System for cancer screening

Abstract

The present application provides a system for cancer screening, comprising: a data acquisition module for acquiring methylation levels and cfDNA related features of a target region of a subject; and a cancer calculation module for predicting whether the subject suffers from cancer based on the methylation levels and cfDNA related features acquired in the data acquisition module. The system of the present application comprehensively utilizes methylation levels and cfDNA feature related indicators to construct a cancer non-invasive screening method that can significantly reduce the cost of cancer screening and improve the accuracy of screening, and has very high sensitivity and specificity.

Description

TECHNICAL FIELD

[0001] The present application relates to a system for cancer screening. BACKGROUND

[0002] Lung cancer is one of the highest incidence and mortality rates of cancer in the world, and the 5-year survival rate is less than 20%. In China, the incidence and mortality rates of lung cancer are also the highest. Among them, the main reason is that lung cancer is usually diagnosed at an advanced stage, and the treatment is far less effective than early stage. The conventional screening methods for lung cancer include low-dose spiral CT (LDCT), and some protein markers, such as carcinoembryonic antigen (CEA), squamous cell carcinoma antigen (SCC), neuron-specific enolase (NSE) and the like. However, these conventional methods are uneven in sensitivity and specificity. At present, DNA methylation has been proved to have tissue characteristics, which can be used for early cancer detection, and can be tracked to the primary tumor site according to the methylation characteristics of circulating tumor DNA (ctDNA).

[0003] Liquid biopsy is to realize early screening, molecular typing, prognosis, drug guidance and recurrence detection of cancer by analyzing cancer components in blood. As a new technology of precision medicine, liquid biopsy plays an increasingly important role in tumor diagnosis and treatment, because it can quantitatively detect tumor cells and DNA directly related to tumors, and has the characteristics of non-invasiveness, convenient sampling and real-time monitoring.

[0004] At present, some studies have proved that cfDNA molecules are not random fragments. The death of specific tissue cells will cause the distribution of DNA fragments of the affected tissue to change, which can be tracked to the primary tumor site for early diagnosis of cancer. In addition, the cfDNA fragment group can comprehensively represent the genomic and chromatin characteristics, so as to identify a large number of changes in circulating tumor derivatives.

[0005] Based on whole genome bisulfite sequencing (WGBS), both methylation level information and cfDNA fragment group can be detected, so the combination of methylation level and fragment group characteristics based on WGBS data provides a new strategy for tumor detection. SUMMARY

[0006] In view of the problems of the prior art, the purpose of the present application is to provide a system for cancer screening.

[0007] Specifically, the technical scheme is as follows:

[0008] 1. A system for cancer screening, comprising:

[0009] a data acquisition module for acquiring methylation level and cfDNA related characteristics of a target region of a subject; and

[0010] a cancer computing module configured to predict whether the subject has cancer based on the methylation level and the cfDNA related features obtained from the data acquisition module.

[0011] 2. The system of item 1, wherein,

[0012] the data acquisition module comprises a sequencing module, a methylation level analysis module and a cfDNA related features extraction module,

[0013] the sequencing module is configured to perform whole genome sequencing of the cfDNA of the subject,

[0014] the methylation level analysis module is configured to analyze the methylation level of the target region from the sequencing data obtained from the sequencing module, and the cfDNA related features extraction module is configured to extract the cfDNA related features from the sequencing data obtained from the sequencing module.

[0015] 3. The system of item 1, wherein,

[0016] the target region comprises any one or two or more of the following regions:

[0017] chromosome 1, 151445000-151450000,

[0018] chromosome 2, 191183500-191188500,

[0019] chromosome 2, 191184000-191189000,

[0020] chromosome 4, 68566500-68571500, or

[0021] chromosome 11, 30601500-30606500.

[0022] 4. The system of item 1, wherein,

[0023] the methylation level of the target region is calculated based on the methylation level of each CG site of the target region, wherein the methylation level of the CG site is the ratio of the number of methylated cytosines detected in all detected sequence results of the site to the sum of the number of methylated cytosines and the number of unmethylated cytosines.

