pattern in the region. In some examples, (d) includes calculating an accessibility score for each binding site for the transcription factor.

[0020] In some examples, (c) includes using a low pass filter. In some examples, (c) includes using a Savitzky-Golay filter.

[0021] In some examples, the subject is a human.

[0022] In some examples, the transcription factor is a cancer-specific transcription factor. In some examples, the transcription factor is selected from GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, FOXA1, HNF-1A, HNF-4A, HNF-4G and DLX-2.

[0023] In some examples, accessibility scores for at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish between healthy subjects and cancer patients, disease progressors and non-progressors, multiple disease subtypes, multiple disease stages, disease treatment responders and non-responders, or any combination thereof.

[0024] In some examples, the transcription factor is selected from GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, and FOXA1.

[0025] In some examples, patient-specific and / or tumor-specific patterns were observed, including inferred binding patterns for the transcription factors AR, HOXB13, and NKX3-1.

[0026] In some examples, (d) includes identifying a signal of higher accessibility of a transcription factor. In some examples, the transcription factor is an epithelial transcription factor. In some examples, the transcription factor is GRH-L2.

[0027] In some examples, transcription factors GRHL2, FOXA1, and ZNF121 were associated with increased accessibility scores or open chromatin accessibility in breast cancer patients.

[0028] In some examples, open accessibility of the transcription factors GRHL2, FOXA1, and ZNF121 is indicative of breast cancer.

[0029] In some examples, transcription factors EVX2, DLX2, HNF1A, HNF4A, GRHL2, and HNF4G were associated with increased accessibility scores or open chromatin accessibility in colon cancer patients.

[0030] In some examples, open accessibility of transcription factors EVX2, DLX2, HNF1A, GRH-L2, HNF4A, and HNF4G is indicative of colon cancer.

[0031] In some examples, transcription factors LYL1 and PU.1 were associated with reduced accessibility scores or closed chromatin accessibility in colon cancer patients.

[0032] In some examples, blocked accessibility of transcription factors LYL1 and PU.1 is indicative of colon cancer.

[0033] In one example, open accessibility of the transcription factor TBX21 or EOMES indicates exhaustion of CD8 + T cells.

[0034] In one example, open accessibility of transcription factors selected from the group consisting of Eomes, Ybx21, Gata3, Rora, Bcl6, Blimp-1, von Hippel-Lindau tumor suppressor (VHL), Foxo1, IRF4, BATF, and NFATcl indicates CD8+ exhaustion. + T cells.

[0035] In some examples, the method further comprises detecting the presence or absence of a disease in a subject with an accuracy of at least about 70%. In some examples, the method further comprises detecting the presence or absence of a disease in a subject with an accuracy of at least about 80%. In some examples, the method further comprises detecting the presence or absence of a disease in a subject with an accuracy of at least about 90%.

[0036] In some examples, the method further comprises detecting the presence of the disease in the subject with a sensitivity of at least about 70%. In some examples, the method further comprises detecting the presence of the disease in the subject with a sensitivity of at least about 80%. In some examples, the method further comprises detecting the presence of the disease in the subject with a sensitivity of at least about 90%.

[0037] In some examples, the method further comprises detecting the absence of the disease in the subject with a specificity of at least about 70%. In some examples, the method further comprises detecting the absence of the disease in the subject with a specificity of at least about 80%. In some examples, the method further comprises detecting the absence of the disease in the subject with a specificity of at least about 90%.

[0038] In some examples, the method further comprises detecting the presence of the disease in the subject with a positive predictive value (PPV) of at least about 70%. In some examples, the method further comprises detecting the presence of the disease in the subject with a positive predictive value (PPV) of at least about 80%. In some examples, the method further comprises detecting the presence of the disease in the subject with a positive predictive value (PPV) of at least about 90%.

[0039] In some examples, the method further comprises detecting the absence of the disease in the subject with a negative predictive value (NPV) of at least about 70%. In some examples, the method further comprises detecting the absence of the disease in the subject with a negative predictive value (NPV) of at least about 80%. In some examples, the method further comprises detecting the absence of the disease in the subject with a negative predictive value (NPV) of at least about 90%.

[0040] In some examples, the method further comprises detecting the presence or absence of a disease in a subject with an area under a receiver operating characteristic (AUROC) of at least about 0.70. In some examples, the method further comprises detecting the presence or absence of a disease in a subject with an area under a receiver operating characteristic (AUROC) of at least about 0.80. In some examples, the method further comprises detecting the presence or absence of a disease in a subject with an area under a receiver operating characteristic (AUROC) of at least about 0.90.

[0041] In some examples, the method further includes applying the trained classifier to the signal to detect the presence or absence of the disease in the subject. In some examples, the method further includes applying the trained classifier to the accessibility score of the binding site of the transcription factor to detect the presence or absence of the disease in the subject. In some examples, the trained classifier includes a trained machine learning classifier. In some examples, the trained machine learning classifier includes a supervised machine learning algorithm. In some examples, the supervised machine learning algorithm includes one or more of the following: regression, support vector machine, tree-based method, neural network and random forest.

[0042] In another aspect, the disclosure provides methods that allow patients to be classified according to tumor type, including, for example, tumor subtype (e.g., subtypes of prostate cancer, colorectal cancer, breast cancer, lung cancer) or tumor stage, which can have important clinical implications for patient management (including treatment planning and responsiveness). Accordingly, provided herein are methods for mapping in vivo tumor-specific transcription factor binding based on patient samples (e.g., blood, plasma, or serum samples), thereby making key portions of the non-coding genome suitable for clinical analysis.

[0043] In some examples, the method includes distinguishing subtypes of the disease.

[0044] In some examples, the method includes distinguishing a subtype of the cancer.

[0045] In some examples, the method includes distinguishing subtypes of prostate cancer, colorectal cancer, breast cancer, and lung cancer.

[0046] In some examples, the method includes distinguishing a prostate cancer subtype, for example, among patients having prostate adenocarcinoma or small cell neuroendocrine prostate cancer.

[0047] In some examples, the method includes distinguishing the stage of the cancer (eg, among stage I, stage II, stage III, and stage IV cancer).

[0048] In some examples, the method includes distinguishing stage I and stage II cancer from stage III and stage IV cancer.

[0049] In some examples, transcription factors GRHL2, FOXA1, HOXB13, AR, and NKX3-1 were associated with increased accessibility scores or open chromatin accessibility in prostate adenocarcinoma patients.

[0050] In some examples, open accessibility of transcription factors GRHL2, FOXA1, HOXB13, AR, and NKX3-1 is indicative of prostate adenocarcinoma.

[0051] In some examples, the transcription factors REST, GRHL2, FOXA1, HOXB13, AR, and NKX3-1 are associated with reduced or closed chromatin accessibility in patients with small cell neuroendocrine prostate cancer.

[0052] In some examples, decreased accessibility of the transcription factors REST, GRHL2, GRHL3, FOXA1, FOXA2, GATA2, GATA3, HOXB13, AR, and NKX3-1 is indicative of small cell neuroendocrine prostate cancer.

[0053] In some examples, increased accessibility of the transcription factors GLIS1, SOX2, and SOX11 is indicative of small cell neuroendocrine prostate cancer.

[0054] In another aspect, the present disclosure provides a system comprising: a computing device comprising at least one computer processor; an operating system configured to execute executable instructions; a memory; and a computer program containing instructions executable by the computing device to provide a computer application for detecting the presence or absence of a disease in a subject, the computer application comprising: a sequence module programmed to acquire a set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject; a coverage module programmed to use the set of sequence reads to generate a coverage pattern of a transcription factor; a signal module programmed to process the coverage pattern to provide a signal; and a detection module programmed to process the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby detecting the presence or absence of the disease in the subject.

[0055] In another aspect, the present disclosure provides a non-transitory computer-readable medium comprising machine executable code, which, when executed by one or more computer processors, implements a method for detecting the presence or absence of a disease in a subject, the method comprising: (a) providing a set of sequence reads from deoxyribonucleic acid (DNA) extracted from a subject; (b) using the set of sequence reads to generate a coverage pattern of a transcription factor; (c) processing the coverage pattern to provide a signal; and (d) processing the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby detecting the presence or absence of the disease in the subject.

[0056] On the other hand, the present disclosure provides a system for detecting the presence or absence of a disease in a subject, the system comprising: a database comprising a set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are programmed individually or collectively to: (a) use the set of sequence reads to generate a coverage pattern of a transcription factor; (b) process the coverage pattern to provide a signal; (d) process the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby detecting the presence or absence of a disease in the subject.

[0057] In another aspect, the present disclosure provides a computer-implemented method for monitoring progression or regression of a disease in a subject, the method comprising: (a) providing a first set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second set of sequence reads from the DNA extracted from the subject at a second time later than the first time; (b) generating a first coverage pattern for a transcription factor using the first set of sequence reads, and generating a second coverage pattern for the transcription factor using the second set of sequence reads; (c) processing the first coverage pattern to provide a first signal, and processing the second coverage pattern to provide a second signal; (d) processing the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (e) processing the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (f) monitoring progression or regression of the disease in the subject based on processing the first signal and the second signal with the reference signal.

[0058] In some examples, accessibility scores for at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish between disease progressors and non-progressors, multiple disease subtypes, multiple disease stages, or any combination thereof.

[0059] In some examples, accessibility scores for at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish between disease treatment responders and non-responders.

[0060] In some examples, the second overlay pattern indicates a phenotypic change in the tumor during the course of the disease.

[0061] In some examples, the phenotypic change is a change in the stage of the cancer from androgen-dependent to androgen-independent.

[0062] In some examples, the DNA is cell-free DNA.

[0063] In some examples, the disease is cancer.

[0064] In some examples, (b) includes aligning the first sequence read set and the second sequence read set with a reference sequence to provide a first aligned sequence pattern and a second aligned sequence pattern, respectively, selecting a region of the first aligned sequence pattern and the second aligned sequence pattern corresponding to a binding site of a transcription factor, and normalizing the first aligned sequence pattern and the second aligned sequence pattern in the region.

[0065] In some examples, (c) includes using a low pass filter. In some examples, (c) includes using a Savitzky-Golay filter.

[0066] In some examples, the subject is a human.

[0067] In some examples, the transcription factor is a cancer-specific transcription factor.

[0068] In some examples, the transcription factor is selected from GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, and FOXA1.

[0069] In some examples, the transcription factor is selected from HNF-1a, HNF-4a, HNF-4g, EVX-2, and DLX-2.

[0070] In some examples, the method further comprises adjusting a treatment regimen for the subject's disease based on (f).

[0071] In another aspect, the present disclosure provides a system comprising: a computing device comprising at least one computer processor; an operating system configured to execute executable instructions; a memory; and a computer program comprising instructions executable by the computing device to provide a computer application for monitoring progression or regression of a disease in a subject, the computer application comprising: a sequence module programmed to acquire a first set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second set of sequence reads from DNA extracted from the subject at a second time later than the first time; a coverage module programmed to use the first sequence to generate a first set of sequence reads from the DNA extracted from the subject at a second time later than the first time; a second ... A column read set generates a first coverage pattern for a transcription factor, and a second sequence read set is used to generate a second coverage pattern for the transcription factor; a signal module, which is programmed to process the first coverage pattern to provide a first signal, and to process the second coverage pattern to provide a second signal; a first processing module, which is programmed to process the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; a second processing module, which is programmed to process the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and a detection module, which is programmed to monitor the progression or regression of a disease in a subject based on processing the first signal and the second signal with the reference signal.

[0072] In another aspect, the present disclosure provides a non-transitory computer-readable medium comprising machine executable code, which when executed by one or more computer processors implements a method for monitoring progression or regression of a disease in a subject, the method comprising: (a) providing a first sequence read set from deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second sequence read set from DNA extracted from the subject at a second time later than the first time; (b) generating a first coverage pattern for a transcription factor using the first sequence read set, and generating a second coverage pattern for the transcription factor using the second sequence read set; (c) processing the first coverage pattern to provide a first signal, and processing the second coverage pattern to provide a second signal; (d) processing the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (e) processing the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (f) monitoring progression or regression of the disease in the subject based on processing the first signal and the second signal with the reference signal.

[0073] On the other hand, the present disclosure provides a system for monitoring the progression or regression of a disease in a subject, the system comprising: a database comprising a first sequence read set from deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second sequence read set from DNA extracted from the subject at a second time later than the first time; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually or collectively programmed to: (a) generate a first coverage pattern for a transcription factor using the first sequence read set, and generate a second coverage pattern for the transcription factor using the second sequence read set; (b) process the first coverage pattern to provide a first signal, and process the second coverage pattern to provide a second signal; (c) process the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (d) process the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (e) monitor the progression or regression of the disease in the subject based on processing the first signal and the second signal with the reference signal.

[0074] In another aspect, the present disclosure provides a system for determining a transcription factor binding profile in a nucleic acid sample from a subject, the system comprising a processor configured to: (a) analyze a set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject; (b) use the set of sequence reads to generate a coverage pattern of the transcription factor; (c) process the coverage pattern to provide a signal; and (d) process the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby determining the transcription factor binding profile.

[0075] In some examples, the DNA is cell-free DNA.

[0076] In some examples, (c) includes using a low pass filter. In some examples, (c) includes using a Savitzky-Golay filter.

[0077] In some examples, the subject is a human.

[0078] In another aspect, the present disclosure provides a system for detecting the presence or absence of a disease in a subject, comprising a processor configured to: (i) generate a coverage pattern of a transcription factor using a set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject; (ii) process the coverage pattern to provide a signal, wherein the signal has a frequency different from a reference signal; and (iii) process the signal with a reference signal to detect the presence or absence of the disease in the subject.

[0079] In some examples, the present disclosure provides a system for classifying a tumor according to a tumor subtype or a tumor stage, comprising a processor configured to: (i) generate a first coverage pattern of a transcription factor and a second coverage pattern of a transcription factor using a first sequence read set from deoxyribonucleic acid (DNA) extracted from a subject at a first time and a second sequence read set extracted from the DNA from the subject at a second time later than the first time; (ii) process the first coverage pattern to provide a first signal, and process the second coverage pattern to provide a second signal, wherein the first signal and the second signal have different frequencies from a reference signal; and (iii) process the first signal with the reference signal, and process the second signal with the reference signal to monitor progression or regression of a disease in the subject.

[0080] In another aspect, the present disclosure provides a system for monitoring progression or regression of a disease in a subject, comprising a processor configured to: (i) generate a first coverage pattern of a transcription factor and a second coverage pattern of a transcription factor using a first sequence read set from deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second sequence read set extracted from the DNA from the subject at a second time later than the first time; (ii) process the first coverage pattern to provide a first signal, and process the second coverage pattern to provide a second signal, wherein the first signal and the second signal have different frequencies from a reference signal; and (iii) process the first signal with the reference signal, and process the second signal with the reference signal, to monitor progression or regression of the disease in the subject.

[0081] 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, implements any of the methods above or elsewhere herein.

[0082] Another aspect of the present disclosure provides a system comprising one or more computer processors and a computer memory coupled thereto. The computer memory contains machine executable code that implements any method above or elsewhere herein when executed by one or more computer processors.

[0083] On the other hand, the present disclosure provides a method for determining a tumor-specific TFBS pattern, the method comprising: (a) providing a first sequence read set from deoxyribonucleic acid (DNA) extracted from a subject at a first time and a second sequence read set from DNA extracted from the subject at a second time later than the first time; (b) generating a first coverage pattern of a transcription factor using the first sequence read set, and generating a second coverage pattern of the transcription factor using the second sequence read set; (c) processing the first coverage pattern to provide a first signal, and processing the second coverage pattern to provide a second signal; (d) processing the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (e) processing the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (f) determining a tumor-specific TFBS pattern based on processing the first signal and the second signal with the reference signal.

[0084] In another aspect, the present disclosure provides a system comprising: a computing device comprising at least one computer processor; an operating system configured to execute executable instructions; a memory; and a computer program comprising instructions executable by the computing device to provide a computer application for determining a tumor-specific TFBS pattern, the computer application comprising: a sequence module programmed to acquire a first set of sequence reads from deoxyribonucleic acid (DNA) extracted from a subject at a first time and a second set of sequence reads from DNA extracted from the subject at a second time later than the first time; a coverage module programmed to generate a first coverage pattern for a transcription factor using the first set of sequence reads and to generate a second coverage pattern for the transcription factor using the second set of sequence reads; a signal module programmed to process the first coverage pattern to provide a first signal and to process the second coverage pattern to provide a second signal; a first processing module programmed to process the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; a second processing module programmed to process the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and a detection module programmed to determine the tumor-specific TFBS pattern based on processing the first signal and the second signal with the reference signal.

[0085] In another aspect, the present disclosure provides a non-transitory computer-readable medium comprising machine executable code, which, when executed by one or more computer processors, implements a method for determining a tumor-specific TFBS pattern, the method comprising: (a) providing a first set of sequence reads from deoxyribonucleic acid (DNA) extracted from a subject at a first time and a second set of sequence reads from DNA extracted from the subject at a second time later than the first time; (b) generating a first coverage pattern for a transcription factor using the first set of sequence reads, and generating a second coverage pattern sequence reads for the transcription factor using the second set of sequence reads; (c) processing the first coverage pattern to provide a first signal, and processing the second coverage pattern to provide a second signal; (d) processing the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (e) processing the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (f) determining a tumor-specific TFBS pattern based on processing the first signal and the second signal with the reference signal.

[0086] On the other hand, the present disclosure provides a system for monitoring the progression or regression of a disease in a subject, the system comprising: a database comprising a first set of sequence reads from deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second set of sequence reads from DNA extracted from the subject at a second time later than the first time; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are programmed individually or collectively to: (a) generate a first coverage pattern for a transcription factor using the first set of sequence reads, and generate a second coverage pattern for the transcription factor using the second set of sequence reads; (b) process the first coverage pattern to provide a first signal, and process the second coverage pattern to provide a second signal; (c) process the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (d) process the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (e) determine a tumor-specific TFBS pattern based on processing the first signal and the second signal with the reference signal.

[0087] 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, implements any of the methods above or elsewhere herein.

[0088] Another aspect of the present disclosure provides a system comprising one or more computer processors and a computer memory coupled thereto. The computer memory contains machine executable code that implements any method above or elsewhere herein when executed by one or more computer processors.

[0089] Other aspects and advantages of the present disclosure will become apparent to those skilled in the art through the following detailed description, in which only illustrative examples of the present disclosure are shown and described. As will be appreciated, the present disclosure can have other and different examples, and its multiple details can be modified in various obvious aspects, all without departing from the present disclosure. Therefore, the drawings and description are to be regarded as illustrative rather than restrictive in nature.

[0090] Incorporation by reference

[0091] 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 the disclosure contained in the specification, the specification is intended to supersede and / or take precedence over any such conflicting material. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] The features of the present invention are particularly set out in the appended claims. A better understanding of the features and advantages of the present method and system will be obtained by referring to the following detailed description, which sets forth an illustrative example of the principles of the present method and system, and the accompanying drawings, in which (also referred to as "Figures" in this specification):

[0093] Figure 1 A computer system programmed or otherwise configured to perform methods of the present disclosure, such as storing, processing, identifying, or interpreting subject (e.g., patient) data, biological data, biological sequences, reference sequences, transcription factor (TF) binding site (TFBS) data, or TFBS features such as z-scores or TFBS accessibility scores, is shown.

[0094] Fig. 2A-Fig. 2E shows the TF-nucleosome interaction established by cell-free deoxyribonucleic acid (cfDNA). Fig. 2A shows the region with highly organized (e.g., phased) nucleosomes leading to oscillating read depth patterns, wherein the peak read indicates the position of the dyad (e.g., the midpoint of a typical nucleosome). Less clear nucleosome positioning produces a fairly shallow coverage spectrum. Fig. 2B shows the TFBS data of 676 TFs retrieved from GTRD, and it is compared with a selected list of known or possible human TFs. Three different calculations were performed, each with increased stringency. Fig. 2C shows that in all analyzed cfDNAs, the coverage pattern of CCCTC binding factor (CTCF) is similar, which is consistent with DNase hypersensitivity data, and DNase hypersensitivity data show that the accessibility in blood (GM12878) and epithelial tissues (e.g., prostate (LNCaP) and colon (HCT116)) is roughly equal. In this sub-figure and each subsequent sub-figure, the spectrum calculated by the healthy control is shown in gray, while the spectrum of the patient source is shown in the color indicated. Figure 2D shows that the hematopoietic lineage specificity of TF (PU.1, LYL1, SPIB) was confirmed by DNA hypersensitivity assay, and their amplitude was reduced in cancer patient plasma compared to healthy controls. In contrast, the amplitude of epithelial TFGRHL2 was increased in cfDNA from cancer patients. Figure 2E shows the accessibility map and DNase hypersensitivity of TFFOXA1, showing preferential amplitude changes in patients with hormone-dependent cancers (e.g., prostate and breast cancer).

[0095] Figure 3A-Figure 3F shows the accessibility score of TFBS characterization. Figure 3A shows how to determine TF accessibility. In order to measure TF accessibility, different window sizes are used to split the observed original coverage signal (purple in the left sub-figure and black in the right sub-figure) into low-frequency signals (red) and high-frequency signals (blue) by Savitzky-Golay filtering. The right sub-figure shows the superposition of these three signals. The high-frequency signal is used as a measure of accessibility. Figure 3B shows that the range of the high-frequency signal (Y-axis) depends mainly on the number of TFBS (X-axis), because TFs with fewer binding sites have more noise due to smaller averages. The LOESS model (blue) is fitted to correct this deviation. Figure 3C shows a wavelet analysis of GRHL2: a heat map of the cycle along the area around the TFBS of GRHL2 (left sub-figure). The color code represents the quantile of the signal power distribution. The average power of the cycle of the transcription factor GRHL2 (right sub-figure). Figure 3D shows the original (black) and reconstructed (red) nucleosome coverage spectrum of the transcription factor GRHL2 detrended by wavelet analysis. Figure 3E shows that all tested processes (left: >50%-TFBS, Savitzky-Golay filtering; center: power sum, wavelet analysis; right: 1,000-msTFBS, Savitzky-Golay filtering) showed increased values ​​as a measure of accessibility for transcription factors expressed in blood (greater than 10 FPKM), but not in genes with no expression or low expression (<0.1 FPKM). Figure 3F shows that transcription factors with an average DNase hypersensitivity coverage greater than 2 in the GM12878 DNase data from the ENCODE project have a higher adjustment range and higher power sum than factors with an average coverage <1 in all three analyses performed (left: >50%-TFBS, Savitzky-Golay filtering; center: power sum, wavelet analysis; right: 1,000-msTFBS, Savitzky-Golay filtering).

[0096] Fig. 4A-Fig. 4F shows prostate lineage specific TF, its plasticity and suitability for tumor classification.Fig. 4A shows that prostate adenocarcinoma is AR dependent, and therefore has often increased PSA (prostate specific antigen) level and normal NSE (neuron specific enolase) value.On the contrary, t-SCNC no longer depends on AR, and generally has lower PSA and increased NSE level.Identify the multiple TF involved in the transdifferentiation process from adenocarcinoma to t-SCNC, and indicate with arrows.Fig. 4B shows the accessibility spectrum of prostate lineage specific homeobox TF HOXB13 and the respective DNase hypersensitivity assay of prostate cancer cell line LNCaP.In this sub-figure and subsequent sub-figure, the spectrum calculated by healthy control is shown in gray, and the spectrum of patient origin is shown in the color indicated.Fig. 4C shows the accessibility pattern and DNA hypersensitivity assay of NKX3-1, which is one of the earliest genes expressed in the prostate epithelial maturation process. Fig. 4D shows the AR accessibility of all AR binding sites in GTRD and AR binding sites (T-ARBS) with higher binding strength in tumors and sites with high binding strength in normal samples (normal AR binding sites, N-ARBS) (Pomerantz et al., 2015). DNA hypersensitivity assay confirmed the well-established lineage specificity of AR. Fig. 4E shows the coverage pattern changes during the transdifferentiation process from adenocarcinoma to neuroendocrine carcinoma, which is established by two plasma samples from patient P148, for hormone dependence (AR, FOXA1), tissue identity specificity (HOXB13, NKX3-1) and neuroendocrine reprogramming (REST, N-MYC) TF. Fig. 4F shows the analysis of the same TF as Fig. 4A from 4 plasma samples of patients with neuroendocrine prostate cancer.