[0024] 5. The system of item 2, wherein,

[0025] the cfDNA related features comprise:

[0026] the average of all fragment sizes obtained from the cfDNA sequencing data,

[0027] the mode of all fragment sizes obtained from the cfDNA sequencing data of the subject,

[0028] the average coverage obtained from the cfDNA sequencing data of the subject,

[0029] a correlation coefficient between the pre-stored average vector of short fragment coverage of corresponding intervals of healthy people and the short fragment coverage obtained from the cfDNA sequencing data of the subject,

[0030] a correlation coefficient between the pre-stored average vector of long fragment coverage of corresponding intervals of healthy people and the long fragment coverage obtained from the cfDNA sequencing data of the subject,

[0031] and a correlation coefficient between the pre-stored average vector of short-long fragment ratio of corresponding intervals of healthy people and the short-long fragment ratio obtained from the cfDNA sequencing data of the subject.

[0032] 6. The system according to item 5, wherein,

[0033] the pre-stored average vector of short fragment coverage of corresponding intervals of healthy people refers to the average of short fragment coverage of corresponding intervals of healthy people based on the sequencing data of cfDNA of known healthy people, which is provided to the data acquisition module for calculation;

[0034] the pre-stored average vector of long fragment coverage of corresponding intervals of healthy people refers to the average of long fragment coverage of corresponding intervals of healthy people based on the sequencing data of cfDNA of known healthy people, which is provided to the data acquisition module for calculation;

[0035] the pre-stored average vector of short-long fragment ratio of corresponding intervals of healthy people refers to the average of short-long fragment ratio of corresponding intervals of healthy people based on the sequencing data of cfDNA of known healthy people, which is provided to the data acquisition module for calculation.

[0036] 7. The system according to item 5, wherein,

[0037] the average of all fragment sizes obtained from the cfDNA sequencing data of the subject refers to the ratio of the sum of sizes of all fragments to the number of all fragments obtained from the cfDNA sequencing data of the subject;

[0038] the average coverage obtained from the cfDNA sequencing data of the subject refers to the number of all fragments obtained from the cfDNA sequencing data of the subject;

[0039] the short fragment coverage obtained from the cfDNA sequencing data of the subject refers to the number of short fragments obtained from the cfDNA sequencing data of the subject;

[0040] The long fragment coverage obtained from the cfDNA sequencing data refers to the number of long fragments obtained from the cfDNA sequencing data of the subject.

[0041] 8. The system of item 5, wherein,

[0042] The length of the long fragments is 201-320 bp, and the length of the short fragments is 150-200 bp.

[0043] 9. The system of item 2, wherein,

[0044] The cfDNA sequencing data is cfDNA sequencing data after removal of low-quality sequencing fragments.

[0045] 10. The system of item 9, wherein,

[0046] The cfDNA sequencing data is sequencing data after removal of low-quality sequencing fragments and further removal of sequencing data in a low mapping rate interval.

[0047] 11. The system of item 1, wherein,

[0048] In the cancer calculation module, a model based on data fitting of methylation levels and cfDNA-related features of known samples is pre-stored for predicting whether the subject is suffering from cancer,

[0049] The model is obtained based on data of methylation levels and cfDNA-related features of known samples by a gradient boosting tree model with 5-fold cross-validation selection.

[0050] 12. The system of item 1, wherein,

[0051] The system further comprises a bisulfite treatment module for bisulfite treatment of cfDNA of the subject.

[0052] 1. A system for cancer screening, comprising:

[0053] a data acquisition module for acquiring methylation levels and cfDNA-related features of a target region of a subject; and

[0054] a cancer calculation module for predicting whether the subject is suffering from cancer based on the methylation levels and cfDNA-related features acquired in the data acquisition module.

[0055] 2. The system of item 1, wherein,

[0056] The data acquisition module comprises a sequencing module, a methylation level analysis module, and a cfDNA-related feature extraction module,

[0057] the sequencing module is configured to perform whole genome sequencing on the cfDNA of the subject,

[0058] the methylation level analysis module is configured to analyze the methylation level of the target region from the sequencing data obtained from the sequencing module, and the cfDNA related feature extraction module is configured to extract the cfDNA sequencing data related features from the sequencing data obtained from the sequencing module.

[0059] 3. The system of item 1, wherein,

[0060] the target region comprises any one or more of the following regions:

[0061] chromosome 1, 151445000-151450000,

[0062] chromosome 2, 191183500-191188500,

[0063] chromosome 2, 191184000-191189000,

[0064] chromosome 4, 68566500-68571500, or

[0065] chromosome 11, 30601500-30606500.