[0097] Figure 5 Somatic copy number alterations (SCNAs) in plasma samples from cancer patients are shown. SCNAs were identified after whole genome sequencing of eight plasma samples from four patients (C2, P40, P147, and P148).

[0098] Figure 6A-Figure 6E shows TF-nucleosome interaction maps for 676 high-confidence TFs with reliable binding site information. Figure 6A shows the TFBS-nucleosome coverage profiles of two representative TFs (CREM and GATAD1) established from 24 cfDNA samples from healthy controls, each shown with a separate blue line. The MNase-seq coverage pattern of the lymphoblastoid cell line GM12878 obtained from ENCODE is shown in red. Fig.17Additional MNase maps are shown in . Figure 6B shows a heat map of fragment sizes around CTCF binding sites, showing the length of each sequencing read (Y-axis) as a function of the distance from the midpoint of the fragment to the center of the site (X-axis) for each annotated feature. Figure 6C shows a heat map of a single CTCF binding site and surrounding area. The regions were ranked by coverage within the central 50 base pairs (bp) around the TFBS. The spatial density of cfDNA fragments within a 1 kilobase (kb) region centered on the TFBS was calculated and ranked. Figure 6D shows an overlap matrix between TFBSs (left sub-figure: all 676 GRTD TFs; right sub-figure: 505 TFs with 1,000-msTFBSs). Each point represents the percentage of overlap in the binding site definition (within approximately 50 bp). Figure 6E shows that TFBS analysis performed with high molecular weight DNA (not mononucleosomal DNA) produced a uniform, non-oscillating pattern (blue) compared to plasma DNA (green).

[0099] Figure 7 TF-nucleosome interaction maps for various TFs are shown. Additional comparisons between coverage profiles of cfDNA and MNase-seq around transcription factor binding sites are shown.

[0100] Figures 8A-8E show the shapes of TFBS. Figure 8A shows the coverage spectra of TF AP-4 and BCL-3 after calculation for TFBSs within and outside the TSS, respectively. Figure 8B shows that analysis of TFBSs for TF ATF1, CREB, CREM, and ATF-3 can result in uniformly spaced or TSS-like coverage patterns, depending on whether all tissues in the GTRD are included, or whether more strictly only those peaks supported by more than 50% of the maximum number of samples (>50%-TFBS) are included. Figure 8C shows examples of TF-nucleosome spectra calculated for all and>50%-TFBS (upper sub-figure) and 1,000-msTFBS (lower sub-figure), showing variable nucleosome patterns of different TFs in cfDNA. Figure 8D shows that measurements of TFBS width reveal huge differences between various TFBSs. Figure 8E shows a box plot showing the overlap percentage of CpG islands (left sub-figure) and TSSs (right sub-figure).

[0101] Fig. 9 Analysis of pooled shallow coverage cfDNA is shown. Accessibility of pooled cfDNA samples from prostate (n=69), colon (n=100) and breast (n=60) cancer cases for epithelial TF GRHL2 and hematopoietic TFs (PU.1, LYL1 and SPIB) is shown. Accessibility within the prostate cancer cfDNA pool for lineage-specific TFs AR, HOXB13 and NKX3-1 is also shown.

[0102] Figures 10A-10B show transcription factors involved in transdifferentiation from adenocarcinoma to t-SCNC. Figure 10A shows GRHL2 accessibility in plasma samples P148_1 and P148_3 from patient P148. Figure 10B shows analysis of GLIS1 in two plasma samples from patient P148.

[0103] Fig.11 Downsampling of plasma samples P148_1 and P148_3 from patient P148 is shown. Plasma samples P148_1 (819,607,690 reads) and P148_3 (768,763,081 reads) were downsampled to approximately 50 million reads and analyzed for 1,000-ms TFBS (left column) and all and >50%-TFBS (right column). The analysis showed that TFs with a small number of TFBSs were preferentially affected by the increase in noise.

[0104] Fig.12 Comparison of TFBS accessibility in serial analysis is shown. Correlation plot between serial samples from patients C2, P147, P40 and P148. The X-axis represents the first plasma sample and the Y-axis represents the second plasma sample.

[0105] Figures 13A-13B show the establishment of TF-nucleosome interactions. Figure 13A shows the TFBS-nucleosome profiles of four TFs, such as SP1 and SP2, which mainly bind to common sites in the genome and also co-bind with NF-YA and NF-YB. Figure 13B shows the TF-nucleosome interactions depicted by the average nucleosome occupancy profile established from plasma DNA for the hematopoietic cell lineage-specific TFs PU.1, LYL1 and SPIB and the epithelial cell-specific TF GRHL2. The different amplitudes may reflect the different contributions of DNA released from hematopoietic cells and epithelial cells to the circulation.

[0106] Figure 14A-Figure 14B shows that CTCF is an excellent example of characterizing different TFBS, and shows that accessibility scores are characterizations of TFBS. Figure 14A shows various binding sites associated with CTCF and TAD or TSS. The coverage pattern of CTCF is divided into CTCF sites overlapping with TAD boundaries (red) or outside TAD boundaries (orange), CTCF sites near TSS (e.g., within about 2kbp; green) or away from TSS (greater than 2kbp; blue) and CTCF sites (black) (left sub-figure) for the complete GTRD data set ultra-conserved and only those peaks (right sub-figure) supported by more than 50% of the maximum number of analyzed samples. Figure 14B shows TF-nucleosome spectra, showing the variability of their patterns.

[0107] Fig.15 It is shown that the oncogenes c-Jun (upper panel) and JunD (lower panel) showed increased accessibility only in CRC patient C2, and the relative colon specificity was confirmed by DNA hypersensitivity assay.

[0108] Figure 16A-Figure 16C shows the accessibility of TF changes during the transdifferentiation of prostate cancer. In particular, Figure 16A-Figure 16C shows that the coverage pattern changes after neuroendocrine differentiation in sample P148. Figure 16A shows that nucleosome phasing changes significantly in the site defined by GTRD and tumor-specific AR binding site (defined by Pomerantz et al.) in the androgen receptor binding site. Figure 16B shows that nucleosome phasing also decreases sharply in other transcription factors of the AR axis. Phasing is significant in sample 1, but mostly disappears in sample 3. Figure 16C shows that the inhibitory factors (ZNF644, REST) ​​that play a role in neurogenesis in sample 3 are mostly inactivated.

[0109] Fig.17 TF-nucleosome interaction maps of 676 high-confidence TFs with reliable binding site information are shown. TF-nucleosome profiles are classified according to their accessibility scores and the number of TFBSs.

[0110] Figures 18A-18B show analysis of prostate cancer cases with tandem replicon phenotype or chromothripsis, including four cfDNA samples with tandem replicon phenotype (P21_2, P111_1, P111_4, and P166_1) and one case with chromothripsis on chromosome 10 (P143_3). Accessibility of epithelial TFs FOXA1, GRHL1, and GRHL2 and prostate lineage-specific TFs AR, HOXB13, and NKX3-1.

[0111] Fig.19 Shown is the analysis of AR binding sites from plasma samples of P40. This prostate cancer patient was treated with ADT and developed high levels of AR amplification between samples P40_1 and P40_2.

[0112] Fig. 20 A diagram showing how epigenetic control regions affect nucleosome positioning is shown. Histone modifications and enhancers are illustrated.

[0113] Fig.21 Nucleosomal positioning of selected TFs is shown.

[0114] Fig. 22 The coverage pattern of the selected TF is shown.

[0115] Fig.23The overlap of different TFs is shown. These overlap values ​​correspond to the heatmap of Figure 2C.

[0116] Fig.24 The effect of TFBS size is shown.

[0117] Fig.25 Nucleosome patterns of REST and KLF16 from samples of 24 healthy individuals are shown. Each line represents a different individual. In 24 healthy individuals, under the same settings of transcription factors active in blood cells, these patterns appear to be almost identical.

[0118] Figures 26A-26C show the nucleosome positioning of selected TFs for advanced cancer samples. The CTCF patterns in all samples looked similar. The activity of blood-specific TFs including PU.1, Lyl-1 and Spi-B was reduced in cancer samples. Cancer-specific TFs including GRH-L2 (epithelial marker), ASH-2 and HOX-B13 (prostate cancer markers of the androgen receptor axis) and EVX2 (colon cancer marker) were more active.

[0119] Fig. 27 Included is a list of TFs that can be used in the methods and systems provided herein.

[0120] Figure 28A-28B Included are TFs that can be used in the methods and systems provided herein. Fig.28A Includes TFs with binding sites larger than 300 bp, while Fig.28B TFs with binding sites close to the dinucleosome size (between 312-352 bp) were included.

[0121] Figure 29A-Figure 29E provides the identification of transcription factors with accessibility changes in plasma samples from cancer patients. Figure 29A provides TFBS analysis of plasma samples from healthy donors (NPH001). Each point represents a TF, the y-axis shows the accessibility value, and the x-axis shows the overall z score as a measure of accessibility deviation from normal control samples. In samples from healthy donors (compared with all other healthy donors), only a few TFs exceed the z score (grey dotted line) of ± 3, and no TF exceeds the z score (red line) threshold of ± 5. As in Figure 29A, Figure 29B provides an overall z score graph, but the plasma sample is derived from a prostate cancer patient (P40). As in Figure 29A, Figure 29C provides an overall z score graph for plasma samples C2_6. Figure 29D provides a nucleosome position spectrum for TF EVX2, plasma DNA from healthy controls (grey spectrum) and two plasma samples (blue and red) derived from colon cancer patients C2. Figure 29E provides a bar graph of the overall z-score plot for the combined breast, prostate, and colon cancer libraries. The left panel shows TFs with increased accessibility in at least one tumor entity; the right panel summarizes the accessibility of hematopoietic-related TFs.

[0122] Figures 30A-30B provide graphs showing the limits of plasma resolution and early cancer detection based on TFs. Figure 30A provides a graph showing the accessibility comparison of selected TFs based on their tumor fractions in the COAD cohort subsamples. Figure 30B provides a graph showing the logistic regression of samples from the Phase I (left sub-graph) and Phase II (right sub-graph) colon cancer cohorts with all 504 TFs, respectively. All results given are cross-validated test settings. DETAILED DESCRIPTION

[0123] Although various embodiments of the present invention have been shown and described herein, it will be apparent to those skilled in the art that these embodiments are provided by way of example only. Those skilled in the art may conceive of multiple variations, changes and substitutions without departing from the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed.

[0124] Where values ​​are described as ranges, it will be understood that such disclosure includes all possible sub-ranges within those ranges as well as disclosure of specific values ​​falling within those ranges, regardless of whether a specific value or specific sub-range is explicitly stated.

[0125] As used herein, the term "accessibility score" generally refers to a measure of the accessibility of each transcription factor (TF) binding site. Since transcription factor binding can open or "prime" its target enhancers without activating them themselves, the graded values ​​are called "accessibility scores". Accessibility scores can be used to objectively compare the accessibility of TFBSs in serial analyses from the same person or different individuals. The score provides a reliable assessment of TFBS accessibility, and is particularly suitable for use in clinical diagnosis, cancer detection, and treatment monitoring using cfDNA.

[0126] As used herein, the term "aligned sequence pattern" generally refers to the spatial pattern of sequence reads after alignment to a reference genome.

[0127] As used herein, the term "circulating free DNA" or "cell-free DNA" (cfDNA) generally refers to deoxyribonucleic acid (DNA), which was first detected in human plasma in 1948. (Mandel, P. Metais, P., CR Acad. Sci. Paris, 142, 241-243 (1948)). Since then, its connection with disease has been established in several fields. (Tong, YK Lo, YM, Clin Chim Acta, 363, 187-196 (2006)). Studies have shown that many circulating nucleic acids in the blood come from necrotic or apoptotic cells (Giacona, MB et al., Pancreas, 17, 89-97 (1998)), and a large increase in the level of nucleic acids from apoptosis is observed in diseases such as cancer. (Giacona, MB et al., Pancreas, 17, 89-97 (1998); Fournie, GJ et al., Cancer Lett, 91, 221-227 (1995)). Particularly for cancer, where circulating DNA carries hallmarks of disease, including mutations in oncogenes, microsatellite alterations, and, for some cancers, viral genomic sequences, DNA or RNA in plasma have been increasingly investigated as potential biomarkers of disease. 16266-16271 (2008)).

[0128] The cell-free portion can be serum or plasma. As used herein, the term "cell-free portion" of a biological sample generally refers to a portion of a biological sample that is substantially free of cells. As used herein, the term "substantially free of cells" generally refers to a preparation from a biological sample that contains less than about 20,000 cells / mL, less than about 2,000 cells / mL, less than about 200 cells / mL, or less than about 20 cells / mL. Genomic DNA may not be excluded from a non-cellular sample and generally accounts for about 50% to about 90% of the nucleic acids present in the sample.

[0129] As used herein, the term "coverage pattern" generally refers to the spatial arrangement of sequence reads after alignment to a reference genome. Coverage pattern identifies the coverage and depth of a next-generation sequencing method.

[0130] As used herein, the term "derived from" generally refers to an origin or source, and may include naturally occurring, recombinant, unpurified or purified molecules. Nucleic acids derived from original nucleic acids may comprise the original nucleic acids in part or in whole, and may be fragments or variants of the original nucleic acids. Nucleic acids derived from biological samples may be purified from the sample.

[0131] As used herein, the terms "diagnose" or "diagnosis" of a condition or outcome generally refer to predicting or diagnosing a condition or outcome, determining susceptibility to a condition or outcome, monitoring treatment of a subject (e.g., a patient), diagnosing a subject's (e.g., a patient's) response to treatment, and prognosis of the condition or outcome, progression, and response to a particular treatment.

[0132] As used herein, the term "nucleic acid" generally refers to a polynucleotide comprising two or more nucleotides. It can be DNA or RNA. Nucleic acid can be a polymeric form of nucleotides (deoxyribonucleotides (dNTPs) or ribonucleotides (rNTPs) or their analogs) of any length. Nucleic acids can have any three-dimensional structure and can perform any known or unknown function. Non-limiting examples of nucleic acids include deoxyribonucleic acid (DNA), ribonucleic acid (RNA), coding or non-coding regions of genes or gene fragments, loci (gene sites) defined from linkage analysis, exons, introns, messenger RNA (mRNA), transfer RNA, ribosomal RNA, short interfering RNA (siRNA), short hairpin RNA (shRNA), micro RNA (miRNA), ribozymes, cDNA, recombinant nucleic acids, branched nucleic acids, plasmids, vectors, isolated DNA of any sequence, isolated RNA of any sequence, nucleic acid probes and primers. Nucleic acids can include one or more modified nucleotides, such as methylated nucleotides and nucleotide analogs. If modifications to the nucleotide structure exist, they can be performed before or after nucleic acid assembly. The nucleotide sequence of a nucleic acid may be interrupted by non-nucleotide components. The nucleic acid may be further modified after polymerization, for example by conjugation or binding to a reporter. A "variant" nucleic acid is a polynucleotide having a nucleotide sequence identical to that of its original nucleic acid except for having at least one modified (e.g., deleted, inserted, or substituted, respectively) nucleotide. A variant may have a nucleotide sequence that is at least about 80%, 90%, 95%, or 99% identical to the nucleotide sequence of the original nucleic acid.

[0133] As used herein, the term "target nucleic acid" generally refers to a nucleic acid molecule in a starting population of nucleic acid molecules having a nucleotide sequence whose presence, amount and / or sequence or a change in one or more of these is desired to be determined. The target nucleic acid can be any type of nucleic acid, including DNA, RNA and the like. As used herein, "target ribonucleic acid (RNA)" generally refers to a target nucleic acid that is RNA. As used herein, "target deoxyribonucleic acid (DNA)" generally refers to a target nucleic acid that is DNA.

[0134] As used herein, the term "amplification" generally refers to increasing the size or amount of a nucleic acid molecule. Nucleic acid molecules can be single-stranded or double-stranded. Amplification can include generating one or more copies or "amplification products" of nucleic acid molecules. Amplification can be performed, for example, by extension (e.g., primer extension) or connection. Amplification can include performing a primer extension reaction to generate a strand complementary to a single-stranded nucleic acid molecule, and in some cases generating one or more copies of a strand and / or a single-stranded nucleic acid molecule. The term "DNA amplification" generally refers to generating one or more copies or "amplified DNA products" of a DNA molecule. The term "reverse transcription amplification" generally refers to generating deoxyribonucleic acid (DNA) from a ribonucleic acid (RNA) template by the action of a reverse transcriptase.

[0135] The term "transcription factor" generally refers to a protein that controls the transcription rate of genetic information from DNA to messenger RNA by binding to a specific DNA sequence. Transcription factors are proteins that bind to DNA regulatory sequences (e.g., enhancers and silencers), usually located in the 5' upstream region of the target gene to regulate the rate of gene transcription. This can lead to increased or decreased gene transcription, protein synthesis, and subsequent changes in cell function (e.g., cells change in response to the environment (normal or pathological), such as during atrophy, hypertrophy, hyperplasia, metaplasia, or dysplasia). As used herein, specific transcription factors are referred to by nomenclature, although other synonyms may also be used for the transcription factors described herein.

[0136] The term "transcription factor binding profile" generally refers to a multifactorial informative profile of a given transcription factor that includes tissue contributions and biological processes. TFBP also includes an "accessibility score" and a z-score statistic to objectively compare significant changes in TFBS accessibility in different plasma samples. This profile can allow identification of lineage-specific TFs applicable to both tissue of origin and tumor of origin identification.

[0137] As used herein, the term "subject" generally refers to an individual, entity, or medium that has or is suspected of having testable or detectable genetic information or material. The subject can be a person, an individual, or a patient. The subject can be a vertebrate, such as, for example, a mammal. Non-limiting examples of mammals include humans, apes, farm animals, sports animals, rodents, and pets. The subject can show symptoms indicating the health or physiological state or condition of the subject, such as cancer or the staging of cancer in the subject. As an alternative, the subject can be asymptomatic in terms of such health or physiological state or condition.

[0138] As used herein, the term "sample" generally refers to a biological sample obtained from or derived from one or more subjects. A biological sample can be a cell-free biological sample or a substantially cell-free biological sample, or can be processed or fractionated to produce a cell-free biological sample. For example, a cell-free biological sample can include cell-free ribonucleic acid (cfRNA), cell-free deoxyribonucleic acid (cfDNA), cell-free protein and / or cell-free polypeptide. A biological sample can be a tissue (e.g., a tissue obtained by biopsy), blood (e.g., whole blood), plasma, serum, sweat, urine, saliva, or a derivative thereof. Ethylenediaminetetraacetic acid (EDTA) collection tubes, cell-free RNA collection tubes (e.g., Streck) or cell-free DNA collection tubes (e.g., Streck) can be used to obtain or derive a cell-free biological sample from a subject. A cell-free biological sample can be obtained from a whole blood sample by fractionation. A biological sample or its derivatives can contain cells. For example, a biological sample can be a blood sample or its derivatives (e.g., blood or blood drops collected by a collection tube), a tumor sample, a tissue sample, a urine sample, or a cell (e.g., tissue) sample.

[0139] The present disclosure provides methods and systems for modeling transcription factor (TF) binding sites (TFBS) and using TFBS information to detect, evaluate, diagnose and analyze disease states. cfDNA represents a unique analyte produced by endogenous physiological processes to generate an in vivo map of nucleosome occupancy by whole genome sequencing. Ribosome occupancy at transcription factor binding sites (TFBS) can be used to infer expressed genes from cells that release their DNA into circulation. cfDNA nucleosome occupancy can reflect the footprint of TF.

[0140] I. Transcription Factor Binding Site / Nucleosome Occupancy Analysis

[0141] Although next-generation sequencing can provide important information about TFs, a non-invasive method is still needed to measure TF activity or its regulation under treatment (e.g., from blood). Cell-free DNA (cfDNA) (e.g., from plasma), which also contains circulating tumor DNA (ctDNA) in cancer patients, may provide opportunities for non-invasive diagnostic strategies for cancer patients. Since cfDNA may be released from apoptotic cells after enzymatic digestion, it may circulate mainly in the form of mononucleosomal DNA. Therefore, whole-genome sequencing of cfDNA fragments can enable the generation of nucleosome maps, in which dyads with sites with high nucleosome preference (e.g., the midpoint of a typical nucleosome) result in strong peaks in the reads, while dyads with less preferred nucleosomes show reduced peaks or no peaks at all.

[0142] Since TF binding inferred from cfDNA has great diagnostic potential in cancer and other aspects, improved and optimized bioinformatics channels have been developed. This process can parse those components involving nucleosome features at TFBS, thereby objectively evaluating and comparing TFBS accessibility in different plasma samples. Deep whole genome sequencing (WGS) data can be obtained from plasma samples from healthy donors and from plasma samples of cancer (e.g., metastatic prostate cancer, colon cancer or breast cancer) patients. In some examples, cfDNA also includes circulating tumor DNA (ctDNA). In addition, shallow WGS data can also be obtained from plasma samples from patients with the above-mentioned tumor entities. This method can be used to analyze a single TF, rather than using a mixture of cfDNA signals generated by a variety of cell types and analyzing by Fourier transform to establish a general tissue-specific pattern according to other methods. The method and system provided herein also beneficially provide a more nuanced perspective of both tissue contribution and biological process, which allows identification of lineage-specific TFs suitable for analysis of tissue of origin and tumor of origin.

[0143] Certain lineage-specific TFs are applicable to determining the tissue of origin of plasma DNA. However, to determine which TFs can be used in such applications, it is necessary to evaluate the accessibility of TFs (e.g., at their binding sites in cfDNA). Conventional methods may lack the ability to evaluate the accessibility of TFs at their binding sites in cfDNA as a representative of their activity. Calculations are performed separately for TFBSs within and outside the transcription start site (TSS). The average TFBS pattern contains two signals: TSS proximal (within about 2kb of the TSS, resulting in a "low-frequency pattern") and TSS distal (greater than 2kb from the TSS peak, resulting in a "high-frequency pattern"), corresponding to more evenly spaced peak signals. In order to suppress the effects on coverage that are not caused by preferential nucleosome positioning and to eliminate local biases from nucleosome data, a filter can be used for detrending (e.g., a Savitzky-Golay filter). The low-frequency signal obtained can then be used to standardize the high-frequency signal, and the data range of the high-frequency signal can then be recorded (the maximum value of the data value minus the minimum value, corresponding to the amplitude). Since the range of the high-frequency signal depends on the number of TFBSs (except for the 1,000-ms TFBS), these range values ​​can be corrected by smoothing (because they depend on the number of TFBSs) and then used to calculate the level as a measure of the accessibility of each TFBS.

[0144] The metric developed for this analysis, called the "accessibility score," can be used to objectively compare the accessibility of TFBSs in serial analyses of samples obtained from the same person or from different individuals. The ranking of when TF binding opens or "primes" its target enhancers, without necessarily activating them themselves, can be called the "accessibility score." These results demonstrate a reliable method for assessing TFBS accessibility, particularly for the use of cfDNA in clinical diagnostics.

[0145] In contrast to other analyses that may utilize general tissue-specific patterns using a mixture of cfDNA signals generated by multiple cell types and analyzed by Fourier transform, the methods and systems of the present disclosure can dissect individual TFs and thereby establish lineage-specific TFs for clinical applications. Due to the improved resolution of TFBS analysis, monitoring the accessibility of TFBSs from cfDNA may be possible and has been demonstrated in some examples to be useful for revealing TF plasticity during disease progression, e.g., reprogramming to different cell lineages.

[0146] Fig. 20 A diagram showing how epigenetic control regions affect nucleosome positioning is shown. Histone modifications and enhancers are illustrated.

[0147] Fig.21 Nucleosomal positioning of selected TFs is shown.

[0148] Fig. 22 The coverage pattern of the selected TF is shown.

[0149] Fig.23 The overlap of different TFs is shown. These overlap values ​​correspond to the heatmap of Figure 2C.

[0150] Fig.24 The effect of TFBS size is shown.

[0151] Fig.25 Nucleosome patterns of REST and KLF16 from samples of 24 healthy individuals are shown. Each line represents a different individual. In 24 healthy individuals, under the same settings of transcription factors active in blood cells, these patterns appear to be almost identical.

[0152] Figures 26A-26C show the nucleosome positioning of selected TFs for advanced cancer samples. The CTCF patterns in all samples looked similar. The activity of blood-specific TFs including PU.1, Lyl-1 and Spi-B was reduced in cancer samples. Cancer-specific TFs including GRH-L2 (epithelial marker), ASH-2 and HOX-B13 (prostate cancer markers of the androgen receptor axis) and EVX2 (colon cancer marker) were more active.

[0153] II. Transcription factor binding sites

[0154] Transcription factor binding sites are identified from the gene transcription regulation database (GTRD: a database on gene transcription regulation-2019 update. ISYevshin, RNSharipov.SKKolmykov, YVKondrakhin, FAKolpakov.NucleicAcids Res. 2019 January 8; 47(D1): D100-D105) using the statistical thresholds used in the methods and systems of the present invention, and are informative for machine learning model and classifier generation. In some examples, the associated pathways and categories of transcription factors are similarly useful and informative for machine learning model and classifier generation.

[0155] Statistical thresholds are used to identify differential TFs between two or more patient groups for analysis (e.g., healthy versus cancer, progressors versus non-progressors, one of multiple stages (e.g., I, II, III, or IV), one of multiple subtypes, or treatment responders versus non-responders).

[0156] In some examples, the methods and systems described herein can be used to analyze Fig. 27and those transcription factors listed in FIG. 28 .

[0157] In some examples, the transcription factor is selected from GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, and FOXA1.

[0158] In some cases, patient-specific as well as tumor-specific patterns were observed, including inferred binding patterns for the transcription factors AR, HOXB13, and NKX3-1.

[0159] In some examples, the transcription factor is an epithelial transcription factor. In some examples, the transcription factor is GRHL2.

[0160] In some examples, transcription factors GRHL2, FOXA1, and ZNF121 were associated with increased accessibility scores or open chromatin accessibility in breast cancer patients.

[0161] In some examples, open accessibility of at least one transcription factor selected from GRHL2, FOXA1, and ZNF121 is indicative of breast cancer.

[0162] In some examples, transcription factors EVX2, DLX2, HNF1A, HNF4A, GRHL2, and HNF4G were associated with increased accessibility scores or open chromatin accessibility in colon cancer patients.

[0163] In some examples, open accessibility of at least one transcription factor selected from EVX2, DLX2, HNF1A, GRHL2, HNF4A, and HNF4G is indicative of colon cancer.

[0164] In some examples, transcription factors LYL1, EVI1, TAL1, Spi-B, TBX21, and PU.1 were associated with reduced accessibility scores or closed chromatin accessibility in colon cancer patients.

[0165] In some examples, blocked accessibility of at least one transcription factor selected from LYL1, EVI1, TAL1, Spi-B, TBX21, and PU.1 is indicative of colon cancer.

[0166] In some examples, transcription factors GRHL2, FOXA1, HOXB13, AR, and NKX3-1 were associated with increased accessibility scores or open chromatin accessibility in prostate adenocarcinoma patients.

[0167] In some examples, open accessibility of at least one transcription factor selected from GRHL2, FOXA1, HOXB13, AR, and NKX3-1 is indicative of prostate adenocarcinoma.

[0168] In some examples, the transcription factors REST, GRHL2, FOXA1, HOXB13, AR, and NKX3-1 are associated with reduced or closed chromatin accessibility in patients with small cell neuroendocrine prostate cancer.

[0169] In some examples, decreased accessibility of at least one transcription factor selected from REST, GRHL2, FOXA1, HOXB13, AR, and NKX3-1 is indicative of small cell neuroendocrine prostate cancer.

[0170] In one example, a correlation between the accessibility of hematopoietic transcription factors and tissue-specific TFs is associated with the presence of a disease such as cancer.

[0171] In one example, the hematopoietic transcription factor is selected from LYL1, SCL, Bcl11a, Hhex, Lmo2, Spi1, and PU.1. In one example, the hematopoietic transcription factor is selected from LYL1 or PU.1.

[0172] In some examples, low accessibility of hematopoietic transcription factors, such as LYL1, SPIB, and EVI1 (transcription regulator ecotropic viral integration site 1), is associated with prostate cancer.

[0173] In some examples, the transcription factor is selected from GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, and FOXA1.

[0174] In some examples, the transcription factor is selected from HNF-1a, HNF-4A, HNF-4G, EVX-2, and DLX-2.

[0175] In some examples, low accessibility of hematopoiesis-related TFs, such as LYL1, TAL1 (SCL / TAL1 (stem cell leukemia / T-cell acute lymphoblastic leukemia [T-ALL] 1)), EVI1, TBX21 (T-bet), and PU.1, is associated with cancer.

[0176] During persistent exposure to antigens in chronic viral infection or cancer, effector CD8 + T cells acquire an alternative cell differentiation fate, known as T cell exhaustion. They are unable to undergo antigen-independent self-renewal like memory cells and lose their effector functions in a graded manner, which hinders the development of these antigen-specific CD8 + Viral clearance and tumor control by T cells.

[0177] In one example, open accessibility of the transcription factor TBX21 or EOMES indicates exhaustion of CD8 + T cells.

[0178] In one example, open accessibility of the transcription factors Eomesin (EOMES), Blimp-1, von Hippel-Lindau tumor suppressor (VHL), Foxo1, IRF4, BATF, and NFATcl indicated exhausted CD8 + T cells.

[0179] III. Machine Learning Systems and Models

[0180] A. Sample Characteristics

[0181] In some examples, TFBS accessibility scores are used as input features in machine learning models to find correlations between sequence composition and subject (e.g., patient) groups. Examples of such patient groups include the presence, staging, subtype, responders and non-responders, and progressors and non-progressors of a disease or condition. In some examples, a feature matrix is ​​generated to compare samples obtained from individuals with known conditions or characteristics. In some examples, samples are obtained from healthy individuals or individuals without any known indications, as well as samples from patients known to have cancer.

[0182] As used herein, when the term "feature" refers to machine learning and pattern recognition, it refers to a single measurable attribute or property of an observed phenomenon. Features are typically numeric features, but structural features such as strings and graphs can be used for syntactic pattern recognition. The concept of a "feature" is related to the concept of an explanatory variable used in statistical techniques, such as, but not limited to, linear regression. In some examples, the feature is a transcription factor binding profile. In some examples, the feature is an accessibility score calculated from a transcription factor binding profile.

[0183] In some examples, features are input into a feature matrix for machine learning analysis.

[0184] In some examples, accessibility scores for at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish between healthy subjects and cancer patients, disease progressors and non-progressors.

[0185] In some examples, accessibility scores for at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier capable of distinguishing multiple disease subtypes or multiple disease stages.

[0186] In some examples, accessibility scores for at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish between disease treatment responders and non-responders.

[0187] For multiple measurements, the system identifies a set of features to be accepted as input to the machine learning model. The system makes a measurement for each molecular class and forms a feature vector based on the measurements. The system accepts the feature vector as input to the machine learning model and generates an output classification of whether the biological sample has the specified attribute.

[0188] In some examples, the machine learning model generates a classifier that can distinguish between two or more groups or categories of individuals or features in a group of individuals or features of a group. For example, the classifier can be a binary classifier that can distinguish between two groups or categories of individuals or features in a group of individuals or features of a group. As another example, the classifier can be a multi-class classifier that can distinguish between more than two groups or categories of individuals or features in a group of individuals or features of a group. In some examples, the classifier is a trained machine learning classifier.

[0189] In some examples, the characteristics of informative loci or biomarkers in cancer tissue are determined to form a profile. In the case of a binary classifier, a receiver operating characteristic (ROC) curve can be generated to plot the performance of a specific feature (e.g., any biomarker described herein and / or any other biomedical information item) in distinguishing two populations (e.g., individuals who respond and do not respond to a therapeutic agent). In some examples, feature data in an entire population (e.g., cases and controls) is sorted in ascending order based on the value of a single feature.

[0190] In some examples, the specified attribute is selected from healthy versus cancer, disease subtype among multiple disease subtypes, disease stage among multiple disease stages, progressor versus non-progressor, responder versus non-responder, or a combination thereof.

[0191] B. Data Analysis

[0192] In some examples, the present disclosure provides a system, method or kit for realizing data analysis in software applications, computing hardware or both. In some examples, the analysis application or system includes at least a data receiving module, a data preprocessing module, a data analysis module (which can operate one or more types of genomic data), a data interpretation module or a data visualization module. In some examples, the data receiving module may include a computer system that connects laboratory hardware or instruments to a computer system that processes laboratory data. In some examples, the data preprocessing module may include a hardware system or computer software that performs operations on data in preparation for analysis. Examples of operations that can be applied to data in the preprocessing module include affine transformations, denoising operations, data cleaning, reformatting or secondary sampling. A data analysis module that can be specialized for analyzing genomic data from one or more genomic materials, for example, an assembled genomic sequence can be obtained and probability and statistical analysis can be performed to identify abnormal patterns associated with disease, pathology, state, risk, condition or phenotype. The data interpretation module can use, for example, an analytical method derived from statistics, mathematics or biology to support the understanding of the relationship between the identified abnormal pattern and health status, functional status, prognosis or risk. The data visualization module may use mathematical modeling, computer graphics, or rendering methods to create a visual representation of the data that may facilitate understanding or interpretation of the results (eg, by a user such as a subject (eg, a patient) or a physician or other healthcare provider).

[0193] In some examples, machine learning methods are applied to distinguish samples in a sample population. In some examples, machine learning methods are applied to distinguish between healthy and cancer (e.g., advanced adenoma) samples.

[0194] In some examples, one or more machine learning operations used to train the prediction engine include one or more of the following: generalized linear models, generalized additive models, nonparametric regression operations, random forest classifiers, spatial regression operations, Bayesian regression models, time series analysis, Bayesian networks, Gaussian networks, decision tree learning operations, artificial neural networks, recursive neural networks, reinforcement learning operations, linear or nonlinear regression operations, support vector machines, clustering operations, and genetic algorithm operations.

[0195] In some examples, the computer processing method is selected from logistic regression, multivariate linear regression (MLR), dimensionality reduction, partial least squares (PLS) regression, principal component regression, autoencoder, variational autoencoder, singular value decomposition, Fourier base, wavelet, discriminant analysis, support vector machine, decision tree, classification and regression tree (CART), tree-based methods, random forest, gradient boosted tree, logistic regression, matrix factorization, multidimensional scaling (MDS), dimensionality reduction methods, t-distributed stochastic neighbor embedding (t-SNE), multilayer perceptron (MLP), network clustering, neuro-fuzzy and artificial neural network.

[0196] In some examples, the methods disclosed herein may include performing computational analysis on nucleic acid sequencing data from a sample from an individual or from multiple individuals.

[0197] C. Classifier Generation

[0198] In one aspect, the present disclosure provides systems and methods including a classifier generated based on feature information obtained from sequence analysis of a biological sample from cfDNA. The classifier forms part of a prediction engine that can distinguish groups in a population based on sequence features identified in a biological sample (e.g., cfDNA).

[0199] In some examples, a classifier is created by standardizing sequence information by formatting similar portions of the sequence information into a uniform format and uniform scale; storing the standardized sequence information in a columnar database; training a prediction engine by applying one or more machine learning operations to the stored standardized sequence information, the prediction engine mapping a combination of one or more features for a specific population; applying the prediction engine to the accessed field information to identify individuals associated with the group; and classifying the individuals into a group.

[0200] The trained classifier can be configured to accept multiple input variables and generate one or more output values ​​based on the multiple input variables. Multiple input variables can include one or more data sets indicating a disease, a disorder, or an abnormal condition (e.g., cancer). For example, the input variables can include many nucleic acid sequences corresponding to or aligned with a genomic locus associated with a group of diseases. Multiple input variables can also include clinical health data of the subject.

[0201] For example, clinical health data may include one or more quantitative measures of a subject, such as age, weight, height, body mass index (BMI), blood pressure, heart rate, and glucose level. As another example, clinical health data may include one or more categorical measures, such as race, ethnicity, medication or other clinical treatment history, smoking history, drinking history, daily activity or fitness level, genetic test results, blood test results, and imaging results.

[0202] The trained algorithm provided herein can include a classifier, so that each of one or more output values ​​includes one of the possible values ​​of a fixed number of sample classification instructions classifier (e.g., linear classifier, logistic regression classifier, etc.). The trained algorithm can include a binary classifier, so that each of one or more output values ​​includes indicating that the sample is classified by the classifier in two values ​​(e.g., {0, 1}, {positive, negative} or {high risk, low risk}). The trained algorithm can be another type of classifier, so that each of one or more output values ​​includes indicating that the sample is classified by the classifier in more than two values ​​(e.g., {0, 1, 2}, {positive, negative or uncertain} or {high risk, medium risk or low risk}). The output value can include a descriptive label, a numerical value or a combination thereof. Some output values ​​can include a descriptive label. This descriptive label can provide an evaluation or indication of a disease, a disorder or an abnormal condition of a subject, and can include, for example, positive, negative, high risk, medium risk, low risk or uncertainty. This descriptive label can provide the identification of the treatment of the disease, illness or abnormal condition of the assessment of the subject, and can include, for example, the duration of therapeutic intervention, therapeutic intervention and / or the dosage suitable for the therapeutic intervention of the disease, illness or abnormal condition. This descriptive label can provide the identification of the secondary clinical test that may be suitable for the subject, and can include, for example, imaging test, blood test, computed tomography (CT) scan, magnetic resonance imaging (MRI) scan, ultrasound scan, chest X-ray, positron emission tomography (PET) scan, PET-CT scan, cytological determination or any combination thereof. For example, this descriptive label can provide the prognosis of the disease, illness or abnormal condition of the subject. As another example, this descriptive label can provide the relative assessment of the disease, illness or abnormal condition of the subject. Some descriptive labels can be mapped to numerical values, for example, by mapping "positive" to 1 and mapping "negative" to 0.

[0203] Some output values ​​may include numerical values, such as binary, integer or continuous values. Such binary output values ​​may include, for example, {0, 1}, {positive, negative} or {high risk, low risk}. Such integer output values ​​may include, for example, {0, 1, 2}. Such continuous output values ​​may include, for example, a probability value that is at least 0 and not greater than 1. Such continuous output values ​​may include, for example, an unstandardized probability value that is at least 0. Such continuous output values ​​may indicate the prognosis of a disease, illness or abnormal condition of a subject. Some numerical values ​​may be mapped to descriptive labels, for example, by mapping 1 to "positive" and 0 to "negative".

[0204] Some output values ​​can be assigned based on one or more cutoff values. For example, if the sample indicates that the subject has at least a 50% probability of suffering from a disease, disorder, or abnormal condition, the binary classification of the sample can be assigned an output value of "positive" or 1. For example, if the sample indicates that the subject has less than a 50% probability of suffering from a disease, disorder, or abnormal condition, the binary classification of the sample can be assigned an output value of "negative" or 0. In this case, a single cutoff value of 50% is used to classify the sample into one of two possible binary output values. Examples of single cutoff values ​​may include about 1%, about 2%, about 5%, about 10%, about 15%, about 20%, about 25%, about 30%, about 35%, about 40%, about 45%, about 50%, about 55%, about 60%, about 65%, about 70%, about 75%, about 80%, about 85%, about 90%, about 91%, about 92%, about 93%, about 94%, about 95%, about 96%, about 97%, about 98%, and about 99%.

[0205] As another example, if the sample indicates that the subject has a probability of at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 85%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99% or more of having a disease, disorder, or abnormal condition, then the classification of the sample may be assigned an output value of "positive" or 1. If the sample indicates that the subject has greater than about 50%, greater than about 55%, greater than about 60%, greater than about 65%, greater than about 70%, greater than about 75%, greater than about 80%, greater than about 85%, greater than about 90%, greater than about 91%, greater than about 92%, greater than about 93%, greater than about 94%, greater than about 95%, greater than about 96%, greater than about 97%, greater than about 98%, or greater than about 99% probability of having a disease, disorder, or abnormal condition, then the classification of the sample can be assigned an output value of "positive" or 1.

[0206] If the sample indicates that the subject has less than about 50%, less than about 45%, less than about 40%, less than about 35%, less than about 30%, less than about 25%, less than about 20%, less than about 15%, less than about 10%, less than about 9%, less than about 8%, less than about 7%, less than about 6%, less than about 5%, less than about 4%, less than about 3%, less than about 2%, or less than about 1% probability of having a disease, disorder, or abnormal condition, then the classification of the sample may be assigned an output value of "negative" or 0. If the sample indicates that the subject has no greater than about 50%, no greater than about 45%, no greater than about 40%, no greater than about 35%, no greater than about 30%, no greater than about 25%, no greater than about 20%, no greater than about 15%, no greater than about 10%, no greater than about 9%, no greater than about 8%, no greater than about 7%, no greater than about 6%, no greater than about 5%, no greater than about 4%, no greater than about 3%, no greater than about 2%, or no greater than about 1% probability of having a disease, disorder, or abnormal condition, then the classification of the sample may be assigned an output value of "negative" or 0.

[0207] If the sample is not classified as "positive", "negative", "1" or "0", the classification of the sample can be assigned an output value of "uncertain" or 2. In this case, a set of two cutoff values ​​will be used to classify the sample into one of three possible output values. Examples of cutoff value sets may include {1%, 99%}, {2%, 98%}, {5%, 95%}, {10%, 90%}, {15%, 85%}, {20%, 80%}, {25%, 75%}, {30%, 70%}, {35%, 65%}, {40%, 60%}, and {45%, 55%}. Similarly, a set of n cutoff values ​​may be used to classify the sample into one of n+1 possible output values, where n is any positive integer.

[0208] A trained classifier can be trained with multiple independent training samples. Each independent training sample can include a sample from a subject, a related data set (as described elsewhere herein) obtained by measuring the sample, and one or more known output values ​​corresponding to the sample (e.g., a clinical diagnosis, prognosis, absence or therapeutic effect of a disease, disorder or abnormal condition of the subject). Independent training samples can include samples and related data sets and outputs obtained from or derived from multiple different subjects. Independent training samples can include samples and related data sets and outputs obtained from the same subject at multiple different time points (e.g., regularly, such as weekly, biweekly or monthly). Independent training samples can be related to the presence of a disease, disorder or abnormal condition (e.g., training samples include samples and related data sets and outputs obtained from or derived from multiple subjects known to have a disease, disorder or abnormal condition). Independent training samples can be related to the absence of a disease, disorder or abnormal condition (e.g., training samples include samples and related data sets and outputs obtained from or derived from multiple subjects known to have no previous diagnosis of a disease, disorder or abnormal condition or have received a negative test result for a disease, disorder or abnormal condition).

[0209] The trained classifier can be trained with at least about 5, at least about 10, at least about 15, at least about 20, at least about 25, at least about 30, at least about 35, at least about 40, at least about 45, at least about 50, at least about 100, at least about 150, at least about 200, at least about 250, at least about 300, at least about 350, at least about 400, at least about 450, or at least about 500 independent training samples. The independent training samples can include samples associated with the presence of a disease, disorder, or abnormal condition and / or samples associated with the absence of a disease, disorder, or abnormal condition. The trained classifier can be trained with no more than about 500, no more than about 450, no more than about 400, no more than about 350, no more than about 300, no more than about 250, no more than about 200, no more than about 150, no more than about 100, or no more than about 50 independent training samples associated with the presence of a disease, disorder, or abnormal condition. In some embodiments, the samples are independent of the samples used to train the trained classifier.

[0210] The trained classifier can be trained with a first number of independent training samples associated with the presence of a disease, disorder, or abnormal condition and a second number of independent training samples associated with the absence of a disease, disorder, or abnormal condition. The first number of independent training samples associated with the presence of a disease, disorder, or abnormal condition may be no more than the second number of independent training samples associated with the absence of a disease, disorder, or abnormal condition. The first number of independent training samples associated with the presence of a disease, disorder, or abnormal condition may be as many as the second number of independent training samples associated with the absence of a disease, disorder, or abnormal condition. The first number of independent training samples associated with the presence of a disease, disorder, or abnormal condition may be more than the second number of independent training samples associated with the absence of a disease, disorder, or abnormal condition.

[0211] The trained classifier can be configured to detect the presence of a single molecule of at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 100%, at least about 101%, at least about 102%, at least about 103%, at least about 104%, at least about 105%, at least about 106%, at least about 107%, at least about 108%, at least about 109%, at least about 110%, at least about 111%, at least about 112%, at least about 113%, at least about 114%, at least about 115%, at least about 116%, at least about 117%, at least about 118%, at least about 119%, 8%, at least about 99% or more accuracy in identifying the presence or absence of a disease, disorder or abnormal condition; for at least about 5, at least about 10, at least about 15, at least about 20, at least about 25, at least about 30, at least about 35, at least about 40, at least about 45, at least about 50, at least about 100, at least about 150, at least about 200, at least about 250, at least about 300, at least about 350, at least about 400, at least about 450, or at least about 500 independent training samples. The accuracy of identifying the presence or absence of a disease, disorder or abnormal condition by a trained algorithm can be calculated as the percentage of independent test samples (e.g., subjects known to have the disease, disorder or abnormal condition, or subjects with a negative clinical test result for the disease, disorder or abnormal condition) that are correctly identified or classified as having or not having the disease, disorder or abnormal condition.

[0212] The trained classifier can be configured to identify the presence of a disease, disorder, or abnormal condition with a positive predictive value (PPV) of at least about 5%, at least about 10%, at least about 15%, at least about 20%, at least about 25%, at least about 30%, at least about 35%, at least about 40%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, or more. The PPV for identifying the presence of a disease, disorder, or abnormal condition using a trained classifier can be calculated as the percentage of samples identified or classified as having the disease, disorder, or abnormal condition that correspond to subjects who actually have the disease, disorder, or abnormal condition.

[0213] The trained classifier can be configured to identify the absence of a disease, disorder, or abnormal condition with a negative predictive value (NPV) of at least about 5%, at least about 10%, at least about 15%, at least about 20%, at least about 25%, at least about 30%, at least about 35%, at least about 40%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, or more. The NPV for identifying a disease, disorder, or abnormal condition using a trained classifier can be calculated as the percentage of samples identified or classified as not having the disease, disorder, or abnormal condition that corresponds to subjects who truly do not have the disease, disorder, or abnormal condition.

[0214] The trained classifier can be configured to detect the presence of a single antibody in a given sample of at least about 5%, at least about 10%, at least about 15%, at least about 20%, at least about 25%, at least about 30%, at least about 35%, at least about 40%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 100%, at least about 101%, at least about 102%, at least about 103%, at least about 104%, at least about 105%, at least about 106%, at least about 107%, at least about 108%, at least about 109%, at least about 110%, at least about 111%, at least about 112%, at least about 113%, at least about 114%, at least about 115%, The clinical specificity of identifying a disease, disorder or abnormal condition is at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 99.1%, at least about 99.2%, at least about 99.3%, at least about 99.4%, at least about 99.5%, at least about 99.6%, at least about 99.7%, at least about 99.8%, at least about 99.9%, at least about 99.99%, at least about 99.999% or more. As used herein, specificity refers to "the probability of testing negative among people who do not suffer from the disease". It is equal to the number of disease-free people who test negative divided by the total number of disease-free individuals. The clinical specificity of identifying the absence of a disease, disorder or abnormal condition using a trained classifier can be calculated as the percentage of independent test samples (e.g., subjects whose clinical test results for a disease, disorder or abnormal condition are negative) that are correctly identified or classified as not suffering from the disease, disorder or abnormal condition. In some examples, the model, classifier or predictive test has at least about 40%, at least about 45%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 85%, at least about 90%, at least about 95% or at least about 99% specificity.