[0066] 4. The system of item 1, wherein,

[0067] the methylation level of the target region is calculated based on the methylation level of each CG site in the target region, wherein the methylation level of the CG site is the ratio of the number of methylated cytosines detected at the site to the total number of methylated cytosines and non-methylated cytosines detected at the site in all sequence results of the site.

[0068] 5. The system of item 2, wherein,

[0069] the cfDNA related features comprise:

[0070] the average of all fragment sizes obtained from the cfDNA sequencing data,

[0071] the mode of all fragment sizes obtained from the cfDNA sequencing data,

[0072] the average coverage obtained from the cfDNA sequencing data,

[0073] the correlation coefficient between the pre-stored average short fragment coverage vector of the corresponding interval of healthy people and the short fragment coverage obtained from the cfDNA sequencing data of the subject,

[0074] a correlation coefficient between a pre-stored average vector of long fragment coverage in corresponding intervals of healthy people and long fragment coverage obtained from cfDNA sequencing data of the subject,

[0075] and a correlation coefficient between a pre-stored average vector of short-long fragment ratio in corresponding intervals of healthy people and short-long fragment ratio obtained from cfDNA sequencing data of the subject.

[0076] 6. The system of item 5, wherein,

[0077] The pre-stored average vector of short fragment coverage in corresponding intervals of healthy people refers to an average of short fragment coverage in corresponding intervals of known healthy people based on sequencing data of cfDNA of the known healthy people, which is provided to the data acquisition module for calculation.

[0078] The pre-stored average vector of long fragment coverage in corresponding intervals of healthy people refers to an average of long fragment coverage in corresponding intervals of known healthy people based on sequencing data of cfDNA of the known healthy people, which is provided to the data acquisition module for calculation.

[0079] The pre-stored average vector of short-long fragment ratio in corresponding intervals of healthy people refers to an average of short-long fragment ratio in corresponding intervals of known healthy people based on sequencing data of cfDNA of the known healthy people, which is provided to the data acquisition module for calculation.

[0080] 7. The system of item 5, wherein,

[0081] The average of all fragment sizes obtained from cfDNA sequencing data refers to a ratio of the sum of sizes of all fragments obtained from cfDNA sequencing data of the subject to the number of all fragments.

[0082] The average coverage obtained from cfDNA sequencing data refers to the number of all fragments obtained from cfDNA sequencing data of the subject.

[0083] The short fragment coverage obtained from cfDNA sequencing data refers to the number of short fragments obtained from cfDNA sequencing data of the subject.

[0084] The long fragment coverage obtained from cfDNA sequencing data refers to the number of long fragments obtained from cfDNA sequencing data of the subject.

[0085] 8. The system of item 5, wherein,

[0086] The length of the long fragment is 201-320 bp, and the length of the short fragment is 150-200 bp.

[0087] 9. The system of item 2, wherein,

[0088] The cfDNA sequencing data is the cfDNA sequencing data after removing low quality sequencing fragments.

[0089] 10. The system of item 9, wherein,

[0090] The cfDNA sequencing data is the sequencing data after removing low quality sequencing fragments and further removing sequencing data in low mapping rate intervals.

[0091] 11. The system of item 1, wherein,

[0092] In the cancer computing module, a model based on data fitting of methylation levels of known samples and cfDNA related features is pre-stored for predicting whether the subject suffers from cancer,

[0093] The model is based on data of methylation levels of known samples and cfDNA related features by gradient boosting tree model, and is obtained by 5-fold cross-validation selection.

[0094] 12. The system of item 1, wherein,

[0095] The system further comprises a bisulfite treatment module for bisulfite treatment of cfDNA of the subject.

[0096] 13. A method for cancer screening, comprising:

[0097] a data collection step for obtaining methylation levels of target regions and cfDNA related features of a subject; and

[0098] a cancer computing step for predicting whether the subject suffers from cancer based on the methylation levels and cfDNA related features obtained in the data collection step.

[0099] 14. The method of item 13, wherein,

[0100] The data collection step comprises a sequencing step, a methylation level analysis step and a cfDNA related feature extraction step,

[0101] The sequencing step is for whole genome sequencing of cfDNA of the subject,

[0102] The methylation level analysis step is for analyzing methylation levels of target regions from sequencing data obtained in the sequencing step, and the cfDNA related feature extraction step is for extracting cfDNA related features from the sequencing data obtained in the sequencing step.