[0215] The trained classifier can be configured to detect the presence of a single antibody in a given sample of at least about 5%, at least about 10%, at least about 15%, at least about 20%, at least about 25%, at least about 30%, at least about 35%, at least about 40%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 100%, at least about 101%, at least about 102%, at least about 103%, at least about 104%, at least about 105%, at least about 106%, at least about 107%, at least about 108%, at least about 109%, at least about 110%, at least about 111%, at least about 112%, at least about 113%, at least about 114%, at least about 115%, The clinical sensitivity of at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 99.1%, at least about 99.2%, at least about 99.3%, at least about 99.4%, at least about 99.5%, at least about 99.6%, at least about 99.7%, at least about 99.8%, at least about 99.9%, at least about 99.99%, at least about 99.999% or more identifies the presence of a disease, disorder or abnormal condition. As used herein, sensitivity refers to the "probability of testing positive among people suffering from the disease". It is equal to the number of sick individuals who test positive divided by the total number of sick individuals.

[0216] In some examples, the model, classifier or predictive test has a sensitivity of at least about 40%, at least about 45%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 85%, at least about 90%, at least about 95%, or at least about 99%. The clinical sensitivity of using a trained classifier to identify the presence of a disease, disorder, or abnormal condition can be calculated as the percentage of independent test samples (e.g., subjects known to have the disease, disorder, or abnormal condition) associated with the presence of the disease, disorder, or abnormal condition that are correctly identified or classified as having the disease, disorder, or abnormal condition.

[0217] The trained classifier can be configured to identify the presence or absence of a disease, disorder, or abnormal condition with an area under the receiver operating characteristic (AUROC) of at least about 0.50, at least about 0.55, at least about 0.60, at least about 0.65, at least about 0.70, at least about 0.75, at least about 0.80, at least about 0.81, at least about 0.82, at least about 0.83, at least about 0.84, at least about 0.85, at least about 0.86, at least about 0.87, at least about 0.88, at least about 0.89, at least about 0.90, at least about 0.91, at least about 0.92, at least about 0.93, at least about 0.94, at least about 0.95, at least about 0.96, at least about 0.97, at least about 0.98, at least about 0.99, or more. AUROC can be calculated as the integral of a receiver operating characteristic (ROC) curve associated with a trained classifier in classifying a sample as having or not having a disease, disorder, or abnormal condition (e.g., the area under the ROC curve, or AUC).

[0218] The trained classifier can be adjusted or tuned to improve one or more of the performance, accuracy, PPV, NPV, clinical sensitivity, clinical specificity, or AUC for identifying a disease, disorder, or abnormal condition. The trained classifier can be adjusted or tuned by adjusting the parameters of the trained classifier (e.g., a set of cutoff values ​​for classifying samples as described elsewhere herein, or the weights of a neural network). The trained classifier can be continuously adjusted or tuned during or after the training process.

[0219] After the trained classifier is preliminarily trained, the most influential or important subset of input can be identified to be included in the high-quality classification. For example, the most influential or important subset of multiple input variables can be identified to be included in the high-quality classification or identification for the assessment of a disease, illness or abnormal condition. Multiple input variables or their subsets can be graded based on a classification index, which indicates the impact or importance of each input variable on the assessment of a disease, illness or abnormal condition being classified or identified. This index can be used to (significantly in some cases) reduce the number of input variables (e.g., predictor variables) that can be used to train a trained classifier to a desired performance level (e.g., based on desired minimum accuracy, PPV, NPV, clinical sensitivity, clinical specificity, AUC or its combination). For example, if training a trained classifier with a plurality of input variables comprising tens or hundreds of input variables produces a classification accuracy exceeding 99%, then training the trained classifier with only a selected subset of no more than about 5, no more than about 10, no more than about 15, no more than about 20, no more than about 25, no more than about 30, no more than about 35, no more than about 40, no more than about 45, no more than about 50, or no more than about 100 of the most influential or important input variables among the plurality can result in a reduced but still acceptable accuracy. The invention also provides a degree of classification accuracy (e.g., at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, or at least about 99%). The subset can be selected by ranking all of the multiple input variables and selecting a predetermined number (e.g., no more than about 5, no more than about 10, no more than about 15, no more than about 20, no more than about 25, no more than about 30, no more than about 35, no more than about 40, no more than about 45, no more than about 50, or no more than about 100) of input variables with the best classification indicators.

[0220] D. Digital processing device

[0221] In some examples, the subject matter described herein may include a digital processing device or its use. In some examples, the digital processing device may include one or more hardware central processing units (CPUs), graphics processing units (GPUs), or tensor processing units (TPUs) that perform the functions of the device. In some examples, the digital processing device may include an operating system configured to execute executable instructions. In some examples, the digital processing device may be connected to a computer network. In some examples, the digital processing device may be connected to the Internet. In some examples, the digital processing device may be connected to a cloud computing infrastructure. In some examples, the digital processing device may be connected to an intranet. In some examples, the digital processing device may be connected to a data storage device.

[0222] Non-limiting examples of suitable digital processing devices include server computers, desktop computers, laptop computers, notebook computers, subnotebook computers, netbook computers, netpad computers, set-top computers, handheld computers, Internet appliances, mobile smart phones and tablet computers. Suitable tablet computers may include, for example, those computers having brochure, tablet and convertible configurations.

[0223] In some examples, the digital processing device may include an operating system configured to execute executable instructions. For example, an operating system may include software (including programs and data) that manages the hardware of the device and provides services for executing application programs. Non-limiting examples of operating systems include Ubuntu, FreeBSD, OpenBSD, Linux, Mac OS X Windows and Non-limiting examples of suitable personal computer operating systems include Mac OS and UNIX-like operating systems (such as ). In some examples, the operating system may be provided by cloud computing, and the cloud computing resources may be provided by one or more service providers.

[0224] In some examples, the device may include a storage and / or memory device. The storage and / or memory device may be one or more physical devices for temporarily or permanently storing data or programs. In some examples, the device may be a volatile memory and require power to maintain the stored information. In some examples, the device may be a non-volatile memory and retain the stored information when the digital processing device is not powered. In some examples, the non-volatile memory may include flash memory. In some examples, the non-volatile memory may include dynamic random access memory (DRAM). In some examples, the non-volatile memory may include ferroelectric random access memory (FRAM). In some examples, the non-volatile memory may include phase change random access memory (PRAM). In some examples, the device may be a storage device, including, for example, a CD-ROM, a DVD, a flash memory device, a disk drive, a tape drive, an optical drive, and a cloud computing-based storage device. In some examples, the storage and / or memory device may be a combination of those devices as disclosed herein. In some examples, the digital processing device may include a display for sending visual information to a user. In some examples, the display may be a cathode ray tube (CRT). In some examples, the display may be a liquid crystal display (LCD). In some examples, the display may be a thin film transistor liquid crystal display (TFT-LCD). In some examples, the display may be an organic light emitting diode (OLED) display. In some examples, the OLED display may be a passive matrix OLED (PMOLED) or an active matrix OLED (AMOLED) display. In some examples, the display may be a plasma display. In some examples, the display may be a video projector. In some examples, the display may be a combination of those devices as disclosed herein.

[0225] In some examples, the digital processing device may include an input device to receive information from a user. In some examples, the input device may be a keyboard. In some examples, the input device may be a pointing device, including, for example, a mouse, a trackball, a trackpad, a joystick, a game controller, or a stylus. In some examples, the input device may be a touch screen or a multi-touch screen. In some examples, the input device may be a microphone for capturing voice or other sound input. In some examples, the input device may be a camera for capturing motion or visual input. In some examples, the input device may be a combination of those devices as disclosed herein.

[0226] E. Non-transitory computer-readable storage medium

[0227] In some examples, the subject matter disclosed herein may include one or more non-transitory computer-readable storage media encoded with a program, the program including instructions executable by an operating system. The operating system may be part of a networked digital processing device. In some examples, the computer-readable storage medium may be a tangible component of the digital processing device. In some examples, the computer-readable storage medium may be removed from the digital processing device. In some examples, the computer-readable storage medium may include, for example, a CD-ROM, a DVD, a flash memory device, a solid-state memory, a disk drive, a tape drive, an optical drive, a cloud computing system and service, etc. In some examples, the program and instructions may be permanently, substantially permanently, semi-permanently, or non-temporarily encoded on the medium.

[0228] F.Computer System

[0229] The present disclosure provides computer systems programmed to implement the methods described herein. Figure 1 A computer system 101 is shown, which is programmed or otherwise configured to perform methods of the present disclosure, such as storing, processing, identifying or interpreting subject (e.g., patient) data, biological data, biological sequences, reference sequences, TFBS data or TFBS features, such as z scores or TFBS accessibility scores. Computer system 101 can process various aspects of subject (e.g., patient) data, biological data, biological sequences or reference sequences of the present disclosure. Computer system 101 can be an electronic device of a user or a computer system remotely located relative to an electronic device. The electronic device can be a mobile electronic device.

[0230] Computer system 101 includes a central processing unit (CPU, also referred to herein as a "processor" and "computer processor") 105, which may be a single-core or multi-core processor, or multiple processors for parallel processing. Computer system 101 also includes memory or memory location 110 (e.g., random access memory, read-only memory, flash memory), an electronic storage unit 115 (e.g., a hard disk), a communication interface 120 (e.g., a network adapter) for communicating with one or more other systems, and peripherals 125, such as cache, other memory, data storage, and / or an electronic display adapter. Memory 110, storage unit 115, interface 120, and peripherals 125 communicate with CPU 105 via a communication bus (solid lines) such as a motherboard. Storage unit 115 may be a data storage unit (or data repository) for storing data. Computer system 101 may be operatively coupled to a computer network ("network") 130 by means of communication interface 120. Network 130 may be the Internet, an internetwork, and / or an extranet, or an intranet and / or an extranet in communication with the Internet. In some examples, network 130 is a telecommunications and / or data network. Network 130 may include one or more computer servers that may implement distributed computing, such as cloud computing. In some examples, with the aid of computer system 101, network 130 may implement a peer-to-peer network that may enable devices coupled to computer system 101 to behave as clients or servers.

[0231] The CPU 105 may execute a series of machine-readable instructions, which may be embodied in a program or software. The instructions may be stored in a memory location, such as the memory 110. The instructions may be directed to the CPU 105, which may then program or otherwise configure the CPU 105 to implement the methods of the present disclosure. Examples of operations performed by the CPU 105 may include fetching, decoding, executing, and writing back.

[0232] CPU 105 may be part of a circuit (eg, an integrated circuit). One or more other components of system 101 may be included in the circuit. In some examples, the circuit is an application specific integrated circuit (ASIC).

[0233] Storage unit 115 may store files, such as drivers, libraries, and saved programs. Storage unit 115 may store user data, such as user preferences and user programs. In some examples, computer system 101 may include one or more additional data storage units external to computer system 101, such as located on a remote server that communicates with computer system 101 via an intranet or the Internet.

[0234] Computer system 101 may communicate with one or more remote computer systems via network 130. For example, computer system 101 may communicate with a user's remote computer system. Examples of remote computer systems include personal computers (e.g., portable PCs), tablet computers or tablet computers (e.g., iPad, Galaxy Tab), phones, smartphones (e.g. iPhone, Android-supported devices, ) or a personal digital assistant. Users can access the computer system 101 through the network 130.

[0235] The methods described herein may be implemented by machine (e.g., computer processor) executable code stored in an electronic storage location of computer system 101, such as memory 110 or electronic storage unit 115. Machine executable or machine readable code may be provided in the form of software. During use, the code may be executed by processor 105. In some examples, the code may be retrieved from storage unit 115 and stored in memory 110 for access by processor 105 at any time. In some examples, electronic storage unit 115 may not be included, and machine executable instructions may be stored in memory 110.

[0236] The code may be precompiled and configured for use with a machine having a processor suitable for executing the code, or may be interpreted or compiled during runtime. The code may be provided in a programming language that may be selected to enable the code to be executed in a precompiled, interpreted, or just-in-time compiled manner.

[0237] Aspects of the systems and methods provided herein, such as computer system 101, can be implemented in programming. Various aspects of technology can be considered as "products" or "articles", usually in the form of machine (or processor) executable code and / or associated data carried on or implemented in the type of machine-readable medium. Machine executable code can be stored in an electronic storage unit, such as a memory (e.g., read-only memory, random access memory, flash memory) or a hard disk. "Storage" type media can include any or all tangible memories of a computer, processor, etc., or its related modules, such as various semiconductor memories, tape drives, disk drives, etc. that can provide non-temporary storage for software programming at any time. All or part of the software can sometimes communicate through the Internet or other various telecommunications networks. For example, such communication can enable software to be loaded from one computer or processor to another computer or processor, for example, from a management server or host to a computer platform of an application server. Therefore, another type of medium that can carry software elements includes light waves, radio waves and electromagnetic waves, such as light waves, radio waves and electromagnetic waves used between physical interfaces between local devices through wired and optical landline networks and through various air links. Physical elements that carry such waves, such as wired or wireless links, optical links, etc., can also be considered as media that carry the software. As used herein, unless limited to non-transitory, tangible "storage" media, terms such as computer or machine "readable media" refer to any medium that participates in providing instructions to a processor for execution.

[0238] Thus, machine-readable media, such as computer executable code, may take a variety of forms, including but not limited to tangible storage media, carrier media, or physical transmission media. Non-volatile storage media include, for example, optical or magnetic disks, such as any storage device in any computer, etc., which may be used, for example, to implement the databases shown in the accompanying drawings, etc. Volatile storage media include dynamic memory, such as the main memory of such computer platforms. Tangible transmission media include coaxial cables; copper wire and optical fiber, including the wires that make up the bus within a computer system. Carrier transmission media may take the form of electrical or electromagnetic signals, as well as acoustic or light waves, such as those generated during radio frequency (RF) and infrared (IR) data communications. Thus, common forms of computer readable media include, for example: a floppy disk, a flexible disk, a hard disk, a magnetic tape, any other magnetic medium, a CD-ROM, a DVD or DVD-ROM, any other optical medium, a punched card tape, any other physical storage medium with a pattern of holes, a RAM, a ROM, a PROM and an EPROM, a FLASH-EPROM, any other memory chip or cartridge, a carrier wave that transmits data or instructions, a cable or link that transmits such a carrier wave, or any other medium from which a computer can read programming code and / or data. Many of these forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to a processor for execution.

[0239] The computer system 101 may include or communicate with an electronic display 135 that includes a user interface (UI) 140 for providing, for example, nucleic acid sequences, enriched nucleic acid samples, transcription factor binding profiles, accessibility scores, expression profiles, and expression profile analysis. Examples of UIs include, but are not limited to, graphical user interfaces (GUIs) and web-based user interfaces.

[0240] The methods and systems of the present disclosure may be implemented by one or more algorithms. The algorithms may be implemented by software when executed by the central processing unit 105. For example, the algorithms may detect multiple regulatory elements, sequence nucleic acid samples, enrich nucleic acid samples, determine expression profiles of nucleic acid samples, analyze expression profiles of nucleic acid samples, and archive or disseminate results of expression profile analysis.

[0241] In some examples, the subject matter disclosed herein may include at least one computer program or the purpose of the computer program. The computer program may execute a series of instructions that may be executed in a CPU, GPU, or TPU of a digital processing device and written to perform a specified task. Computer-readable instructions may be implemented as program modules that perform specific tasks or implement specific abstract data types, such as functions, objects, application programming interfaces (APIs), data structures, etc. For example, computer programs may be written in various versions of various languages.

[0242] The functionality of the computer-readable instructions may be combined or distributed in various environments as desired. In some examples, the computer program may include one instruction sequence. In some examples, the computer program may include multiple instruction sequences. In some examples, the computer program may be provided from one location. In some examples, the computer program may be provided from multiple locations. In some examples, the computer program may include one or more software modules. In some examples, the computer program may include, in part or in whole, one or more web applications, one or more mobile applications, one or more stand-alone applications, one or more web browser plug-ins, extensions, add-ons, or add-ons, or a combination thereof.

[0243] In some examples, the computer processing can be a method of statistics, mathematics, biology, or any combination thereof. In some examples, the computer processing method includes a dimensionality reduction method, including, for example, logistic regression, dimensionality reduction, principal component analysis, autoencoders, singular value decomposition, Fourier basis, singular value decomposition, wavelets, discriminant analysis, support vector machines, tree-based methods, random forests, gradient boosted trees, logistic regression, matrix decomposition, network clustering, and neural networks.

[0244] In some examples, the computer processing methods are supervised machine learning methods, including, for example, regression, support vector machines, tree-based methods, and networks.

[0245] In some examples, the computer processing methods are unsupervised machine learning methods, including, for example, clustering, networks, principal component analysis, and matrix factorization.

[0246] G. Database

[0247] In some examples, the subject matter disclosed herein may include one or more databases, or the one or more databases are used to store subject (e.g., patient) data, biological data, biological sequences, or reference sequences. Reference sequences can be obtained from a database. For example, many databases can be suitable for storing and retrieving sequence information. In some examples, suitable databases may include, for example, relational databases, non-relational databases, object-oriented databases, object databases, entity-relationship model databases, associated databases, and XML databases. In some examples, the database may be based on the Internet. In some examples, the database may be based on the web. In some examples, the database may be based on cloud computing. In some examples, the database may be based on one or more local computer storage devices.

[0248] 676 TFs from the Gene Transcription Regulation Database (GTRD; Version 18.01) were used; because they contain detailed TFBS information based on ChIP-seq data from various tissue samples. TFs were annotated with the latest catalog list of 1,639 known or possible human TFs (Figure 6A). Due to the potential high number of TFBSs that TFs bind to at variable frequencies, three different stringency criteria were defined (Figure 6A): first, all TFBSs of all tissue samples in GTRD; second, those peaks supported by more than 50% of the maximum number of samples (subsequently referred to as ">50%-TFBS"; in both analyses, all 676 GTRD TFs were included); third, 1,000 TFBS / TFs supported by most samples ("1,000-msTFBS"; 505 TFs met this criterion).

[0249] In some examples, the reference genome is selected from GrCH38, GrCH37, hg19, or hg38.

[0250] In some examples, a reference genome database is used for the alignment and mapping steps of the methods disclosed herein.

[0251] IV. How to use

[0252] A. Diagnostic and Subject Characterization Methods and Systems

[0253] The methods and systems provided herein can use artificial intelligence-based methods to perform predictive analysis to analyze TFBS data obtained from a subject (e.g., a patient) to generate an output of an assessment (e.g., diagnosis, prognosis, treatment selection, treatment monitoring, staging, or subtype) of a subject with cancer (e.g., colorectal cancer, breast cancer, prostate cancer). For example, the application can apply a predictive algorithm to the acquired TFBS data to generate an assessment (e.g., diagnosis, prognosis, treatment selection, treatment monitoring, staging, or subtype) of a subject with cancer. The predictive algorithm can include an artificial intelligence-based predictor, such as a machine learning-based model, which is configured to process the acquired TFBS data to generate an assessment (e.g., diagnosis, prognosis, treatment selection, treatment monitoring, staging, or subtype) of a subject with cancer.

[0254] The machine learning predictor can be trained using a dataset (e.g., a dataset generated by performing TFBS analysis on individual biological samples from one or more cohorts of patients suffering from cancer) as input and using the known diagnosis (e.g., stage and / or tumor grade, subtype, treatment responders versus non-responders, progressors versus non-progressors) results of the subjects as outputs of the machine learning predictor.

[0255] A training data set (e.g., a data set generated by performing a multi-analyte assay on an individual's biological sample) can be generated from, for example, one or more groups of subjects having common features (features) and results (labels). The training data set may include a set of features and labels corresponding to features relevant to diagnosis. Features may include characteristics, such as certain ranges or categories measured by a cfDNA assay, such as z scores, accessibility scores, etc. For example, a set of features collected from a given subject at a given time point can collectively serve as a diagnostic marker, which can indicate that a cancer has been identified in the subject at a given time point. Features may also include labels indicating a diagnostic result for a subject, such as for one or more cancers.

[0256] The label can include results, such as a known diagnosis of a subject (e.g., stage, subtype, tumor grade, or progressor versus non-progressor). The results can include characteristics associated with a subject's cancer. For example, a characteristic can indicate that a subject has one or more cancers.

[0257] A training set (e.g., a training data set) can be selected by random sampling of a set of data corresponding to one or more groups of subjects (e.g., retrospective and / or prospective cohorts of subjects (e.g., patients) with or without one or more cancers. Alternatively, a training set (e.g., a training data set) can be selected by proportional sampling of a set of data corresponding to one or more groups of subjects (e.g., retrospective and / or prospective cohorts of subjects (e.g., patients) with or without one or more cancers. The training set can be balanced between data sets corresponding to one or more groups of subjects (e.g., patients from different clinical sites or trials). A machine learning predictor can be trained until certain predetermined conditions for accuracy or performance are met, such as having a minimum expected value corresponding to a diagnostic accuracy measure. For example, a diagnostic accuracy measure can correspond to a prediction of a diagnosis, stage, or subtype of one or more cancers in a subject.

[0258] Examples of diagnostic accuracy metrics may include sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and the area under the curve (AUC) of a receiver operating characteristic (ROC) curve corresponding to the diagnostic accuracy of detecting or predicting cancer (e.g., colorectal cancer).

[0259] In one aspect, the present disclosure provides a computer-implemented method for detecting the presence or absence of a disease in a subject or diagnosing a disease in a subject, the method comprising: (a) providing a set of sequence reads of deoxyribonucleic acid (DNA) extracted from a subject; (b) using the set of sequence reads to generate a coverage pattern of a transcription factor; (c) processing the coverage pattern to provide a signal; (d) processing the signal with a reference signal, wherein the signal and the reference signal have different frequencies, thereby detecting the presence or absence of a disease in a subject or diagnosing a disease in a subject.

[0260] In some examples, the DNA is cell-free DNA.

[0261] In some examples, the disease is cancer.

[0262] In some examples, (b) includes aligning the sequence reads to a reference sequence to provide an aligned sequence pattern, selecting regions of the aligned sequence pattern corresponding to binding sites of the transcription factor, and normalizing the aligned sequence pattern in these regions.

[0263] In some examples, (c) includes using a low pass filter. In some examples, (c) includes using a Savitzky-Golay filter.

[0264] In some examples, the subject is a human.

[0265] In some examples, the transcription factor is a cancer-specific transcription factor.

[0266] In some examples, accessibility scores for at least 2, or at least 5, or at least 10, or at least 15, or at least 20, or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish healthy subjects from cancer patients.

[0267] In some examples, accessibility scores for at least 2, or at least 5, or at least 10, or at least 15, or at least 20, or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish disease progressors from non-progressors, disease subtypes among multiple disease subtypes, disease stages among multiple disease stages, or any combination thereof.

[0268] In some examples, accessibility scores for at least 2, or at least 5, or at least 10, or at least 15, or at least 20, or at least 25 transcription factor binding sites are determined and input into a machine learning model to train a classifier that can distinguish between disease treatment responders and non-responders.

[0269] On the one hand, the methods described herein allow patients to be classified by tumor type, including, for example, tumor subtypes (e.g., subtypes of prostate cancer, colorectal cancer, breast cancer, lung cancer) that may have important clinical significance for patient management (including treatment planning and responsiveness). Therefore, the methods provided herein for locating tumor-specific transcription factor binding in vivo based on patient samples (e.g., blood, plasma or serum samples) make key parts of the non-coding genome suitable for clinical analysis.

[0270] In some examples, the method distinguishes subtypes of the disease.

[0271] In some examples, the method distinguishes subtypes of cancer.

[0272] In some examples, the method distinguishes subtypes of prostate cancer, colorectal cancer, breast cancer, and lung cancer.

[0273] In some examples, the method distinguishes patients with prostate cancer subtypes as having prostate adenocarcinoma or small cell neuroendocrine prostate cancer.

[0274] In another aspect, the present disclosure provides a computer-implemented method for monitoring progression or regression of a disease in a subject, the method comprising: (a) providing a first set of sequence reads of deoxyribonucleic acid (DNA) extracted from the subject at a first time and a second set of sequence reads of the DNA extracted from the subject at a second time, the second time being later than the first time; (b) generating a first coverage pattern of a transcription factor using the first set of sequence reads, and generating a second coverage pattern of the transcription factor using the second set of sequence reads; (c) processing the first coverage pattern to provide a first signal, and processing the second coverage pattern to provide a second signal; (d) processing the first signal with a reference signal, wherein the first signal and the reference signal have different frequencies; (e) processing the second signal with the reference signal, wherein the second signal and the reference signal have different frequencies; and (f) monitoring progression or regression of the disease in the subject based on processing the first signal and the second signal with the reference signal.

[0275] In some examples, the DNA is cell-free DNA.

[0276] In some examples, the disease is cancer.

[0277] In some examples, (b) includes aligning the first group of sequence reads and the second group of sequence reads with a reference sequence to provide a first aligned sequence pattern and a second aligned sequence pattern, respectively, selecting a region of the first aligned sequence pattern and the second aligned sequence pattern corresponding to a binding site of a transcription factor, and normalizing the first aligned sequence pattern and the second aligned sequence pattern in the region.

[0278] In some examples, (c) includes using a low pass filter. In some examples, (c) includes using a Savitzky-Golay filter.

[0279] In some examples, the subject is a human.

[0280] In some examples, the transcription factor is a cancer-specific transcription factor.

[0281] In another aspect, the present disclosure provides a system for detecting or diagnosing a disease in a subject, comprising a processor configured to: (i) use sequence reads of deoxyribonucleic acid (DNA) extracted from the subject to generate a coverage pattern of a transcription factor; (ii) process the coverage pattern to provide a signal, wherein the signal has a frequency different from a reference signal; and (iii) provide detection or diagnosis of the disease for the subject based on the signal.

[0282] In another aspect, the present disclosure provides a system for monitoring progression or regression of a disease during or after a course of treatment for a subject, comprising a processor configured to: (i) use a first set of sequence reads of deoxyribonucleic acid (DNA) extracted from the subject at a first time during or after treatment, and a second set of sequence reads of DNA extracted from the subject at a second time later than the first time to generate a first coverage pattern of transcription factors corresponding to the first set of sequence reads and a second coverage pattern of transcription factors corresponding to the second set of sequence reads; (ii) process the first coverage pattern to provide a first signal, and process the second coverage pattern to provide a second signal, wherein the first signal and the second signal have a different frequency from a reference signal; and (iii) monitor progression or regression of the disease during or after a course of treatment for the subject based on processing the first signal and the second signal using the reference signal.

[0283] In another aspect, the present disclosure provides a system for detecting or diagnosing a disease in a subject, comprising a processor configured to: (i) use sequence reads of deoxyribonucleic acid (DNA) extracted from the subject to generate a coverage pattern of a transcription factor; (ii) process the coverage pattern to provide a signal, wherein the signal has a frequency different from a reference signal; and (iii) provide detection or diagnosis of the disease for the subject based on the signal.

[0284] In some embodiments, a trained classifier can determine that a subject is at risk for a disease, disorder, or abnormal condition (e.g., cancer) by at least about 5%, at least about 10%, at least about 15%, at least about 20%, at least about 25%, at least about 30%, at least about 35%, at least about 40%, at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, or more.

[0285] The trained classifier can be at least about 50%, at least about 55%, at least about 60%, at least about 65%, at least about 70%, at least about 75%, at least about 80%, at least about 81%, at least about 82%, at least about 83%, at least about 84%, at least about 85%, at least about 86%, at least about 87%, at least about 88%, at least about 89%, at least about 90%, at least about 91%, at least about 92%, at least about 93%, at least about 94%, at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 100%, at least about 101%, at least about 102%, at least about 103%, at least about 104%, at least about 105%, at least about 106%, at least about 107%, at least about 108%, at least about 109%, at least about 110%, at least about 111%, at least about 112%, at least about 113%, at least about 114%, at least about 115%, at least about 116%, at least about 117%, at least about 118%, at least about 119%, The invention can determine that a subject is at risk for a disease, disorder or abnormal condition with an accuracy of at least about 95%, at least about 96%, at least about 97%, at least about 98%, at least about 99%, at least about 99.1%, at least about 99.2%, at least about 99.3%, at least about 99.4%, at least about 99.5%, at least about 99.6%, at least about 99.7%, at least about 99.8%, at least about 99.9%, at least about 99.99%, at least about 99.999% or more.

[0286] After the subject is identified as having a disease, disorder, or abnormal condition, a therapeutic intervention may be provided to the subject (e.g., an appropriate course of treatment is prescribed to treat the subject's disease, disorder, or abnormal condition). The therapeutic intervention may include prescribing an effective dose of a drug, further testing or evaluation of the disease, disorder, or abnormal condition, further monitoring of the disease, disorder, or abnormal condition, or a combination thereof. If the subject is currently being treated for a disease, disorder, or abnormal condition by a course of treatment, the therapeutic intervention may include a subsequent different course of treatment (e.g., to improve the effectiveness of the treatment because the current course of treatment is ineffective).

[0287] Therapeutic intervention may include recommending that the subject undergo secondary clinical testing to confirm the diagnosis or other evaluation of the disease, disorder, or abnormal condition. The secondary clinical testing may include imaging tests, blood tests, computed tomography (CT) scans, magnetic resonance imaging (MRI) scans, ultrasound scans, chest x-rays, positron emission tomography (PET) scans, PET-CT scans, cytology assays, or any combination thereof.

[0288] Multiple input variables (e.g., TFBS information) can be evaluated over a period of time to monitor patients (e.g., subjects suffering from a disease, disorder, or abnormal condition or being treated for a disease, disorder, or abnormal condition). In this case, the input variables (e.g., TFBS information) of the patient sample can change during the course of treatment. For example, the TFBS information of a patient whose risk of a disease, disorder, or abnormal condition is reduced due to effective treatment can be shifted toward the spectrum or distribution of a healthy subject (e.g., a subject without a disease, disorder, or abnormal condition). Conversely, for example, the TFBS information of a patient whose risk of a disease, disorder, or abnormal condition is increased due to ineffective treatment can be shifted toward the spectrum or distribution of a subject with a higher risk of a disease, disorder, or abnormal condition, or a more advanced state of a disease, disorder, or abnormal condition.

[0289] The disease, disorder or abnormal condition of the subject can be monitored by monitoring the course of treatment of the disease, disorder or abnormal condition of the subject. Monitoring can include evaluating the TFBS information of the subject at two or more time points. The evaluation can be based at least on the TFBS information determined at each of the two or more time points.

[0290] In some examples, a difference in TFBS information determined between two or more time points can indicate one or more clinical indications, such as (i) detection or diagnosis of a disease, disorder, or abnormal condition in a subject, (ii) prognosis of a disease, disorder, or abnormal condition in a subject, (iii) an increased risk of a disease, disorder, or abnormal condition in a subject, (iv) a decreased risk of a disease, disorder, or abnormal condition in a subject, (v) an effective course of treatment for the disease, disorder, or abnormal condition in a subject, and (vi) an ineffective course of treatment for the disease, disorder, or abnormal condition in a subject.

[0291] In some examples, the difference in TFBS information determined between two or more time points can indicate a diagnosis of a disease, illness, or abnormal condition of a subject. For example, if a disease, illness, or abnormal condition is not detected in a subject at an earlier time point, but a disease, illness, or abnormal condition is detected in a subject at a later time point, the difference indicates the detection or diagnosis of a disease, illness, or abnormal condition of the subject. Based on this indication of the detection or diagnosis of a disease, illness, or abnormal condition of a subject, a clinical action or decision can be made, for example, a new therapeutic intervention is prescribed to the subject. Clinical action or decision may include advising the subject to undergo a secondary clinical test to confirm the diagnosis of a disease, illness, or abnormal condition. The secondary clinical test may include an imaging test, a blood test, a computed tomography (CT) scan, a magnetic resonance imaging (MRI) scan, an ultrasound scan, a chest X-ray, a positron emission tomography (PET) scan, a PET-CT scan, a cytological assay, or any combination thereof.

[0292] In some examples, differences in TFBS information determined between two or more time points can indicate a prognosis for a disease, disorder, or abnormal condition in a subject.

[0293] In some examples, the difference of TFBS information determined between two or more time points can indicate that the subject has an increased risk of a disease, illness or abnormal condition. For example, if a disease, illness or abnormal condition is detected in the subject at an earlier time point and a later time point, the difference can indicate that the subject has an increased risk of the disease, illness or abnormal condition. Based on this indication of the increased risk of a disease, illness or abnormal condition, clinical action or decision can be made, for example, a new therapeutic intervention is prescribed for the subject or a therapeutic intervention is changed (for example, the current treatment is terminated and a new treatment is prescribed). Clinical action or decision may include advising the subject to perform a secondary clinical test to confirm the increased risk of a disease, illness or abnormal condition. The secondary clinical test may include imaging tests, blood tests, computed tomography (CT) scans, magnetic resonance imaging (MRI) scans, ultrasound scans, chest x-rays, positron emission tomography (PET) scans, PET-CT scans, cytological determinations or any combination thereof.

[0294] In some examples, the difference in the TFBS information determined between two or more time points can indicate that the subject has a reduced risk of a disease, illness or abnormal condition. For example, if a disease, illness or abnormal condition is detected in the subject at an earlier time point and a later time point, the difference can indicate that the subject has a reduced risk of the disease, illness or abnormal condition. Based on this indication of the reduced risk of a disease, illness or abnormal condition, clinical action or decision can be made, for example, a new therapeutic intervention or a change in therapeutic intervention (for example, continuing or ending current treatment) is prescribed for the subject. Clinical action or decision may include advising the subject to perform a secondary clinical test to confirm that the risk of a disease, illness or abnormal condition increases. The secondary clinical test may include imaging tests, blood tests, computed tomography (CT) scans, magnetic resonance imaging (MRI) scans, ultrasound scans, chest x-rays, positron emission tomography (PET) scans, PET-CT scans, cytological determinations or any combination thereof.

[0295] In some examples, the difference in TFBS information determined between two or more time points can indicate that the course of treatment for the disease, disorder or abnormal condition of the subject is effective. For example, if a disease, disorder or abnormal condition is detected in the subject at an earlier time point, but not detected in the subject at a later time point, the difference can indicate that the course of treatment for the disease, disorder or abnormal condition of the subject is effective. Based on this indication that the course of treatment is effective for the disease, disorder or abnormal condition of the subject, clinical action or decision can be made, for example, to continue or end the current therapeutic intervention for the subject. Clinical action or decision may include advising the subject to undergo a secondary clinical test to confirm that the course of treatment for the disease, disorder or abnormal condition is effective. The secondary clinical test may include imaging tests, blood tests, computed tomography (CT) scans, magnetic resonance imaging (MRI) scans, ultrasound scans, chest X-rays, positron emission tomography (PET) scans, PET-CT scans, cytological determinations, or any combination thereof.

[0296] In some examples, the difference in TFBS information determined between two or more time points can indicate that the course of treatment for the disease, illness or abnormal condition of the subject is invalid. For example, if the disease, illness or abnormal condition is detected in the subject at both the earlier time point and the later time point, and if effective treatment is indicated at the earlier time point, the difference can indicate that the course of treatment for the disease, illness or abnormal condition of the subject is invalid. Based on this indication that the course of treatment is invalid for the disease, illness or abnormal condition of the subject, clinical action or decision can be made, for example, ending the current therapeutic intervention and / or changing to (for example, prescribing) another new therapeutic intervention for the subject. Clinical action or decision may include advising the subject to perform a secondary clinical test to confirm that the course of treatment for the disease, illness or abnormal condition is invalid. The secondary clinical test may include imaging tests, blood tests, computed tomography (CT) scans, magnetic resonance imaging (MRI) scans, ultrasound scans, chest X-rays, positron emission tomography (PET) scans, PET-CT scans, cytological determinations or any combination thereof.

[0297] B. Indications

[0298] Non-limiting examples of cancers that may be inferred by the disclosed methods include acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), adrenocortical carcinoma, Kaposi's sarcoma, anal cancer, basal cell carcinoma, bile duct cancer, bladder cancer, bone cancer, osteosarcoma, malignant fibrous histiocytoma, brain stem glioma, brain cancer, craniopharyngioma, ependymoma, ependymoma, medulloepithelioma, pineal parenchymal tumor, breast cancer, bronchial tumor, Burkitt's lymphoma, non-Hodgkin's lymphoma, carcinoid tumor, cervical cancer, chordoma, chronic lymphocytic leukemia (CLL), chronic myeloid leukemia (CML), colon cancer, colorectal cancer, cutaneous T-cell lymphoma, ductal carcinoma in situ, endometrial cancer, esophageal cancer, Ewing's sarcoma, eye cancer, intraocular melanoma, retinoblastoma, Fibrous histiocytoma, gallbladder cancer, gastric cancer, glioma, hairy cell leukemia, head and neck cancer, heart cancer, hepatocellular (liver) cancer, Hodgkin's lymphoma, hypopharyngeal cancer, kidney cancer, laryngeal cancer, lip cancer, oral cancer, lung cancer, non-small cell carcinoma, small cell carcinoma, melanoma, oral cancer, myelodysplastic syndrome, multiple myeloma, medulloblastoma, nasal cancer, paranasal sinus cancer, neuroblastoma, nasopharyngeal cancer, oral cancer, oropharyngeal cancer, osteosarcoma, ovarian cancer, Pancreatic cancer, papilloma, paraganglioma, parathyroid cancer, penile cancer, pharyngeal cancer, pituitary tumor, plasma cell neoplasms, prostate cancer, rectal cancer, renal cell carcinoma, rhabdomyosarcoma, salivary gland cancer, Sezary syndrome, skin cancer, small intestine cancer, soft tissue sarcoma, squamous cell carcinoma, testicular cancer, laryngeal cancer, thymoma, thyroid cancer, urethral cancer, uterine cancer, uterine sarcoma, vaginal cancer, vulvar cancer, Waldenstrom's macroglobulinemia, and Wilms' tumor.

[0299] In various examples, the tumor is a colorectal disease selected from colorectal cancer, advanced adenoma, ulcerative colitis, Crohn's disease, irritable bowel syndrome (IBS).

[0300] In some examples, colorectal cancer is classified by stage, such as stage 0, stage I, stage IIA, stage IIB, stage IIC, stage IIIA, stage IIIB, stage IIIC, stage IVA, stage IVB, or stage IVC.

[0301] Example

[0302] Example 1: Inference of cell-specific transcription factor binding from cell-free DNA enables tumor subtype prediction and early detection of cancer

[0303] According to the methods and systems of the present disclosure, an analytical procedure was developed to determine the accessibility of transcription factor binding sites and applied to 244 cfDNA samples from patients with prostate, breast, or colon cancer.

[0304] Inferring TF binding from cfDNA has great diagnostic potential in cancer and beyond, and an improved and optimized bioinformatics pipeline was developed. This process is able to parse those components involved in nucleosome features at TFBSs, thereby objectively assessing and comparing the accessibility of TFBSs in different plasma samples. To validate the pipeline for clinical purposes, deep whole-genome sequencing (WGS) data were obtained from 24 plasma samples from healthy donors and 15 plasma samples from patients with metastatic prostate, colon, or breast cancer, where cfDNA also included circulating tumor DNA (ctDNA). In addition, shallow WGS data of 229 plasma samples from patients with the above tumor entities were generated (totaling more than 18.5 billion mapped plasma sequence reads). An additional 769 plasma samples from patients with colon cancer (n=592) and healthy controls (n=177) were also included (providing approximately 238 billion mapped plasma sequence reads). This approach profiles individual TFs rather than using a mixture of cfDNA signals generated by multiple cell types and analyzing them by Fourier transformation as other methods do to establish general tissue-specific patterns. The methods and systems provided herein provide insights into both tissue contributions and biological processes, which allow the identification of lineage-specific TFs suitable for analysis of tissues of origin and tumors of origin. In addition, through two relevant clinical application examples, the plasticity of TFBS in cfDNA of cancer patients and the potential of TF for prostate cancer subtype classification are demonstrated. First, these TF-based cfDNA assays can distinguish prostate adenocarcinoma from small cell neuroendocrine prostate cancer, which has important therapeutic implications. Secondly, large colon cancer groups can accurately establish the resolution limit and can explore the use of TF-based plasma analysis to detect early cancer staging.

[0305] Analysis of a small group of individuals with advanced cancer (n=5) showed that cfDNA fragmentation patterns can be used to detect non-hematopoietic features. To explore the potential of mapping TF-nucleosome interactions from cfDNA in more detail, known hematopoietic TF-nucleosome footprints were confirmed in plasma samples from healthy controls. The TFBS catalog list from the Gene Transcription Regulation Database (GTRD) was annotated with a recently published list of known or possible human TFs to generate a comprehensive TFBS-nucleosome occupancy map of 676 TFs from cfDNA. Using the bioinformatics pipeline provided in this article, different stringency criteria were evaluated to measure nucleosome features at TFBSs, and an index (called "accessibility score") and z-score statistics were established to objectively compare significant changes in TFBS accessibility in different plasma samples. For clinical purposes, a set of lineage-specific TFs are used to identify the tissue of origin of cfDNA or the tumor of origin of cancer patients. Finally, the accessibility score and z-score statistics are used to illustrate the changes in TFBS accessibility in cfDNA of cancer patients.

[0306] Knowing the precise position of nucleosomes relative to TF binding sites (TFBS) in the genome is useful for understanding how genes are regulated. To this end, analysis of cell-free DNA (cfDNA) in plasma, which also contains circulating tumor DNA (ctDNA) in cancer patients, provides an improved opportunity to study TFBS non-invasively in humans. Since cfDNA is primarily released from apoptotic cells after enzymatic digestion, it circulates primarily as mononucleosomal DNA. Therefore, sequencing cfDNA fragments generates nucleosome maps in which dyads of “perfectly positioned” nucleosomes, e.g., sites with high nucleosome preference, result in strong peaks of reads reflecting nucleosome phasing, while dyads of less preferentially positioned nucleosomes show diminished peaks or none at all. Therefore, cfDNA represents a unique analyte generated by endogenous physiological processes, allowing the generation of in vivo maps of nucleosome occupancy by whole-genome sequencing. This can be used to infer expressed genes through detailed analysis of nucleosome occupancy at transcription start sites (TSSs). The occupancy of cfDNA nucleosomes can reflect the footprint of TFs. In a small cohort of individuals with cancer (n = 5), cfDNA fragmentation patterns were matched to a reference dataset to detect non-hematopoietic signatures. However, TF-nucleosome interactions remain largely unmapped, and measurements of real-time TF dynamics at a genomic scale within the human body are needed.

[0307] The nucleosome mapping strategy and bioinformatics pipeline from cfDNA were used to address the following questions: (1) whether cfDNA accurately reflects known TF-nucleosome interactions; (2) generate the most comprehensive TF-nucleosome interaction map including data on 676 TFs; (3) establish an improved metric, called “accessibility score”, to objectively compare the accessibility of TFBSs in serial analyses within the same individual or between different individuals; and (4) define a set of lineage-specific TFs that are suitable for identifying the tissue of origin of cfDNA or the tumor of origin of cancer patients. In addition, this study examined whether tracking TFBSs from cfDNA of cancer patients can elucidate the accessibility of TFBSs and associated pathways. To this end, high-coverage whole-genome sequencing (WGS) data were obtained from 24 plasma samples from healthy donors (12 men and 12 women) and 16 plasma samples from patients with metastatic prostate, colon, or breast cancer. Furthermore, for validation purposes, shallow-coverage sequencing data of cfDNA from 229 patients were used for the above tumor entities to generate a total of more than 18.3 billion mapped plasma sequence reads, providing a broad in vivo view of important parts of the non-coding genome.

[0308] Nucleosome occupancy inferred from cfDNA reveals characteristic TF binding footprints

[0309] Nucleosome occupancy maps at TFBSs were prepared and tested for similarities and differences between healthy individuals and cancer patients. To this end, high-coverage cfDNA samples were obtained from 24 healthy controls (12 men and 12 women), in which the vast majority (more than 90%) of cfDNA was derived from apoptosis of leukocytes with little contribution from other tissues, and 11 plasma samples from 7 patients with 3 common tumor entities (e.g., 4 prostate cancers (P40, P147, P148, and P190), 1 colorectal cancer (CRC; C2), and 2 breast cancers (B7 and B13), with ctDNA fractions ranging from 18-78% ( Figure 5 )).

[0310] 676 TFs from the Gene Transcription Regulation Database (GTRD; Version 18.01) were used; because they contain detailed TFBS information based on ChIP-seq data for various tissue samples. TFs were annotated with the latest catalog list of 1,639 known or possible human TFs (Fig. 6A). Due to the potential high number of TFBSs that TFs bind to at variable frequencies, three different stringency criteria were defined (Fig. 6A): first, all TFBSs of all tissue samples in GTRD; second, those peaks supported by more than 50% of the maximum number of samples (subsequently referred to as ">50%-TFBS"; in both analyses, all 676 GTRD TFs were included); third, 1,000 TFBS / TFs supported by most samples ("1,000-msTFBS"; 505 TFs met this criterion).

[0311] Establishment of TF-nucleosome interactions

[0312] Fig. 2A-Fig. 2E shows the TF-nucleosome interaction established by cell-free deoxyribonucleic acid (cfDNA). Fig. 2A shows a highly organized, for example, phased nucleosome region resulting in an oscillating read depth pattern, wherein the peak read indicates the position of the dyad, for example, the midpoint of a typical nucleosome. The less clear positioning of the nucleosome produces a fairly shallow coverage spectrum. Fig. 2B shows the TFBS data of 676 TFs retrieved from GTRD, and compared with the cataloged list of known or possible human TFs. Three different calculations were performed, each with increased stringency. Fig. 2C shows that in all analyzed cfDNAs, the coverage pattern of CCTCC binding factor (CTCF) is similar, which is consistent with DNase hypersensitivity data, showing that the accessibility of blood (GM12878) and epithelial tissues (e.g., prostate (LNCaP) and colon (HCT116)) is roughly the same. In this group and subsequent groups, the spectrum calculated from the healthy control is shown in gray, while the spectrum of the patient source is shown in the indicated color. Figure 2D shows that the hematopoietic lineage specificity of TFs (PU.1, LYL1, SPIB) was confirmed by DNA hypersensitivity assays and their amplitudes were reduced in cancer patient plasma compared to healthy controls. In contrast, the amplitude of the epithelial TF GRHL2 was increased in cfDNA from cancer patients. Figure 2E shows the accessibility map and DNase hypersensitivity of the TF FOXA1, illustrating preferential amplitude changes in hormone-dependent cancers, e.g., prostate and breast cancer patients.

[0313] Using 24 samples of cfDNA from healthy controls, an average of 435,135,450 (range: 352,904,231-556,303,420) sequencing reads were obtained per sample. TF binding sites are often located on the sides of a series of strongly positioned nucleosomes, visible as a periodic oscillating pattern (Figure 6A-Figure 6E). In contrast, negative control normal high molecular weight DNA was used to observe uniform coverage on TFBS (Figure 8A-Figure 8E). The oscillating coverage pattern produced by CTCF binding sites surrounded by a strongly positioned nucleosome array remained similar in all samples analyzed, regardless of whether the cfDNA was from healthy controls or from cancer patients (Figure 2C). These results are consistent with DNase hypersensitivity assays in the Encyclopedia of DNA Elements (ENCODE) database of cell lines GM12878 (B lymphocyte cell line from a female donor of European descent), LNCaP (androgen-sensitive human prostate adenocarcinoma cell line), and HCT116 (human colon cancer cell line) (Figure 2C).