[0103] 15. The method of item 13, wherein,

[0104] The target region comprises any one or more of the following regions:

[0105] chromosome 1 at 151445000-151450000,

[0106] chromosome 2 at 191183500-191188500,

[0107] chromosome 2 at 191184000-191189000,

[0108] chromosome 4 at 68566500-68571500, or

[0109] chromosome 11 at 30601500-30606500.

[0110] 16. The method of item 13, wherein,

[0111] The methylation level of the target region is calculated based on the methylation level of each CG site of the target region, wherein the methylation level of the CG site is the ratio of the number of the site detected as methylated cytosine to the sum of the number of the site detected as methylated cytosine and the number of the site detected as unmethylated cytosine in all sequence results of the site detected.

[0112] 17. The method of item 14, wherein,

[0113] The cfDNA-related features comprise:

[0114] the average of all fragment sizes obtained from the cfDNA sequencing data,

[0115] the mode of all fragment sizes obtained from the cfDNA sequencing data,

[0116] the average coverage obtained from the cfDNA sequencing data,

[0117] the correlation coefficient between the pre-stored average vector of short fragment coverage of the corresponding interval of healthy people and the short fragment coverage obtained from the cfDNA sequencing data of the subject,

[0118] the correlation coefficient between the pre-stored average vector of long fragment coverage of the corresponding interval of healthy people and the long fragment coverage obtained from the cfDNA sequencing data of the subject,

[0119] and the correlation coefficient between the pre-stored average vector of the ratio of short and long fragments of the corresponding interval of healthy people and the ratio of short and long fragments obtained from the cfDNA sequencing data of the subject.

[0120] 18. The method of item 17, wherein,

[0121] The pre-stored average short-interval-coverage vector of healthy people refers to the average short-interval-coverage of known healthy people based on the sequencing data of cfDNA of the known healthy people, which is provided to the data collection step for calculation;

[0122] The pre-stored average long-interval-coverage vector of healthy people refers to the average long-interval-coverage of known healthy people based on the sequencing data of cfDNA of the known healthy people, which is provided to the data collection step for calculation;

[0123] The pre-stored average short-long-interval-coverage vector of healthy people refers to the average short-long-interval-coverage of known healthy people based on the sequencing data of cfDNA of the known healthy people, which is provided to the data collection step for calculation.

[0124] 19. The method of item 17, wherein,

[0125] The average of all fragment sizes obtained from the cfDNA sequencing data refers to the ratio of the sum of the sizes of all fragments obtained from the cfDNA sequencing data of the subject to the number of all fragments;

[0126] The average coverage obtained from the cfDNA sequencing data refers to the number of all fragments obtained from the cfDNA sequencing data of the subject;

[0127] The short-interval-coverage obtained from the cfDNA sequencing data refers to the number of short-interval-fragments obtained from the cfDNA sequencing data of the subject;

[0128] The long-interval-coverage obtained from the cfDNA sequencing data refers to the number of long-interval-fragments obtained from the cfDNA sequencing data of the subject.

[0129] 20. The method of item 17, wherein,

[0130] The length of the long-interval-fragments is 201-320 bp, and the length of the short-interval-fragments is 150-200 bp.

[0131] 21. The method of item 14, wherein,

[0132] The cfDNA sequencing data is cfDNA sequencing data after removing low-quality sequencing fragments.

[0133] 22. The method of item 21, wherein,

[0134] The cfDNA sequencing data is sequencing data after further excluding sequencing data in a low mapping rate interval after removing low-quality sequencing fragments.

[0135] 23. The method according to item 13, wherein,

[0136] In the cancer computing step, a model based on data fitting of methylation levels of known samples and cfDNA related features is pre-stored for predicting whether the subject is suffering from cancer,

[0137] The model is based on data of known samples and cfDNA related features by gradient boosting tree model, and is obtained by 5-fold cross-validation selection.

[0138] 24. The method according to item 13, wherein,

[0139] The method further comprises a bisulfite treatment step for bisulfite treatment of cfDNA of the subject.