[0314] Compared with healthy controls, ctDNA in the plasma of cancer patients altered the balance between hematopoietic and epithelial cell DNA, for example, resulting in reduced amplitudes of the lineage-restricted hematopoietic TFs purine-rich box 1 (PU.1), LYL1 (lymphoblastic leukemia 1), and the lymphocyte lineage-restricted transcription factor SPIB in samples derived from cancer, and increased amplitudes of the TF GRHL2 (a pioneer TF of epithelial cells) (Figure 2D). The lineage specificity of these TFs was also confirmed with data from publicly available DNase hypersensitivity assays (Figure 2D). As another example of a well-established TF, FOXA1 was analyzed, which synergizes with nuclear hormone receptors in endocrine-driven breast and prostate tumors. Consistent with the DNase hypersensitivity assay, preferentially increased accessibility of FOXA1 was observed in plasma samples from patients with prostate and breast cancer. Comparisons with ENCODE data were also performed, in which mononucleosome-bound DNA fragments were generated by micrococcal nuclease (MNase) digestion (Figures 6A-6E and Figure 7 ). A coverage-independent analysis was performed (Figure 6B), and the spatial density of cfDNA fragments associated with individual recognition sequences was calculated (Figure 6C). Sequence-specific TFs can have typical motifs and important secondary motifs that can correspond to motifs of other TFs. A catalog of TFBSs was also generated that can be affected by the co-binding of more than one TF from all 676 TFs and 505 TTFs from the 1,000-msTFBS (Figure 6D). In addition, uniform coverage was observed on the TFBS using purified high molecular weight DNA as a negative control (Figure 6E). Therefore, these results indicate that the corresponding TFBS coverage profiles are very similar to each other, indicating that the method has high accuracy and that the patterns obtained for any given TF are reproducible in all samples.

[0315] Since sequence-specific TFs may have typical motifs and important secondary motifs that may correspond to motifs of other TFs, the overlap between various TFBSs was calculated (Figure 6D). A list of TFBSs that may be affected by co-binding of more than one TF was generated ( Fig. 27 ). Examples of such overlapping effects are TFs SP1, SP2, NF-YA and NF-YB, whose overlap ranges from 10% to 36%, where the TFBS-nucleosome profiles are indeed similar (Figure 13A). The main origin of cfDNA in the blood is particularly reflected in the well-positioned nucleosomes flanking the binding sites of lineage-restricted hematopoietic TFs such as purine-rich box 1 (PU.1), LYL1 (lymphoblastic leukemia 1), and lymphocyte lineage-restricted transcription factor SPI-B (Figure 13A). In contrast, the TFBS profile of GRHL2, a pioneer TF of epithelial cells, showed a significantly reduced amplitude (Figure 13A).

[0316] The three different stringency criteria were applied to evaluate CTCF binding sites surrounded by strongly positioned nucleosome arrays, and the expected oscillation patterns of >50%-TFBS and 1,000-msTFBS were preferentially observed (Figure 2C-2E). In addition, different binding sites were evaluated using CTCF ( Fig.17 ), and an independent analysis was performed as additional confirmation coverage ( Fig.17 ), and calculated the spatial density of cfDNA fragments associated with a single recognition sequence. The resulting heat map shows that nucleosome phasing is uniform in most of the analyzed sites, which is consistent with the coverage profile.

[0317] Figure 7 TF-nucleosome interaction maps for various TFs are shown. Additional comparisons between cfDNA and MNase-seq coverage profiles around transcription factor binding sites are shown. CTCF as a special example of a TF with multiple distinct binding sites

[0318] To explore different TFBSs for the same TF, CCCTC-binding factor (CTCF) was used. CTCF is present at 55,000-65,000 binding sites in the mammalian genome. Of these sites, about 5,000 are ultra-conserved, about 50% are located in intergenic regions, about 15% are located near promoters, and about 40% are intragenic. In addition, chromosomes are divided into evolutionarily conserved higher-order chromosomal structures named topologically associating domains (TADs), and their boundaries are enriched with binding sites for CTCF and cohesin. In mammals, 15% of genomic CTCF binding sites are located at TAD boundaries, while the other 85% are located inside TADs.

[0319] CTCF sites that overlap or exceed TAD boundaries, TSS proximal (e.g., within about 2 kbp) or distal (greater than 2 kbp), and ultra-conserved sites were analyzed separately. All tissue types in GTRD were analyzed, and different CTCF coverage patterns were obtained, among which the ultra-conserved CTCF sites had the largest amplitude (Figure 14A, left sub-figure). When the analysis was limited to those binding sites that were called in more than 50% of all samples in GTRD, the resulting spectra became more similar to each other (Figure 14A, right sub-figure). As a coverage-independent confirmation of the TFBS signal, the length of each cfDNA fragment was plotted as a function of the distance from the midpoint of the fragment to the CTCF binding site. The resulting heat map confirmed the signal periodicity consistent with the coverage-based oscillation pattern (Figure 6B). In addition, in order to more closely analyze the distribution of fragments associated with a single recognition sequence, the spatial density of cfDNA fragments within a 2 kb region centered on the TFBS was calculated, and the sites were graded according to the coverage of the central 40 bp. The resulting heat map showed that nucleosome phasing was uniform in most of the sites analyzed ( FIG. 6C ), again consistent with the coverage profile.

[0320] “Accessibility score” can accurately infer TF binding from cfDNA

[0321] Binding sites where nucleosomes are repositioned by intervening TF binding ensure that the corresponding DNA is accessible to proteins and the transcription and replication machinery. Some TFs show evenly spaced nucleosome peaks, including their binding sites (e.g., PU.1 and GRHL2 in Figure 13B), while other TFs have wider troughs in their binding sites (e.g., Figure 13A), similar to the troughs of TSSs. For the latter, basic binding site width differences of TFs were measured (Figure 8D). This measurement identified 55 TFBSs with TFBSs exceeding 300 bp, of which 26 had binding sites close to the size of a dinucleosome (312-352 bp) ( Figure 28A-28B To test whether these patterns were a side effect of binding to CpG island promoters, the CpG density (box plots) and colocalization with CpG islands (bar / pie charts) of the 55 broad TFs were plotted versus TFs with narrowly defined binding sites.

[0322] Certain lineage-specific TFs are suitable for determining the tissue of origin of plasma DNA. However, to determine which TFs may be useful in this application, it is necessary to evaluate the accessibility of TFs (e.g., at their binding sites in cfDNA). Conventional methods may not be able to evaluate the accessibility of TFs at their binding sites in cfDNA as a proxy for their activity. To implement this approach, the coverage profiles of TF-specific nucleosomes were investigated. Calculations were performed separately for TFBSs within and outside the transcription start site (TSS) (Figure 8A) and for all GTRD tissues with >50%-TFBS (Figure 8B). These analyses showed that the average TFBS pattern included two signals: TSS-proximal (within 2 kb of the TSS, generating a “low-frequency pattern”) and TSS-distal (more than 2 kb away from the TSS peak, generating a “high-frequency pattern”), corresponding to more evenly spaced peak signals. To suppress effects on coverage that are not caused by preferential nucleosome positioning and to remove local biases from the nucleosome data, a Savitzky-Golay filter was used for detrending (Figure 3A). The obtained low-frequency signal was then used to normalize the high-frequency signal, and the data range of the high-frequency signal was subsequently recorded (the maximum value minus the minimum value of the data value, corresponding to the amplitude). Since the range of the high-frequency signal depends on the number of TFBSs (Figure 3B) (except for the 1,000-ms TFBS), these range values ​​were corrected by LOESS smoothing because they depend on the number of TFBSs (Figure 3B), and the rank was then calculated as a measure of the accessibility of each TFBS. Figure 3C shows a wavelet analysis of GRHL2: a heat map of the periodicity of the area around the TFBS along GRHL2 (left sub-figure). The color code represents the quantile of the signal power distribution. The average power of the transcription factor GRHL2 period (right sub-figure). Figure 3D shows the detrended original (black) and reconstructed (red) nucleosome coverage spectra of the transcription factor GRHL2 derived from the wavelet analysis.

[0323] To test potential alternatives for TF accessibility assessment, an unbiased, detrended signal was reconstructed for the segment between 135 and 235 bp by wavelet analysis, and the power of the signal was summed over 2,000 bp of flanking TFBSs (Fig. 3E-3F). To benchmark the performance of Savitzky-Golay filtering and wavelet analysis, cfRNA data were used, compared with expressed (e.g., greater than 10 FPKM) TFs (>50%-TFBS; Savitzky-Golay filtering: p = 1.75 × 10 -13 ; Power sum (wavelet analysis): p = 0.0004049; 1,000-ms TFBS; Savitzky-Golay filter: p = 1.254 × 10 -11A significant decrease in accessibility was observed for unexpressed TFs (e.g., <0.01 FPKM [fragments per kilobase million]) compared to controls (Figure 14B). When the adjusted range was compared to the average DNase coverage (>50%-TFBS; Savitzky-Golay filter: p < 2.2 × 10 -16 ; Power sum (wavelet analysis): p<2.2×10 -16 ; 1,000-ms TFBS; Savitzky-Golay filter: p < 2.2 × 10- 16 These differences were also significant when the 1,000-ms TFBSs were compared using the 1,000-ms TFBS-based accessibility test ( 1,000-ms TFBS-based accessibility test, 2,000-ms TFBS-based accessibility test, and 1,000-ms TFBS-based accessibility test, respectively). The detection threshold for TFBS accessibility was then defined as ±3 mean standard deviations from the normal sample ( z score of 3) due to the slightly better performance of the Savitzky-Golay filter. For the evaluation based on all or >50%-TFBSs, the detection threshold for the normalized accessibility score was ±253 and ±88 for the 1,000-ms TFBSs, which had fewer analyzable TFs ( Fig. 8C ).

[0324] In addition, a comprehensive TF-nucleosome interaction map of 676 GTRD TFs was generated from cfDNA ( Figure 14B ; Fig.17 TF-nucleosome interactions can be characterized by, for example, mapping using ChIP-seq datasets from the ENCODE Consortium, which characterize the chromatin structure around 119 human TFs. These efforts led to the TF-centric web repository Factorbook, which includes data for 167 TFs. However, these data are based on ex vivo tissue samples, whereas this paper investigates in vivo accessibility generated by endogenous processes.

[0325] These results demonstrate a reliable method for assessing TFBS accessibility specifically using cfDNA in clinical diagnostics.

[0326] TFBS accessibility in cfDNA across multiple cell types

[0327] Plasma samples from three common tumor entities were used to demonstrate clinical application. This study first analyzed 11 plasma samples from seven patients, for example, four with prostate cancer (P40, P147, P148, and P190), one with colorectal cancer (CRC; C2), and two with breast cancer (B7 and B13) (Figure 4A-Figure 4F). cfDNA from C2, P40, P147, and P148 were sequenced on the Illumina NovaSeq platform (with an average of 688,482,254 (range: 541,216,395-870,285,698) sequencing reads), while B7 (328,515,075 reads) and B13 (379,733,061 reads) were sequenced on the Illumina NextSeq platform.

[0328] CTCF is a special transcription factor that is active in every tissue because it regulates the 3D structure of chromosomes that is conserved throughout the tissue. The amplitude of CTCF remained similar in all samples analyzed, regardless of whether the cfDNA was derived from healthy controls or from cancer patients (Figure 2C). This is consistent with DNase hypersensitivity assays from the ENCODE database for the cell lines GM12878, LNCaP (androgen-sensitive human prostate adenocarcinoma cell line), and HCT116 (human colon cancer cell line), showing increased accessibility of CTCF binding sites in various tissues (Figure 2C). However, the proportion of ctDNA in cancer patients increases, which changes the balance between DNA from hematopoietic cells and epithelial cells within cfDNA. Therefore, the amplitude of hematopoietic TFs (PU.1, Lyl-1, and Spi-B) decreased, while the amplitude of epithelial TF GRH-L2 increased, indicating that the contribution of the hematopoietic system was diluted and the contribution of epithelial cells increased (Figure 2D). These observations are again consistent with the DNase hypersensitivity assays (Figure 2D).

[0329] As another example of a confirmed TF, we analyzed FOXA1, a TF widely expressed in different tissues, where it controls cell differentiation and organ function. Moreover, FOXA1 cooperates with nuclear hormone receptors in endocrine-driven breast and prostate tumors, and its expression in the prostate is associated with castration-resistant prostate cancer (CRPC). In fact, consistent with the DNase hypersensitivity assay, preferentially increased accessibility of FOXA1 was observed in plasma samples of prostate and breast cancer patients (Figure 2E).

[0330] Inference of TF binding from cfDNA supports molecular subtyping of prostate cancer

[0331] In some cases, it may be important to assess the extent to which tissue-specific TFs are suitable for identifying tumors of origin and molecular subtyping. To this end, prostate cancer is a tumor entity of particular interest, as a common mechanism (approximately 20%) for developing therapeutic resistance to novel agents targeting the AR pathway (such as abiraterone or enzalutamide) is the transdifferentiation of adenocarcinomas into treatment-induced small cell neuroendocrine prostate cancers (t-SCNCs). This transdifferentiation has enormous clinical implications, as it requires changes in therapeutic approaches, and the involvement of multiple TFs in this transdifferentiation process can be investigated (Figure 4A).

[0332] Increased accessibility of multiple TFs was detected in one tumor entity but not in others. For example, plasma samples from CRC patient C2 showed increased accessibility to c-Jun and JunD oncogenes ( Fig.15 ), and colon preference was confirmed by DNA hypersensitivity assay ( Fig.15 ).

[0333] Another analysis was performed on prostate cancer samples. The expression of human TFs across tissues and various cell types provided by (Lambert et al., 2018) and the publicly available Human Protein Atlas data were screened and confirmed prostate lineage specificity for TFsAR, HOXB13, and NKX3-1, which was also reflected in the DNase hypersensitivity assay of the prostate cancer cell line LNCaP (Figure 4B-4D).

[0334] HOXB13 is a highly lineage-specific homeobox TF gene that is important in prostate development and maintains high expression levels into adulthood in the normal prostate (Figure 4B). The NKX3-1 homeobox gene is one of the earliest genes expressed during prostate epithelial maturation and is essential for prostate epithelial differentiation and required for prostate tumor progression (Figure 4C). Both TFs showed increased accessibility at their binding sites only in cfDNA from patients with prostate cancer, and in addition, tissue specificity was confirmed by DNase hypersensitivity assay (Figure 4C). Fig.15 and Figures 4B-4D ).

[0335] Therefore, these TFs showed increased accessibility at their binding sites only in cfDNA of patients with prostate cancer. As AR is particularly relevant in prostate cancer, not only the AR binding sites defined by GTRD were used, but also those reported by (Pomerantz et al., 2015), thereby identifying 9,179 tumor AR binding sites with high binding strength in tumors (tumor AR binding sites, T-ARBS) and 2,690 normal AR binding sites with high binding strength in normal samples (normal AR binding sites, N-ARBS) by analyzing the AR cis-trans group. In fact, while N-ARBS were inaccessible from both controls and patients, T-ARBS showed increased accessibility in patient plasma samples (Figure 4D).

[0336] Confirmation of lineage-specific transcription factors in pooled samples

[0337] This approach can also be applied to samples with less coverage and that are largely rearranged. To test this, the TF analysis was repeated after downsampling P148_1 (819,607,690 reads) and P148_3 (768,763,081 reads) to approximately 50 million reads. This comparison showed that the same TFs were identified as having increased or decreased accessibility, indicating that samples with fewer sequencing reads are suitable for these analyses. Four cfDNA samples (P21_2, P111_1, P111_4, P166_1) with a tandem replicon phenotype (Viswanathan et al., 2018) and one case (P143_3) with chromosome breakage on chromosome 10 were subsequently analyzed (mean: 52,869,911; range: 41,780,819-84,049,593) (Figures 18A-B). In these cases, the epithelial TFs FOXA1, GRHL1, and GRHL2, as well as the prostate lineage-specific TFs AR, HOXB13, and NKX3-1 again showed increased accessibility (Figure 18A-18B), indicating that results can be obtained even when demand is impaired, and further suggesting that altered accessibility of these TFs appears to be a universal feature of prostate cancer.

[0338] As further confirmation of the robustness and reproducibility of lineage-specific TFs in cfDNA, analysis of multiple cfDNA sample libraries generated with shallow coverage (<0.2x) showed that these TFs had increased accessibility in most or all samples, e.g., lineage-specific TFs had higher accessibility scores, while others reached the average. To this end, cfDNA samples were pooled separately for prostate cancer cases (n=69), colon cancer cases (n=100), and breast cancer cases (n=60), and the analysis was repeated. The epithelial TF GRHL2 persisted with increased accessibility, while the accessibility of hematopoietic TFs decreased (Figure 8A-Figure 8E). In the prostate cancer cfDNA pool, the lineage-specific TFs AR (340; 4.0), HOXB13 (712; 8.4), and NKX3-1 (253; 3.0) showed increased accessibility, indicating that changes in accessibility of these TFs are universal features in prostate cancer (Figure 8A-Figure 8E) and that these features are prevalent in prostate cancer and may be suitable for identifying the tumor of origin from cfDNA.

[0339] Fig. 9 Analysis of pooled shallow coverage cfDNA is shown. Accessibility of pooled cfDNA samples of prostate cancer (n=69), colon cancer (n=100), and breast cancer (n=60) for epithelial TF GRHL2 and hematopoietic TFs (PU.1, LYL1, and SPIB) is shown. Accessibility of lineage-specific TFs AR, HOXB13, and NKX3-1 is also shown in the prostate cancer cfDNA pool.

[0340] These analyses suggest that the accessibility of these TF binding sites in cfDNA may also be used to identify the tumor of origin, as some lineage-specific TFs are commonly altered in epithelial and prostate cancers, respectively.

[0341] For tumor subclassification, the index case P148 was used, in which two plasma samples (P148_1, P148_3) taken 12 months apart were analyzed, during which time prostate adenocarcinoma transdifferentiated into t-SCNC. These two samples showed significant changes in TFBS accessibility ((Kendall's Tau: 0.7573), specifically reflected in multiple TFs. t-SCNC is no longer an androgen-dependent stage of prostate cancer, and therefore, accessibility to AR binding sites is no longer observed in sample P148_3 (Figure 4E). Due to its close collaboration with nuclear hormone receptors, accessibility to FOXA1 is correspondingly reduced (Figure 4E). In addition, changes in cell type identity become apparent with reduced accessibility to binding sites for prostate-specific lineage TFs HOXB13 and NKX3-1 (Figure 4E) and epithelial TF GRHL2 (Figure 10A). TF changes associated with neuronal development include GLI-similar 1 (GLIS1) accessibility was increased (Figure 10B), a TF whose expression was significantly increased under hypoxic conditions. Hypoxia was discussed to promote the progression of prostate adenocarcinoma to an androgen-independent state and further to downregulate the repressor factor 1 (RE-1) silencing transcription factor (REST), which induced neuroendocrine reprogramming and was observed to indeed significantly reduce the accessibility of REST (Figure 4E). In addition, N-MYC is involved in AR signaling inhibition and neuroendocrine program regulation, which was reflected in increased accessibility (Figure 4E). These observations suggest that TFBS can have highly plastic impact pathways at some cancer disease stages.

[0342] To demonstrate that prostate cancer subtype classification based on TFBS from cfDNA is feasible, plasma samples from four t-SCNC cases (P170_2, P179_4, P198_5, and P240_1) were further analyzed. For these cases, it was shown that the approach is also applicable to cfDNA sequenced at a lower coverage by downsampling the plasma samples P148_1 (819,607,690 reads) and P148_3 (768,763,081 reads) to approximately 50 million reads. The reduction in reads leads to an increase in the noise level, which depends on the number of TFBSs but is negligible for TFs with more than 1,000 TFBSs ( Fig.11), so the analysis of the above highly associated TFs was not affected. The analysis was repeated for the above 4 samples, each sequenced with approximately 50 million reads, and reduced accessibility of TFs AR, FOXA1, HOX-B13 and NKX3-1, or increased accessibility of N-MYC was again observed (Figure 4F). Only two of these four cases showed reduced accessibility of REST (P170_2 and P198_5; Figure 4F), which is consistent with the observation of REST downregulation in 50% of neuroendocrine prostate cancer cases. Only in these two cases, GLIS1 again did have increased accessibility (z scores: P170_2: 4.3; P198_5: 4.4), suggesting that this hypoxia-related TF may be associated with REST downregulation.

[0343] Accessibility of TFBSs can change during the course of disease

[0344] To address the question of whether TF accessibility remains stable over time, serial samples were obtained and analyzed from four patients (P40, P147, P148, C2). The analysis was limited to 1,000 ms TFBS, and three of the four pairs of plasma samples did not show significant differences (control: median: 0.8404 ± 0.0196 (IQR); P40: 0.8620; P147: 0.8370; C2: 0.8719; Kendall's Tau, each) ( Fig.12 ).

[0345] Between P147_1 and P147_3, a new high-amplitude amplification including the RET gene was formed, while C2_7 lost the amplification including KRAS, which was observed in C2_6. Both RET in prostate cancer and KRAS in CRC can affect the PI3K / AKT / mTOR pathway, so downstream targets such as TF CREB were studied; however, accessibility was not different from control plasma samples and remained unchanged. Between P40_1 and P40_2, resistance to androgen deprivation therapy (ADT) had been formed, which was reflected in the high-level amplification of the AR gene. However, if AR expanded its transcriptional target library, it was not obvious at the T-ARBS and N-ARBS mentioned above (Figure 18A-Figure 18B). This analysis adopts a conservative approach, and changes are only observed when the accessibility score between one analysis and the next differs by ≥100, and can explain the reduced or limited differences between these samples.

[0346] For case P148 (Pearson correlation: 0.777291), there was a significant change in the accessibility of TFs, in which the tumor transdifferentiated from a prostate adenocarcinoma (P148_1) to a neuroendocrine tumor (P148_3). Neuroendocrine tumors are no longer the androgen-dependent stage of prostate cancer, and therefore no longer require the accessibility of AR binding sites, which is reflected accordingly in these analyses (Figures 16A-16C). Due to the loss of accessibility to the binding sites that determine the epithelial cell fate of TFs GRHL2 and GRHL3, the change in cell type identity of this prostate cancer case is obvious (Figures 16A-16C). In addition, similar accessibility reductions were observed for other prostate-specific lineage TFs HOXB13, NKX3-1, FOXA1, GATA2, and GATA3.

[0347] Importantly, as mentioned above, changes in TFs related to neuronal development were also observed. Hypoxia frequently occurs in advanced solid tumors, can promote the development of prostate adenocarcinoma into an androgen-independent state, and can induce neuroendocrine programs. In fact, increased accessibility of GLI-similar1 (GLIS1) was observed, a TF whose expression was significantly increased under hypoxic conditions. In addition, hypoxia downregulates the repressor factor 1 (RE-1) silencing transcription factor (REST) ​​in prostate cancer epithelium and induces the expression of neuronal genes involved in neuroendocrine reprogramming. REST is a key mediator of neuroendocrine differentiation caused by androgen depletion, and indeed a decrease in the accessibility of REST was observed (Figure 16A-Figure 16C).

[0348] Differences related to stem cell characteristics were also observed. TF SOX2 and SOX11 are upregulated during neuroendocrine transdifferentiation. Increased accessibility of these two TFs was observed; however, these were already present in plasma sample P148_1 and therefore preceded other changes (Figure 6E). This example confirms another feature of this real-time analysis, for example, the order of events can be determined. Another change associated with stem cells is a reduced accessibility of FOXA2 (Figure 16A-Figure 16C).

[0349] Changes were also observed in less well-characterized TFs, such as TF ZNF644 (ZNF644 is one of the core subunits of the G9a / GLP complex, which mediates mono- and dimethylation of Lys9 of histone H3 at specific gene loci, which is associated with transcriptional repression) or ZNF701 (Figures 16A-16C), whose potential role in the transdifferentiation process is still unclear.

[0350] Prostate cancer classification based on TFs from cfDNA

[0351] To show that the method is applicable to samples sequenced at lower coverage, plasma samples P148_1 (819,607,690 reads) and P148_3 (768,763,081 reads) were downsampled to approximately 50 million reads. The reduction in reads resulted in an increase in noise level, which depended on the number of TFBSs and was negligible for TFs with more than 1,000 TFBSs (Figure 18A-Figure 18B). Therefore, the accessibility analysis of the above highly relevant TFs involved in transdifferentiation to neuroendocrine carcinomas was not affected.

[0352] PSA NSE REST P170_2 3.5 133 down P179_4 0.56 218 nl P198_5 29.4 >370 Downward adjustment P240_1 3.2 542.4 nl

[0353] discuss

[0354] This study provides methods and a bioinformatics software pipeline for inferring the binding of tumor cell-specific transcription factors from cell-free DNA in blood, relevant for clinical diagnosis and non-invasive tumor classification. Although some studies have adopted a gene-centric approach when evaluating somatically acquired alterations, this analysis uses a significant portion of the noncoding genome, focusing on TFBS. Since many TFs preferentially bind in open chromatin and must therefore interact with nucleosomes, cfDNA, which is primarily mononucleosomal, was used as it allows mapping of nucleosome positions. The unique feature of this method, compared to technical variants associated with in vitro assays, is that it is used to generate in vivo data on TFBS through endogenous physiological processes. However, these data are highly correlated with DNase I hypersensitivity data for the cell lines GM12878, LNCaP, or HCT116, demonstrating the reliability of this method.

[0355] In contrast to other analyses that can use general tissue-specific patterns of mixtures of cfDNA signals generated by multiple cell types and analyze them by Fourier transform, the methods and systems of the present disclosure can dissect individual TFs and thereby establish lineage-specific TFs for clinical applications. Due to the improved resolution of TFBS analysis, it is possible to monitor the accessibility of TFBSs from cfDNA and reveal their plasticity during disease processes, such as reprogramming to different cell lineages. In addition, whereas other analyses may require more than 1.5 billion reads / sample, this study shows that approximately 50 million reads are sufficient for in-depth TF analysis, making this approach more efficient and economical for clinical applications. Importantly, this cfDNA TFBS bioinformatics pipeline can classify tumors, thereby filling important diagnostic gaps in the management of patients with prostate cancer, for example.

[0356] This work provides several substantial improvements to current techniques for TF profiling. First, using cfDNA, a cataloged list of TFBSs from the GTRD (annotated with a list of high-confidence TFs), 676 TFs can be analyzed from cfDNA. Second, this bioinformatics pipeline is used to establish an improved metric (accessibility score) to allow comparison of the accessibility of TFBSs between different cfDNA samples. Third, the use of z-score statistics based on comparisons of control samples (e.g., reference samples) and case samples (e.g., test samples obtained from subjects) allows identification of significant changes in TFBS accessibility. Fourth, for cfDNA-based clinical applications, the use of lineage-specific TFs for blood (PU.1, LYL1, and SPIB), epithelial (GRHL2), and prostate lineages (AR, HOXB13, NKX3-1) is shown. This is in contrast to other approaches, which involve a mixture of signals generated by multiple cell types contributing to cfDNA, and analysis by Fourier transforms. The current evaluation of clinical utility suggests that these TFs are widely applicable to identifying individuals with epithelial or prostate cancer. We demonstrate the ability to monitor TFBS accessibility over time, specifically during transdifferentiation of prostate adenocarcinoma to neuroendocrine tumors, allowing for noninvasive assessment from peripheral blood of dramatic changes involving AR, epithelial, prostatic, and neuronal lineages.

[0357] TF nucleosome interaction maps can be heterogeneous, including signals from all cell types that produce cfDNA. Plasma samples from individuals that appear to have a large ctDNA load can be used, which may affect the sensitivity of the measurement. In addition, this method uses whole genome sequencing with relatively high coverage (about 50 million reads), which exceeds the shallow sequencing plasma method used to establish SCNA.

[0358] However, advanced prostate cancer (the tumor entity analyzed here) is a classic example of the stubbornness and lethality of metastatic cancer. Clinical biopsies of metastatic lesions are not routinely performed, so detailed knowledge of the molecular mechanisms controlling prostate cancer cell survival and progression is lacking. Indeed, tumor research lacks dynamic models, particularly dynamic profiling of clinical samples, to explore the conversion and interaction between pathways. Due to the potential of TF for regulating gene transcription in the entire genome, and its acute lineage-specific mode, their detailed analysis provides a unique opportunity for improving clinical diagnosis. Through improved transcriptional regulation profile analysis methods, this data can also provide a basis for further careful analysis of non-coding genomes.

[0359] method

[0360] Subjects

[0361] This study was approved by the Ethics Committee of the Medical University of Graz (approval numbers 21-227ex 09 / 10 [breast cancer], 21-228ex 09 / 10 [prostate cancer], 21-229ex 09 / 10 [colorectal cancer], and 29-272ex 16 / 17 [high-resolution analysis of plasma DNA]) and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all patients and healthy probands, respectively. Some plasma samples have been analyzed in other studies, for example, plasma samples from patients B7 and B13 and P40, P147, and P148.

[0362] B7 and B13: These studies analyzed plasma DNA and primary tumors obtained from matched and synchronized pairs of two metastatic breast cancer cases (B7, B13) by whole-genome sequencing and RNA-Seq. Plasma DNA was sequenced at high coverage (B7: ~411 million reads; ~8.2x; B13: ~455 million reads; ~9.1x), and copy number changes were calculated. Focal amplifications that frequently occur in breast cancer were identified, such as amplification of 11q13.3 (15 genes including CCND1) in B7 or amplification of 8p11 (31 genes including FGFR1) and 17q12 (46 genes including ERBB2) in B13.

[0363] P40: Initial plasma DNA analysis of patient P40 showed multiple copy number changes on most autosomes, whereas no copy number changes were observed on chromosome X (Figure 3B). Prior to this treatment, the patient had undergone localized radiotherapy. Due to disease progression, treatment was switched to the third-generation LHRH antagonist degarelix. However, despite switching to this treatment, progression occurred 10 months later, and repeated plasma analysis showed a focal amplification on chromosome Xq12, which carries the AR gene, despite the same autosomal changes.

[0364] B7 and B13: These studies analyzed plasma DNA and primary tumors obtained from matched and synchronized pairs of two metastatic breast cancer cases (B7, B13) by whole-genome sequencing and RNA-Seq. Plasma DNA was sequenced at high coverage (B7: ~411 million reads; ~8.2x; B13: ~455 million reads; ~9.1x), and copy number changes were calculated. Focal amplifications that frequently occur in breast cancer were identified as previously defined (Ulz et al., 2016b), such as amplification of 11q13.3 (15 genes including CCND1) in B7 or amplification of 8p11 (31 genes including FGFR1) and 17q12 (46 genes including ERBB2) in B13.

[0365] P21: Patient P21 was diagnosed with metastatic castration-resistant prostate cancer (CRPC). After 4 months of treatment with the LHRH antagonist degarelix, the patient showed signs of clinical progression followed by an increase in PSA values. At the progression stage (P21_2), a cfDNA profile with a tandem replicon phenotype was observed.

[0366] P40: At diagnosis, patient P40 was classified as having castration-sensitive prostate cancer (CSPC). As the patient did not show a clinical response to previous radiotherapy, treatment was switched to the use of the LHRH antagonist degarelix. Initially, the patient showed a good response to androgen blockade (PSA value decreased from 425.3 ng / mL to 115.3 ng / mL), but after 10 months, he progressed to CRPC (PSA: 656.0 ng / mL). The cfDNA profile of the patient at progression (P40_2) showed high-level AR amplification on chromosome X.

[0367] P40: Initial plasma DNA analysis of patient P40 showed multiple copy number changes on most autosomes, while no copy number changes were observed on chromosome X (Figure 4B). Prior to this treatment, the patient underwent local radiotherapy. Due to disease progression, the patient's treatment was switched to the third-generation LHRH antagonist degarelix (Rick et al., 2013). However, despite the switch to this treatment, progression occurred 10 months later, and repeated plasma analysis showed focal amplification on chromosome Xq12, which carries the AR gene, despite the same changes on the autosomes.

[0368] P111: The first sample P111_1 was obtained at the time of diagnosis of prostate cancer. The patient already had multiple malignant lesions in the bones, lymph nodes and kidneys. First-line treatment was the GnRH-analog goserelin, followed by radiotherapy. Between the two samples P111_1 and P111_4, the patient responded well to treatment (CSPC, PSA dropped to 15.5 ng / mL). Two months before the sampling of P111_4, clinical progression occurred and the patient developed CRPC. The cfDNA sample (P111_4) was analyzed during the progression phase, and partial AR amplification was observed. In addition, the patient received chemotherapy (docetaxel), but no further response was found.

[0369] P143: Patient P143 was diagnosed with metastatic prostate cancer 6 years before sample P143_3 was collected. Previously, the patient had undergone different anti-androgen therapies, including a second generation anti-androgen (abiraterone). Therefore, when sample P143_3 was obtained, he was heavily pretreated. Due to disease progression and after multiple treatment failures with different ADTs, chemotherapy (microtubule inhibitor – cabazitaxel) was introduced.

[0370] P147: The first blood sample (P147_1) was obtained 5 years after diagnosis. The patient had multiple bone metastases characterized by CRPC. During these 5 years, he underwent radiotherapy treatment and received multiple antiandrogens. At the time of collection of the P147_1 sample, a new PSA increase occurred. cfDNA analysis revealed high-level amplifications on chromosomes Xq12 (AR) and 5q14.3. After 6 months of treatment with chemotherapy (docetaxel) and antiandrogens (abiraterone and enzalutamide), a new cfDNA sample (P147_3) was analyzed. This analysis revealed a new RET amplification on chromosome 10. The development of new lesion events and an increase in ctDNA content (P147_1ichorCNA: 52%; 147_3ichorCNA: 73%) were associated with clinical progression.

[0371] In patient P147, the time interval between prostatectomy and the first plasma sample was 56 months. Twenty months after surgery, PSA levels were found to be elevated and radiotherapy was started. Twenty-eight months after diagnosis, PSA levels were elevated again. The patient was treated with nonsteroidal anti-androgen bicalutamide for 13 months, and in the following 4 months, GnRH-analog leuprorelin was additionally administered, and later the monoclonal antibody denosumab was added due to the detection of bone metastasis. When PSA levels were elevated, plasma samples were obtained, and new high-level amplification was detected on Xq12 (AR) and 5q14.3 chromosomes. Between the collection of the first plasma sample and the second plasma sample (e.g., P147_1 and P147_2), further high-level lesion amplification occurred on chromosome 10q11.21; the time interval between these two samples was 6 months. During this period, the patient underwent chemotherapy, e.g., docetaxel.

[0372] P148: P148 was diagnosed with prostate adenocarcinoma. The patient received ADT in the period before the first blood draw (P148_1). Clinicians reported disease progression with new bone and lymph node lesions (PSA: 694.4). At the time of collection of the P148_1 sample, chemotherapy (docetaxel) was introduced. Multiple focal events were identified (MYC amplification; PTEN loss; FOXP1, RYBP, SHQ1 loss; TMPRSS2-ERG fusion), including AR amplification (the patient was previously characterized as CRPC).

[0373] Six months after the first sample collection, sample P147_3 was obtained. During this period, a large number of progressions occurred with multiple liver and bone metastases, PSA levels were 52.0ng / mL, and NSE values ​​were greater than 370ng / mL. Interestingly, AR amplification was not detected in sample P147_3, which is a feature of adenocarcinoma to neuroendocrine prostate cancer transdifferentiation (as described in Ulz ​​et al., 2016, Belic et al., 2018). After a brief response to palliative treatment with carboplatin and etoposide, disease progression occurred, and the patient died 2 months later.

[0374] Patient P148 was diagnosed with prostate adenocarcinoma. The first plasma sample was obtained 16 months after the initial diagnosis, at which time the patient's disease had clearly progressed with increased metastases to the bones and newly diagnosed lymphadenopathy. Due to disease progression (PSA: 694.41 ng / mL), the patient underwent docetaxel treatment for 7 months. A second plasma DNA analysis during this period confirmed the presence of high-level AR amplification. Five months after the last docetaxel treatment, there was massive progression with multiple liver and bone metastases, a PSA level of 52.0 ng / mL, and an NSE value greater than 370 ng / mL. The patient underwent palliative treatment with carboplatin and etoposide, with an initial partial response lasting 3 months. Thereafter, his disease progressed and he died two months later.

[0375] P166: A blood sample of patient P166 was obtained 2 years after the initial diagnosis of metastatic prostate cancer. He underwent anti-androgen bicalutamide treatment, but the disease progressed (CRPC). AR amplification on chromosome X was identified in sample P166_1 as a sign of progression. Since the patient developed CRPC, chemotherapy (docetaxel) was further introduced. After 3 months of chemotherapy, the patient showed a partial response.

[0376] P190: Five years before blood collection, the patient was diagnosed with localized prostate cancer. Two years later, he developed local progression and developed metastatic disease, primarily bone metastases, during the next two years. During these years, he was treated with the anti-androgen bicalutamide. The patient developed bone metastases and disease progression, and some of the bone metastases were characterized by a neuroendocrine phenotype. Therefore, clinicians treated him with carboplatin-based chemotherapy (carboplatin / etoposide). cfDNA analysis was performed 3 months after the start of carboplatin / etoposide treatment. At the time of this analysis, the patient showed a good response, with a PSA response of ≥50%, and normalization of the NSE value.

[0377] Tandem replicon phenotype:

[0378] P21_2: 59,849,368 reads

[0379] P111_1: 58,258,680 reads

[0380] P111_4: 61,085,342 reads

[0381] P166_1: 52,829,575 reads

[0382] Chromosome breakage (Chromosome 10):

[0383] P143_3: 111,958,416 reads (fewest PC-specific changes)

[0384] Paired comparison of plasma samples

[0385] To address whether TF accessibility remained stable over time, two samples each from patients P40, P147, and C2 were analyzed. However, under very stringent criteria, for example, by limiting the analysis to 1,000-ms TFBS, no significant differences were observed in these plasma sample pairs (control: median: 0.8404 ± 0.0196 (IQR); P40: 0.8620; P147: 0.8370; C2: 0.8719; Kendall's Tau for each) ( Fig.12 ).

[0386] Between samples P147_1 and P147_3 collected from patient P147, a new high-amplitude amplification including the RET gene appeared, while C2_7 lost the amplification including KRAS, which had been previously observed in sample C2_6. Both RET in prostate cancer and KRAS in CRC can affect the PI3K / AKT / mTOR pathway, so downstream targets such as TF CREB were investigated; however, accessibility did not differ from control plasma samples and remained unchanged. Between samples P40_1 and P40_2 from patient P40, resistance to androgen deprivation therapy (ADT) had developed, which was reflected in high-level amplification of the AR gene. However, if AR has expanded its transcriptional target repertoire, this is not obvious at the T-ARBS and N-ARBS mentioned above ( Fig.19 ). A very conservative approach of limiting the analysis to 1,000-ms TFBSs may explain why no differences between these samples were observed.

[0387] Shape of TFBS

[0388] TF-specific nucleosome coverage profiles were investigated, as some TFs displayed evenly spaced nucleosome peaks including their binding sites (e.g., PU.1 and GRHL2 in Figure 2D ), whereas others had broader troughs at their binding sites (e.g., CREM in Figure 6A ), similar to those observed for TSSs. A total of 55 TFBSs were identified with TFBSs larger than 300 bp, and from these, 26 had binding sites close to the size of a dinucleosome (312–352 bp; Figure 8D ). For these patterns, CpG islands (p = 4.2 × 10 -11 ; Mann-Whitney U test) and TSS (p = 8.5 × 10 -12 ; Mann-Whitney U test) both identified a significant increase in overlap ( Figure 8E ).

[0389] CTCF as a unique example of a TF with multiple distinct binding sites

[0390] To explore different TFBSs for the same TF, CCTC binding factor (CTCF) was used. CTCF is present at 55,000-65,000 binding sites in the mammalian genome. Of these sites, about 5,000 are ultra-conserved, about 50% are located in intergenic regions, about 15% are located near promoters, and about 40% are intragenic. In addition, chromosomes are divided into evolutionarily conserved higher-order chromosomal structures named topologically associating domains (TADs), and their boundaries are enriched with binding sites for CTCF and cohesin. In mammals, 15% of genomic CTCF binding sites are located at TAD boundaries, while the other 85% of genomic CTCF binding sites are located inside TADs.

[0391] Blood sampling and library preparation

[0392] Peripheral blood was collected from patients with metastatic prostate, breast, and colon cancer at the Department of Oncology and from anonymous healthy donors with no known chronic or malignant diseases at the Department of Hematology, Medical University of Graz. CfDNA was isolated from plasma using the QIAamp Circulating Nucleic Acid Kit (QIAGEN, Hilden, Germany) according to the manufacturer's protocol. Library preparation for WGS was performed as previously described (Heitzer et al., 2013).

[0393] Sequencing

[0394] High coverage tumor samples and controls were sequenced at 2×150 bp on an Illumina NovaSeq S4 flow cell by the Biomedical Sequencing Facility at CeMM Vienna, Austria. For control samples, an average of 435,135,450 (range: 352,904,231-556,303,420) paired-end reads were obtained. For tumor samples (P40_1, P40_2, P147_1, P147_3, P148_1, P148_3, C2_6, and C2_7), an average of 688,482,253 reads (range: 541,216,395-870,285,698) were sequenced. Additional samples were sequenced using the Illumina NextSeq platform (B7_1, B13_1, and P190_3; average sequencing yield: 296,733,931 reads; range: 181,953,656-379,733,061) and HiSeq platform (P21_2, P111_1, P111_4, P143_3, and P166_1; average sequencing yield: 52,869,911 reads; range: 41,780819-84,049,593), respectively.

[0395] Low-coverage tumor samples used to create single-entity libraries were sequenced on Illumina Next-Seq or MiSeq platforms. This generated 382,306,130 reads from 69 prostate cancer samples, 254,490,128 reads from 60 breast cancer samples, and 604,080,473 reads from 100 colon cancer samples.

[0396] Characterization of plasma samples

[0397] Some plasma samples, for example, those from patients B7 and B13 and P40, P147, and P148, were analyzed, and information on mutations, specific SCNAs, and tumor content of the plasma samples was collected based on the ichorDNA algorithm.

[0398] ETS family of oncogenic transcription factors (inspired by (Sizemore et al., 2017)). Approximately 50% of localized prostate cancers and approximately 40% of metastatic prostate cancers contain TMPRSS2-ETS fusions. Recurrent gene fusions of the 5' untranslated region of androgen-regulated TMPRSS2 with ERG (TMPRSS2-ERG gene fusions), observed in approximately 50% of primary prostate cancers, result in AR hijacking of ETS expression and transcriptional programs.

[0399] ERG has also been found to block prostate neuroendocrine cell differentiation. A possible mechanism for TMPRSS2-ERG-mediated maintenance of prostate stem and progenitor cells is through SOX9.

[0400] Given their role as transcription factors, it is not surprising that ETS factors mediate tumorigenesis through a variety of mechanisms, ranging from basic survival cues to complete epigenetic reprogramming. ETS factors also affect nucleotide, energy and steroid metabolism.

[0401] P40: mutation in BRCA1: NM_007294: Q975R; specific SCNA: TMPRSS2-ERG fusion; AR amplification in sample 2; chr12 amplification (including ARID2, HDAC7); tumor content: P40_1: 30%, P40_2: 24%. Additional lesion amplification on chromosome 15 (including SNORD (small nucleolar RNA, C / D box) genes, 16 (2x) and 19 (BRD4); P40_1ichorCNA: 30%; P40_2ichorCNA: 24%.

[0402] P147: Mutations: BRCA2: T298fs; TP53: F338I; Specific SCNAs: RET amplification in sample 3; AR amplification; BRAF amplification (7q34); PTEN loss; Tumor content: P147_1: 52%; P147_3: 73%. Additional foci amplified on chromosomes 5 (XRCC4) and 21 (RBM11); P147_1ichorCNA: 52%; 147_3ichorCNA: 73%.

[0403] P148: mutation: TP53: R213X; specific SCNA: MYC amplification; PTEN loss; FOXP1, RYBP, SHQ1 loss; TMPRSS2-ERG fusion; AR amplification (disappeared in P148_3); tumor content: P148_1: 38%; 148_3: 49%.

[0404] C2: Specific SCNA: high-level amplification on chromosome 12 (KRAS) in C2_6, not seen in C2_7; tumor content: C2_6: 18%; C2_7: 28%.

[0405] Transcription factor binding site definition

[0406] Data were downloaded from the GTRD database and individual BED files were extracted for each TF. Positions were recalculated by focusing on the reporting sites with the highest ChIP-seq signal for the meta cluster. Additional BED files were created that included only peaks supported by more than 50% of the maximum number of samples analyzed for that particular transcription factor. All BED files were then converted to hg19 (from the original hg38) using the liftOver tool provided by UCSC.

[0407] Overlapping transcription factor binding sites

[0408] To check whether the binding sites of transcription factors overlap, the regions of binding sites from GTRD (sites supported by more than 50% of the samples) were increased by 25 bp, 50 bp and 100 bp on both sides using bedtools slope. The overlap number was then calculated for each transcription factor with each other transcription factor by using bedtools intersection via pybedtools.

[0409] Single-end sequencing data preparation

[0410] To enhance nucleosome signal, sequencing reads were trimmed to remove the portion of sequencing reads associated with the adapter region. Therefore, the forward sequencing reads were trimmed to contain only bases 53-113 (which can correspond to the central 60 bp of the 166-bp fragment). The reads were then aligned to the human hg19 genome using the Burrows-Wheeler aligner (bwa), and PCR duplicates were removed using samtoolsrmdup. Average coverage was calculated by bedtools genomecov.

[0411] Paired-end sequencing data preparation

[0412] Paired-end sequencing reads were aligned to the human hg19 genome using bwa-mem and PCR duplicates were marked with Picard MarkDuplicates.

[0413] MNase-seq data preparation

[0414] Download the BAM file of the MNase-seq experiment of GM12878 from the ENCODE portal. Use pysam to trim the sequencing reads in the BAM file directly from the BAM file. Briefly, the leftmost alignment position in the BAM file was shifted 53 bp in each direction and the sequence length was adjusted to 60 bp. The coverage pattern was then calculated in the same way as the trimmed cell-free DNA sequencing data.

[0415] Coverage patterns at transcription factor binding sites

[0416] For each transcription factor in GTRD, the coverage pattern was calculated. To do this, coverage data for each region was extracted in the region ±1000bp around the defined binding site using pysam count_coverage. Coverage data at each site was normalized by the regional copy number variation and the average coverage. For each position around the TFBS, the coverage was averaged and a 95% confidence interval was calculated. If more than 100,000 sites were defined for a transcription factor, 100,000 sites were randomly selected for analysis.

[0417] Insert size around transcription factor binding sites

[0418] To assess whether there was a bias in fragment size around transcription factor binding sites, insert size data from a paired-end analysis were used. Each position from -1000 bp to 1000 bp from the binding site was traversed and (single-end) sequencing reads with the center 3 bp around the midpoint located at this position were obtained using pysam. Similarly, paired-end alignments were obtained from the same samples and insert size information was assigned to each read. All insert sizes at a specific position relative to the TFBS were then summarized and 1000 data points were sampled and plotted for each position within the range of -1000 bp to 1000 bp from the TFBS.

[0419] Measuring transcription factor binding site size

[0420] To measure the size of transcription factor binding sites, the respective coverage patterns were smoothed using a third-order Savitzky-Golay filter (window size: 31). Peaks were identified by searching for data points that were larger than 20 adjacent data points on either side. Peaks were removed if they were within 50 bp of the center of the putative binding site. The distance between the closest peaks to the binding site peak was designated as the transcription factor binding site size.

[0421] Since binding site estimation is only reasonable if nucleosome synchronization is detectable, the signal is filtered by various criteria:

[0422] High frequency signal amplitude is greater than 0.1

[0423] The average normalized coverage of the central 100 bp is less than 1

[0424] · Peaks less than 15

[0425] The median distance between peaks is greater than 150 bp

[0426] The binding site set includes more than 500 sites

[0427] A total of 228 binding site sets passed these filters and were used for binding site estimation. Measuring transcription factor accessibility using Savitzky-Golay filters

[0428] Two different signals make up the coverage pattern, and the signals of the two different frequencies are extracted into lower and higher range frequencies. The lower range frequency data is extracted through a Savitzky-Golay filter (third order polynomial and window size of 1001). The high frequency signal is extracted through a different Savitzky-Golay filter (third order polynomial and window size of 51). The high frequency signal is then normalized by dividing the result by the low frequency signal. The data range of the high frequency signal is then recorded. Since the coverage spectrum of transcription factors with few described binding sites has more inherent noise, LOESS (local weighted smoothing) is performed on the signal range and the number of described binding sites. The range value is corrected by the smoothed LOESS, and the level of the adjusted range is calculated. Measuring transcription factor accessibility using wavelet transform

[0429] As another method for measuring transcription factor accessibility, wavelet transform was applied by using the R package "WaveletComp". For each signal, peak values ​​were recorded along the segment between 2bp and 512bp in the power spectrum. The highest peak in the range of 135bp to 235bp (185bp ± 50bp) was used to reconstruct the high-frequency nucleosome signal of denoising in this specific segment. In addition, any residual baseline was removed using the detrending of the original data sequence. Three parameters of the reconstructed signal were analyzed: the maximum amplitude of the signal, the sum of the signal power (amplitude squared) and the absolute amplitude sum of 2000bp around the transcription factor binding site.