[0140] Effects of the invention

[0141] The system of the present application comprehensively utilizes methylation levels and cfDNA feature related indicators for construction, which can greatly reduce the cost of cancer screening and improve the accuracy of cancer screening, and has very high sensitivity and specificity. BRIEF DESCRIPTION OF DRAWINGS

[0142] Various other advantages and benefits of the present application will become apparent to those of ordinary skill in the art, upon reading the following detailed description of the preferred embodiments. The accompanying drawings that are included herein serve only to illustrate preferred embodiments of the present application, and should not be considered limiting of its scope. It should be readily understood that the drawings, described below, are not meant to be exhaustive or limiting of the scope of the present application. Indeed, these drawings merely illustrate selected embodiments of the present application that can make various changes without departing from its spirit and scope. Wherever possible, like reference numbers have been used in the drawings to denote like incremental elements. It should be noted that all references to the methods and processes of the present application are meant to include all structures and equivalents thereof.

[0143] Figure 1 ROC curve in the training set for constructing a model based on 5 regions;

[0144] Figure 2 ROC curve in the test set for constructing a model based on 5 regions;

[0145] Figure 3 Result graph of the cancer screening model constructed for methylation levels;

[0146] Figure 4 Result graph of the cancer screening model constructed for cfDNA features;

[0147] Figure 5 Results plots for cancer screening models constructed for methylation levels and cfDNA features. DETAILED DESCRIPTION

[0148] Specific embodiments of the present application will now be described in greater detail with reference to the figures. Although specific embodiments of the application are illustrated, it will be appreciated that the application can be carried out in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and fully convey the scope of the application to those skilled in the art.

[0149] It should be noted that certain terms have been used throughout the specification and claims which have been used for the purpose of clarity in describing the application. Those skilled in the art will appreciate from the detailed description and drawings that various changes can be made to the application without departing from the scope thereof. Accordingly, it will be understood that the application is not limited to the embodiments set forth herein but is to be given the full scope that includes any and all equivalents of the claimed application.

[0150] Definitions

[0151] Unless otherwise defined, all other technical and scientific terms used herein have the meaning commonly understood by one of ordinary skill in the art to which this application belongs.

[0152] Methylation

[0153] Methylation is an important modification of proteins and nucleic acids, which regulates gene expression and inactivation, is closely related to many diseases such as cancer, aging, Alzheimer's disease, and is one of the important research contents of epigenetics. The most common methylation modifications are DNA methylation and histone methylation.

[0154] DNA methylation refers to the methylation process of the 5th carbon atom on the cytosine in the CpG dinucleotide. As a stable modification state, it can be inherited to the newly born offspring DNA in the process of DNA replication under the action of DNA methyltransferase, and is an important epigenetic mechanism. When DNA methylation occurs, the methylation of the gene promoter region can lead to the transcriptional silence of tumor suppressor genes, so it is closely related to the occurrence of tumors. Abnormal methylation includes high methylation of tumor suppressor genes and DNA repair genes, low methylation of repetitive sequence DNA, and loss of imprinting of certain genes, which is related to the occurrence of many tumors.

[0155] In this paper, ROC curve can reflect the classification effect of the classifier to some extent. AUC is actually the area under the ROC curve. AUC intuitively reflects the classification ability expressed by the ROC curve.

[0156] Whole-genome methylation sequencing

[0157] Whole-genome bisulfite sequencing (WGBS) is considered the "gold standard" of methylation sequencing. Its principle is to use bisulfite treatment to convert unmethylated C bases in the genome into U, which becomes T after PCR amplification, and is distinguished from C bases with methylation modification, and then combined with high-throughput sequencing technology and reference sequence alignment, which can determine whether CpG / CHG / CHH sites are methylated.

[0158] Specificity

[0159] Specificity refers to the proportion of samples of patients without a specific clinical disease whose test results are negative.

[0160] Sensitivity

[0161] Sensitivity refers to the proportion of samples of patients with a clear clinical disease whose test results are positive.

[0162] PPV

[0163] The proportion of people who are truly positive among those who are predicted to be positive.

[0164] NPV

[0165] The proportion of people who are truly negative among those who are predicted to be negative.

[0166] Mode

[0167] The mode refers to the value that appears most frequently in a set of data. The mode of all fragment sizes obtained from cfDNA sequencing data refers to the value that appears most frequently in cfDNA sequencing data.