[0430] To compare tumor to normal samples, the mean and standard deviation of the respective parameter was recorded in the normal samples for each transcription factor, and the Z score was calculated by taking the respective parameter in the cancer sample, subtracting the mean of the normal value, and then dividing by the standard deviation.

[0431] Comparison of tumor and control samples

[0432] To compare tumor samples to control samples, the rank of each transcription factor in the regulation range was compared. The rank difference between the tumor sample and each control sample was calculated, and the average rank difference was recorded. Furthermore, a z score was calculated for each transcription factor based on the accessibility rank by taking the corresponding rank, subtracting the average rank of the control samples, and then dividing by the standard deviation of the rank of the transcription factor for the control samples.

[0433] DNase hypersensitivity data analysis

[0434] BAM files from DNase hypersensitivity experiments were downloaded from the ENCODE database for GM12878, LNCaP, and HCT116 cell lines. Binding site regions for transcription factors were increased by 25 bp on either side using bedtools slope. Coverage at individual binding sites was extracted using mosdepth and normalized by 1 million mapped reads / sample.

[0435] Somatic copy number alteration (SCNA) analysis

[0436] For control data, paired-end alignments were subsampled using samtools view to include only 2% of the initial alignments and converted to FastQ using samtools fastq. For cancer samples, separate low-coverage whole-genome sequencing was performed. Plasma-Seq was applied to the subsampled FastQ files and low-coverage data of cancer samples separately. Briefly, sequencing reads were aligned to the human hg19 genome and counted in pre-specified bins. The size of the bin was determined by the number of theoretically mappable positions to account for differences in mappability across the genome. Read counts were normalized to the total amount of reads, and the GC content of the bins was corrected by performing LOESS smoothing on the GC profile. In addition, the corrected read counts were normalized by the average read count / bin of non-cancer controls to control for additional positional variations.

[0437] Accessibility scores can accurately infer TF binding from cfDNA

[0438] Samples from healthy donors showed that TF did not exceed the ±5z score threshold (Figure 29A); however, very different patterns were observed in samples from cancer patients. For example, in the prostate sample P40_1 from patient P40, in addition to GRHL2 and FOXA1, TFs with accessibility higher than the +5z score threshold were also included, which synergized with nuclear hormone receptors and prostate lineage-specific TFs HOXB13, AR and NKX3-1 in endocrine-driven prostate and breast tumors (Figure 29B). In contrast, hematopoietic TFs such as LYL1, SPIB and EVI1 (transcription regulator ecotropic viral integration site 1 (Figure 29B)) have low accessibility. These results are very consistent with the TF grading based on ATAC-seq data. In breast cancer samples B7 and B13, consistent with the ATAC-seq data of GRHL2, FOXA1 and ZNF121 (zinc finger protein), the accessibility detected increased, which is related to cell proliferation regulation and the development of breast cancer.

[0439] In the sample of colon cancer patient C2, it was unexpectedly observed that ATAC-seq data had graded EVX2, a TF not closely associated with cancer, which was the most accessible in COAD. Indeed, in this analysis, EVX2 had the highest accessibility grade (Figure 29C), and the nucleosome position map showed that EVX2 had greatly improved accessibility (Figure 29D). Consistent with the ATAC-seq data, increased accessibility was also observed for TFs HNF4A, GRHL2, DLX2, HNF4G, and HNF1A (Figure 29D).

[0440] Furthermore, as predicted by evaluation of ATAC-seq data, the accessibility of hematopoietic-associated TFs, such as LYL1, TAL1 (SCL / TAL1 (stem cell leukemia / T-cell acute lymphoblastic leukemia [T-ALL] 1), EVI1, TBX21 (T-bet), and PU.1, was decreased in all tumor samples ( Figures 29B-29C ). To further confirm the robustness and reproducibility of lineage-specific TFs in cfDNA, we showed that the expression of lineage-specific TFs in multiple cfDNA sample pools generated by shallow coverage (<0.2x) was significantly higher in large tumor samples ( Figures 29B-29C ). TFs with increased accessibility in most samples had increased accessibility scores, while others reached average scores. To this end, cfDNA samples were pooled separately for cases of prostate cancer (n=69), colon cancer (n=100), and breast cancer (n=60). When the analysis was repeated, the epithelial TF GRHL2 and hematopoietic TFs reiterated their patterns of increased and decreased accessibility in the three epithelial lineages, respectively. In the colon cfDNA pool, EVX2, DLX2, HNF1A, HNF4A, and HNF4G, and the TFs AR and HOXB13 in the prostate cancer cfDNA pool had increased accessibility, while FOXA1 exceeded the >5z score threshold in both prostate and breast pools. This confirms that TF accessibility estimates from ATAC-seq data can be reliably inferred from plasma DNA nucleosome mapping.

[0441] Figure 29E provides a bar graph of the overall z-score plot for the combined breast, prostate, and colon cancer pools. The left panel shows TFs with increased accessibility in at least one tumor entity; the right panel summarizes the accessibility of hematopoietic-related TFs.

[0442] Figures 30A-30B provide graphs showing the plasma resolution limit and early cancer detection based on TF. Figure 30A provides a graph showing the accessibility comparison of selected TFs in COAD cohort subsamples based on their tumor grade. Figure 30B provides a graph showing the logistic regression of all 504 TFs from samples of stage I (left sub-graph) and stage II (right sub-graph) colon cancer cohorts, respectively. All presented results are cross-validated test settings.

[0443] Although some examples of methods and systems have been shown and described herein, those skilled in the art will recognize that these are provided by way of example only and are not intended to be limited by the specification. Those skilled in the art will now think of various variations, changes and substitutions without departing from the scope described herein. In addition, it should be understood that all aspects of the described methods and systems are not limited to the specific descriptions, configurations or relative proportions that rely on various conditions and variables set forth herein, and that the specification is intended to include such alternatives, modifications, variations or equivalents.

Claims

1. A non-transitory computer readable medium containing machine executable code, which when executed by one or more computer processors implements a method for detecting the presence or absence of cancer in a human subject, the method comprising: (a) providing a plurality of sequence reads obtained by whole genome sequencing of cell-free deoxyribonucleic acid (cfDNA) obtained from or derived from the human subject; (b) aligning each of the plurality of sequence reads to a human reference genome to provide a plurality of aligned sequence reads; (c) for each of a plurality of binding sites for a transcription factor, selecting genomic regions in the plurality of aligned sequence reads that flank the binding site to provide a plurality of flanking genomic regions, wherein each of the plurality of binding sites exhibits differential accessibility to a target enhancer of the transcription factor in a first population of people suffering from the cancer compared to a second population of people not suffering from the cancer, wherein each of the plurality of binding sites is different; (d) for each of the plurality of flanking genomic regions, generating a read depth coverage pattern for a binding site of the transcription factor using at least a portion of the plurality of aligned sequence reads, thereby providing a plurality of read depth coverage patterns for the plurality of binding sites of the transcription factor, respectively; (e) processing the plurality of read depth coverage patterns to provide a signal, wherein the signal is indicative of binding site accessibility of the plurality of binding sites of the transcription factor, (f) splitting each of the plurality of read depth coverage patterns into a low-frequency signal and a high-frequency signal using a low-pass filter, wherein the high-frequency signal is a measure of binding site accessibility for a given binding site, and determining an accessibility score for each of the plurality of binding sites for the transcription factor based at least in part on the high-frequency signal; and (g) detecting the presence or absence of the cancer in the human subject based at least in part on a deviation of the accessibility score from a reference accessibility score.

2. The computer-readable medium of claim 1, wherein the method further determines the accessibility scores of at least 2 or at least 5 or at least 10 or at least 15 or at least 20 or at least 25 transcription factor binding sites and inputs them into a machine learning model to train a classifier capable of distinguishing between healthy subjects and cancer patients, cancer progressors and non-progressors, multiple cancer subtypes, multiple cancer stages, cancer treatment responders and non-responders, or any combination thereof.

3. The computer-readable medium of claim 1, wherein processing the plurality of read depth coverage patterns comprises using a Savitzky-Golay filter. The computer readable medium of claim 1 , wherein the transcription factor is an epithelial transcription factor.

5. The computer-readable medium of claim 1, wherein the transcription factor is a cancer-specific transcription factor.

6. The computer-readable medium of claim 5, wherein the transcription factor is selected from the group consisting of GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, FOXA1, HNF-1A, HNF-4A, HNF-4G, and DLX-2.

7. The computer-readable medium of claim 1, wherein (g) further comprises identifying a signal of higher accessibility of the transcription factor.

8. The computer-readable medium of claim 1, wherein the method detects the presence or absence of the cancer in the human subject with an accuracy of at least 70%.

9. The computer-readable medium of claim 1, wherein the method detects the presence or absence of the cancer in the human subject with an accuracy of at least 80%.

10. The computer-readable medium of claim 1, wherein the method detects the presence or absence of the cancer in the human subject with an accuracy of at least 90%.

11. The computer-readable medium of claim 1, wherein the method detects the presence of the cancer in the human subject with a sensitivity of at least 70%.

12. The computer-readable medium of claim 1, wherein the method detects the presence of the cancer in the human subject with a sensitivity of at least 80%.

13. The computer-readable medium of claim 1, wherein the method detects the presence of the cancer in the human subject with a sensitivity of at least 90%.

14. The computer readable medium of claim 1, wherein the method detects the absence of the cancer in the human subject with a specificity of at least 70%.

15. The computer readable medium of claim 1, wherein the method detects the absence of the cancer in the human subject with a specificity of at least 80%.

16. The computer readable medium of claim 1, wherein the method detects the absence of the cancer in the human subject with a specificity of at least 90%.

17. The computer-readable medium of claim 1, wherein the method detects the presence of the cancer in the human subject with a positive predictive value (PPV) of at least 70%.

18. The computer-readable medium of claim 1, wherein the method detects the presence of the cancer in the human subject with a positive predictive value (PPV) of at least 80%.

19. The computer-readable medium of claim 1, wherein the method detects the presence of the cancer in the human subject with a positive predictive value (PPV) of at least 90%.

20. The computer readable medium of claim 1, wherein the method detects the absence of the cancer in the human subject with a negative predictive value (NPV) of at least 70%.

21. The computer readable medium of claim 1, wherein the method detects the absence of the cancer in the human subject with a negative predictive value (NPV) of at least 80%.

22. The computer-readable medium of claim 1, wherein the method detects the absence of the cancer in the human subject with a negative predictive value (NPV) of at least 90%.

23. The computer-readable medium of claim 1, wherein the method detects the presence or absence of the cancer in the human subject with an area under the receiver operating characteristic (AUROC) of at least 0.

70.

24. The computer-readable medium of claim 1, wherein the method detects the presence or absence of the cancer in the human subject with an area under the receiver operating characteristic (AUROC) of at least 0.

80.

25. The computer-readable medium of claim 1, wherein the method detects the presence or absence of the cancer in the human subject with an area under the receiver operating characteristic (AUROC) of at least 0.

90.

26. The computer-readable medium of claim 1, wherein the method comprises applying a trained classifier to the signal to detect the presence or absence of the cancer in the human subject.

27. The computer-readable medium of claim 1, wherein the method comprises applying a trained classifier to the accessibility scores of the binding sites of the transcription factors to detect the presence or absence of the cancer in the human subject.

28. The computer-readable medium of claim 27, wherein the trained classifier comprises a trained machine learning classifier.

29. The computer-readable medium of claim 28, wherein the trained machine learning classifier comprises a supervised machine learning algorithm.

30. The computer-readable medium of claim 29, wherein the supervised machine learning algorithm comprises one or more of: regression, support vector machine, tree-based method, neural network, and random forest.

31. A non-transitory computer readable medium comprising machine executable code which, when executed by one or more computer processors, implements a method for monitoring progression or regression of cancer in a human subject, the method comprising: (a) providing a plurality of first sequence reads obtained by whole genome sequencing of cell-free deoxyribonucleic acid (cfDNA) obtained from or derived from the human subject at a first time and a plurality of second sequence reads obtained by whole genome sequencing of cfDNA obtained from or derived from the human subject at a second time later than the first time; (b) aligning each of the plurality of first sequence reads and the plurality of second sequence reads to a human reference genome to provide a first plurality of aligned sequence reads and a second plurality of aligned sequence reads; (c) for each of a plurality of binding sites for a transcription factor, selecting genomic regions flanking the binding site in the first plurality of aligned sequence reads and the second plurality of aligned sequence reads to provide a first plurality of flanking genomic regions and a second plurality of flanking genomic regions, wherein each of the plurality of binding sites exhibits differential accessibility to a target enhancer of the transcription factor in a first population of people suffering from the cancer compared to a second population of people not suffering from the cancer, wherein each of the plurality of binding sites is different; (d) for each of the first plurality of flanking genomic regions and the second plurality of flanking genomic regions, generating a first read depth coverage pattern for a binding site of the transcription factor using at least a portion of the first plurality of aligned sequence reads, and generating a second read depth coverage pattern for the binding site of the transcription factor using at least a portion of the second plurality of aligned sequence reads, thereby providing first and second plurality of read depth coverage patterns for the plurality of binding sites of the transcription factor, respectively; (e) processing the first plurality of read depth coverage patterns to provide a first signal, and processing the second plurality of read depth coverage patterns to provide a second signal, wherein the first signal and the second signal characterize binding site accessibility of the plurality of binding sites of the transcription factor; (f) using a low pass filter to split each of the first plurality of read depth coverage patterns into a first low frequency signal and a first high frequency signal, and split each of the second plurality of read depth coverage patterns into a second low frequency signal and a second high frequency signal, wherein the first high frequency signal and the second high frequency signal are measures of binding site accessibility for a given binding site; (g) determining an accessibility score for each of the plurality of binding sites of the transcription factor based at least in part on the first high frequency signal and the second high frequency signal; and (h) monitoring the progression or regression of the cancer in the human subject based at least in part on a deviation of the accessibility score from a reference accessibility score.

32. The computer-readable medium of claim 31 , wherein processing the first plurality of read depth coverage patterns comprises using a Savitzky-Golay filter.

33. The computer readable medium of claim 31, wherein the transcription factor is a cancer-specific transcription factor.

34. The computer-readable medium of claim 33, wherein the transcription factor is selected from the group consisting of GRH-L2, ASH-2, HOX-B13, EVX2, PU.1, Lyl-1, Spi-B, FOXA1, HNF-1a, HNF-4a, HNF-4g, and DLX-2.

35. The computer-readable medium of claim 31 , further comprising adjusting a treatment regimen for the cancer in the human subject based on the monitored progression or regression of the cancer in the human subject.

36. A system for monitoring the progression or regression of cancer in a human subject, the system comprising: a database comprising a plurality of first sequence reads obtained by whole genome sequencing of cell-free deoxyribonucleic acid (cfDNA) obtained from or derived from the human subject at a first time, and a plurality of second sequence reads obtained by whole genome sequencing of cfDNA obtained from or derived from the human subject at a second time later than the first time, wherein the plurality of first sequence reads and the plurality of second sequence reads are aligned to a human reference genome to provide a first plurality of aligned sequence reads and a second plurality of aligned sequence reads; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually or collectively programmed to: (a) for each of a plurality of binding sites for a transcription factor, selecting genomic regions flanking the binding site in the first plurality of aligned sequence reads and the second plurality of aligned sequence reads to provide a first plurality of flanking genomic regions and a second plurality of flanking genomic regions, wherein each of the plurality of binding sites exhibits differential accessibility to a target enhancer of the transcription factor in a first population of people suffering from the cancer compared to a second population of people not suffering from the cancer, wherein each of the plurality of binding sites is different; (b) for each of the first plurality of flanking genomic regions and the second plurality of flanking genomic regions, using at least a portion of the first plurality of aligned sequence reads to generate a first read depth coverage pattern for the binding site of the transcription factor, and using at least a portion of the second plurality of aligned sequence reads to generate a second read depth coverage pattern for the binding site of the transcription factor, thereby providing first and second plurality of read depth coverage patterns for the plurality of binding sites of the transcription factor, respectively; (c) processing the first plurality of read depth coverage patterns to provide a first signal, and processing the second plurality of read depth coverage patterns to provide a second signal, wherein the first signal and the second signal characterize binding site accessibility of the plurality of binding sites of the transcription factor; (d) using a low pass filter to split each of the first plurality of read depth coverage patterns into a first low frequency signal and a first high frequency signal, and each of the second plurality of read depth coverage patterns into a second low frequency signal and a second high frequency signal, wherein the first high frequency signal and the second high frequency signal are measures of binding site accessibility for a given binding site; (e) determining an accessibility score for each of the plurality of binding sites of the transcription factor based at least in part on the first high frequency signal and the second high frequency signal; and (f) monitoring the progression or regression of the cancer in the human subject based at least in part on a deviation of the accessibility score from a reference accessibility score.

37. A computer-implemented method for determining tumor-specific transcription factor binding site (TFBS) patterns, the method comprising: (a) providing a plurality of first sequence reads obtained by whole genome sequencing of cell-free deoxyribonucleic acid (cfDNA) obtained from or derived from a human subject at a first time and a plurality of second sequence reads obtained by whole genome sequencing of cfDNA obtained from or derived from the human subject at a second time later than the first time; (b) aligning each of the plurality of first sequence reads and the plurality of second sequence reads to a human reference genome to provide a first plurality of aligned sequence reads and a second plurality of aligned sequence reads; (c) for each of a plurality of binding sites for a transcription factor, selecting genomic regions flanking the binding site in the first plurality of aligned sequence reads and the second plurality of aligned sequence reads to provide a first plurality of flanking genomic regions and a second plurality of flanking genomic regions, wherein each of the plurality of binding sites exhibits differential accessibility to a target enhancer of the transcription factor in a first population of people suffering from the cancer compared to a second population of people not suffering from the cancer, wherein each of the plurality of binding sites is different; (d) for each of the first plurality of flanking genomic regions and the second plurality of flanking genomic regions, generating a first read depth coverage pattern for a binding site of the transcription factor using at least a portion of the first plurality of aligned sequence reads, and generating a second read depth coverage pattern for the binding site of the transcription factor using at least a portion of the second plurality of aligned sequence reads, thereby providing first and second plurality of read depth coverage patterns for the plurality of binding sites of the transcription factor, respectively; (e) processing the first plurality of read depth coverage patterns to provide a first signal, and processing the second plurality of read depth coverage patterns to provide a second signal, wherein the first signal and the second signal characterize binding site accessibility of the plurality of binding sites of the transcription factor; (f) using a low pass filter to split each of the first plurality of read depth coverage patterns into a first low frequency signal and a first high frequency signal, and split each of the second plurality of read depth coverage patterns into a second low frequency signal and a second high frequency signal, wherein the first high frequency signal and the second high frequency signal are measures of binding site accessibility for a given binding site; (g) determining an accessibility score for each of the plurality of binding sites of the transcription factor based at least in part on the first high frequency signal and the second high frequency signal; and (h) determining the tumor-specific TFBS pattern in the human subject based at least in part on a deviation of the accessibility score from a reference accessibility score.

38. A non-transitory computer readable medium comprising machine executable code which, when executed by one or more computer processors, implements a method for determining a tumor-specific transcription factor binding site (TFBS) pattern, the method comprising: (a) providing a plurality of first sequence reads obtained by whole genome sequencing of cell-free deoxyribonucleic acid (cfDNA) obtained from or derived from a human subject at a first time and a plurality of second sequence reads obtained by whole genome sequencing of cfDNA obtained from or derived from the human subject at a second time later than the first time; (b) aligning each of the plurality of first sequence reads and the plurality of second sequence reads to a human reference genome to provide a first plurality of aligned sequence reads and a second plurality of aligned sequence reads; (c) for each of a plurality of binding sites for a transcription factor, selecting genomic regions flanking the binding site in the first plurality of aligned sequence reads and the second plurality of aligned sequence reads to provide a first plurality of flanking genomic regions and a second plurality of flanking genomic regions, wherein each of the plurality of binding sites exhibits differential accessibility to a target enhancer of the transcription factor in a first population of people suffering from the cancer compared to a second population of people not suffering from the cancer, wherein each of the plurality of binding sites is different; (d) for each of the first plurality of flanking genomic regions and the second plurality of flanking genomic regions, generating a first read depth coverage pattern for a binding site of the transcription factor using at least a portion of the first plurality of aligned sequence reads, and generating a second read depth coverage pattern for the binding site of the transcription factor using at least a portion of the second plurality of aligned sequence reads, thereby providing first and second plurality of read depth coverage patterns for the plurality of binding sites of the transcription factor, respectively; (e) processing the first plurality of read depth coverage patterns to provide a first signal, and processing the second plurality of read depth coverage patterns to provide a second signal, wherein the first signal and the second signal characterize binding site accessibility of the plurality of binding sites of the transcription factor; (f) using a low pass filter to split each of the first plurality of read depth coverage patterns into a first low frequency signal and a first high frequency signal, and split each of the second plurality of read depth coverage patterns into a second low frequency signal and a second high frequency signal, wherein the first high frequency signal and the second high frequency signal are measures of binding site accessibility for a given binding site; (g) determining an accessibility score for each of the plurality of binding sites of the transcription factor based at least in part on the first high frequency signal and the second high frequency signal; and (h) determining the tumor-specific TFBS pattern in the human subject based at least in part on a deviation of the accessibility score from a reference accessibility score.

39. A system for determining tumor-specific transcription factor binding site (TFBS) patterns, the system comprising: A database comprising a plurality of first sequence reads obtained by whole genome sequencing of cell-free deoxyribonucleic acid (cfDNA) obtained from or derived from a human subject at a first time, and a plurality of second sequence reads obtained by whole genome sequencing of cfDNA obtained from or derived from the human subject at a second time later than the first time, wherein the plurality of first sequence reads and the plurality of second sequence reads are aligned to a human reference genome to provide a first plurality of aligned sequence reads and a second plurality of aligned sequence reads; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually or collectively programmed to: (a) for each of a plurality of binding sites for a transcription factor, selecting genomic regions flanking the binding site in the first plurality of aligned sequence reads and the second plurality of aligned sequence reads to provide a first plurality of flanking genomic regions and a second plurality of flanking genomic regions, wherein each of the plurality of binding sites exhibits differential accessibility to a target enhancer of the transcription factor in a first population of people suffering from the cancer compared to a second population of people not suffering from the cancer, wherein each of the plurality of binding sites is different; (b) for each of the first plurality of flanking genomic regions and the second plurality of flanking genomic regions, using at least a portion of the first plurality of aligned sequence reads to generate a first read depth coverage pattern for the binding site of the transcription factor, and using at least a portion of the second plurality of aligned sequence reads to generate a second read depth coverage pattern for the binding site of the transcription factor, thereby providing first and second plurality of read depth coverage patterns for the plurality of binding sites of the transcription factor, respectively; (c) processing the first plurality of read depth coverage patterns to provide a first signal, and processing the second plurality of read depth coverage patterns to provide a second signal, wherein the first signal and the second signal characterize binding site accessibility of the plurality of binding sites of the transcription factor; (d) using a low pass filter to split each of the first plurality of read depth coverage patterns into a first low frequency signal and a first high frequency signal, and each of the second plurality of read depth coverage patterns into a second low frequency signal and a second high frequency signal, wherein the first high frequency signal and the second high frequency signal are measures of binding site accessibility for a given binding site; and (e) determining an accessibility score for each of the plurality of binding sites of the transcription factor based at least in part on the first high-frequency signal and the second high-frequency signal, thereby determining the tumor-specific TFBS pattern based at least in part on a deviation of the accessibility score from a reference accessibility score.

Images