[0168] Correlation coefficient

[0169] A correlation coefficient is a quantity that represents the degree of correlation between variables. The correlation coefficient used in the present invention is the Pearson correlation coefficient. The Pearson correlation coefficient, also known as the Pearson product-moment correlation coefficient (PPMCC or PCCs for short), is used to measure the correlation (linear correlation) between two variables X and Y, and its value is between -1 and 1. A value of 1 for the coefficient means that X and Y can be well described by a straight line equation, and all data points fall well on a straight line, and Y increases with the increase of X. A value of -1 for the coefficient means that all data points fall on a straight line, and Y decreases with the increase of X. A value of 0 for the coefficient means that there is no linear relationship between the two variables.

[0170] Gradient boosting tree

[0171] Gradient boosting decision tree (GBDT) is also called MART (Multiple Additive Regression Tree), which is an iterative decision tree algorithm. It was considered to be an algorithm with strong generalization ability when it was first proposed.

[0172] Cross validation

[0173] Cross validation, also known as rotation estimation, is a statistical method of cutting the data sample into smaller subsets. The basic idea of cross validation is to group the original data in some sense, one part as the training set and the other part as the validation set or test set. First, the training set is used to train the classifier, and then the validation set is used to test the trained model to evaluate the performance of the classifier.

[0174] The present application provides a system for cancer screening, comprising a data acquisition module and a cancer calculation module. The data acquisition module is used to obtain the methylation level and cfDNA related features of the target region of the subject. The cancer calculation module is used to predict whether the subject is suffering from cancer based on the methylation level and cfDNA related features obtained by the data acquisition module.

[0175] In this paper, the target region of the subject can be a specific region on the chromosome of the subject, for example, the target region includes any one or more of the following regions:

[0176] chromosome 1 at positions 151445000-151450000,

[0177] chromosome 2 at positions 191183500-191188500,

[0178] chromosome 2 at positions 191184000-191189000,

[0179] chromosome 4 at positions 68566500-68571500, or

[0180] chromosome 11 at positions 30601500-30606500.

[0181] In a specific embodiment, the target region is chromosome 1 at positions 151445000-151450000.

[0182] In a specific embodiment, the target region is chromosome 2 at positions 191183500-191188500.

[0183] In a specific embodiment, the target region is chromosome 2 at positions 191184000-191189000.

[0184] In a specific embodiment, the target region is chromosome 4 at positions 68566500-68571500.

[0185] In a specific embodiment, the target region is chromosome 11 at positions 30601500-30606500.

[0186] In a specific embodiment, the target region is chromosome 1 at positions 151445000-151450000 and chromosome 2 at positions 191183500-191188500.

[0187] In a specific embodiment, the target region is chromosome 1 at positions 151445000-151450000 and chromosome 2 at positions 191184000-191189000.

[0188] In a specific embodiment, the target region is chromosome 1 at positions 151445000-151450000 and chromosome 4 at positions 68566500-68571500.

[0189] In a specific embodiment, the target region is chromosome 1 at positions 151445000-151450000 and chromosome 11 at positions 30601500-30606500.

[0190] In a specific embodiment, the target region is 2 191183500-191188500 and 4 68566500-68571500.

[0191] In a specific embodiment, the target region is 2 191183500-191188500 and 4 68566500-68571500.

[0192] In a specific embodiment, the target region is 2 191183500-191188500 and 11 30601500-30606500.

[0193] In a specific embodiment, the target region is 2 191184000-191189000 and 4 68566500-68571500.

[0194] In a specific embodiment, the target region is 2 191184000-191189000 and 11 30601500-30606500.

[0195] In a specific embodiment, the target region is 1 151445000-151450000, 2 191183500-191188500 and 2 191184000-191189000.

[0196] In a specific embodiment, the target region is 1 151445000-151450000, 2 191183500-191188500 and 4 68566500-68571500.

[0197] In a specific embodiment, the target region is 1 151445000-151450000, 2 191183500-191188500 and 11 30601500-30606500.

[0198] In a specific embodiment, the target region is 2 191183500-191188500, 2 191184000-191189000 and 4 68566500-68571500.

[0199] In a specific embodiment, the target region is 2 191183500-191188500, 2 191184000-191189000, and 11 30601500-30606500.

[0200] In a specific embodiment, the target region is 2 191183500-191188500, 4 68566500-68571500, and 11 30601500-30606500.

[0201] In a specific embodiment, the target region is 2 191184000-191189000, 4 68566500-68571500, and 11 30601500-30606500.

[0202] In a specific embodiment, the target region is 1 151445000-151450000, 2 191183500-191188500, 2 191184000-191189000, and 4 68566500-68571500.

[0203] In a specific embodiment, the target region is 1 151445000-151450000, 2 191183500-191188500, 2 191184000-191189000, and 11 30601500-30606500.

[0204] In a specific embodiment, the target region is 2 191183500-191188500, 2 191184000-191189000, 4 68566500-68571500, and 11 30601500-30606500.

[0205] In a specific embodiment, the target region is 1 151445000-151450000, 2 191184000-191189000, 4 68566500-68571500, and 11 30601500-30606500.

[0206] In one embodiment, the target region is 1 51445000-151450000 on chromosome 1, 191183500-191188500 on chromosome 2, 68566500-68571500 on chromosome 4 and 30601500-30606500 on chromosome 11.

[0207] In one embodiment, the target region is 1 51445000-151450000 on chromosome 1, 191183500-191188500 on chromosome 2, 191184000-191189000 on chromosome 2, 68566500-68571500 on chromosome 4 and 30601500-30606500 on chromosome 11.

[0208] The methylation level of a target region is calculated based on the methylation level of each CG site in the target region, wherein the methylation level of a CG site is the ratio of the number of methylated cytosines detected to the sum of the number of methylated cytosines and the number of unmethylated cytosines in all sequence results of the site detected.

[0209] For each window, the number of CG sites in each window is counted. Since the depth of methylated cytosines of each CG site and the total depth of the site are known, the methylation level of the entire window can be calculated, i.e. the ratio of the sum of the depth of methylated cytosines of all CG sites to the sum of the total depth of all CG sites. Each window will obtain a corresponding methylation level through the above calculation method. The depth of methylated cytosines of each CG site is the number of reads showing that the site is methylated cytosine in the sequencing detection result, i.e. the number of reads showing that the site is C (cytosine) in the sequencing result. The total depth of the site is the total number of all sequencing reads covering the site, i.e. the total number of reads showing that the site is C or T (thymine). The depth of methylated cytosines and the total depth of the site can be directly provided after analysis by sequencing software.

[0210] The methylation levels of cancer and healthy cfDNA in the above target regions have significant differences, and therefore can be used as markers for cancer detection:

[0211] 1 51445000-151450000 on chromosome 1,

[0212] 191183500-191188500 on chromosome 2,

[0213] chromosome 2: 191184000-191189000,

[0214] chromosome 4: 68566500-68571500, or

[0215] chromosome 11: 30601500-30606500.

[0216] cfDNA-related features refer to features about cfDNA obtained from cfDNA sequencing data.

[0217] In one specific embodiment, the cfDNA-related features include:

[0218] an average of all fragment sizes obtained from cfDNA sequencing data,

[0219] a mode of all fragment sizes obtained from cfDNA sequencing data,

[0220] an average coverage obtained from cfDNA sequencing data,

[0221] a correlation coefficient between a pre-stored average short fragment coverage vector of corresponding intervals of healthy people and a short fragment coverage obtained from cfDNA sequencing data of the subject,

[0222] a correlation coefficient between a pre-stored average long fragment coverage vector of corresponding intervals of healthy people and a long fragment coverage obtained from cfDNA sequencing data of the subject,

[0223] and a correlation coefficient between a pre-stored average short-long fragment ratio vector of corresponding intervals of healthy people and a short-long fragment ratio obtained from cfDNA sequencing data of the subject.

[0224] The average of all fragment sizes obtained from cfDNA sequencing data refers to a ratio of a sum of sizes of all fragments obtained from cfDNA sequencing data of the subject to a number of all fragments.

[0225] The average coverage obtained from cfDNA sequencing data refers to a number of all fragments obtained from cfDNA sequencing data of the subject.

[0226] The short fragment coverage obtained from cfDNA sequencing data refers to a number of short fragments obtained from cfDNA sequencing data of the subject.

[0227] The long fragment coverage obtained from cfDNA sequencing data refers to a number of long fragments obtained from cfDNA sequencing data of the subject.

[0228] The pre-stored average vector of coverage of short interval in healthy people refers to the average coverage of short interval in healthy people based on sequencing data of cfDNA of known healthy people, which is used to provide the data acquisition module for calculation.

[0229] The pre-stored average vector of coverage of long interval in healthy people refers to the average coverage of long interval in healthy people based on sequencing data of cfDNA of known healthy people, which is used to provide the data acquisition module for calculation.

[0230] The pre-stored average vector of ratio of short and long interval in healthy people refers to the average ratio of short and long interval in healthy people based on sequencing data of cfDNA of known healthy people, which is used to provide the data acquisition module for calculation.

[0231] In a specific embodiment, the length of the long interval is 201-320 bp, and the length of the short interval is 150-200 bp.

[0232] Further, the cfDNA sequencing data is cfDNA sequencing data after removing low-quality sequencing fragments.

[0233] Further, the cfDNA sequencing data is sequencing data after removing low-quality sequencing fragments and excluding sequencing data in a low-mapping rate interval. Specifically, the low-mapping rate interval provided at https: / / genome.ucsc.edu / cgi-bin / hgFileUi?db=hg19&g=wgEncodeMapability can be referred to.

[0234] The data acquisition module of the present application can further comprise three sub-modules, i.e., a sequencing module, a methylation level analysis module and a cfDNA related feature extraction module. The sequencing module is used to perform whole genome sequencing on cfDNA of a subject. The methylation level analysis module is used to analyze the methylation level of a target region from sequencing data obtained from the sequencing module. The cfDNA related feature extraction module is used to extract cfDNA related features from sequencing data obtained from the sequencing module.

[0235] In the cancer calculation module of the present application, a model based on data fitting of methylation level and cfDNA related features of known samples is pre-stored for predicting whether a subject is suffering from cancer. The methylation level and cfDNA related features of a target region of a subject obtained in the data acquisition module are substituted into the model in the cancer calculation module, and a prediction result of whether the subject is suffering from cancer is obtained.

[0236] The model is based on the methylation level and cfDNA related feature data of known samples by gradient boosting tree model, and is obtained by 5-fold cross-validation selection.

[0237] The system also further comprises a bisulfite treatment module for bisulfite treatment of cfDNA of the subject. The bisulfite treated cfDNA is used for subsequent cfDNA sequencing.

[0238] The application also provides a method for cancer screening, comprising a data acquisition step for obtaining the methylation level and cfDNA related feature of the target region of the subject; and

[0239] A cancer calculation step for predicting whether the subject is suffering from cancer based on the methylation level and cfDNA related feature obtained in the data acquisition step.

[0240] Further, the data acquisition step comprises a sequencing step, a methylation level analysis step and a cfDNA related feature extraction step,

[0241] The sequencing step is used for whole genome sequencing of cfDNA of the subject,

[0242] The methylation level analysis step is used to analyze the methylation level of the target region from the sequencing data obtained in the sequencing step, and the cfDNA related feature extraction step is used to extract the related features of the cfDNA sequencing data from the sequencing data obtained in the sequencing step.

[0243] The target region, the methylation level of the target region, and the cfDNA related feature are as described above for the system for cancer screening.

[0244] In the cancer calculation step, a model fitted based on the methylation level and cfDNA related feature data of known samples is pre-stored for predicting whether the subject is suffering from cancer,

[0245] The model is based on the methylation level and cfDNA related feature data of known samples by gradient boosting tree model, and is obtained by 5-fold cross-validation selection.

[0246] The method further comprises a bisulfite treatment step for bisulfite treatment of cfDNA of the subject.

[0247] Embodiment

[0248] Example 1 Calculation of differentially methylated regions and calculation of fragment group features

[0249] 1.1 cfDNA extraction and purification

[0250] 1.1.1 Preparation of plasma samples:

[0251] Centrifuge blood sample at 2000 g for 10 min at 4°C and transfer plasma to a new centrifuge tube. Centrifuge plasma sample at 16000 g for 10 min at 4°C. Depending on the type of collection tube used, proceed to next step. In this experiment the type of collection tube used was Other.

[0252]

[0253] 1.1.2 Lysis and binding

[0254] 1.1.2.1. Prepare Binding Solution / Beads Mix according to the table below and mix thoroughly.

[0255]

[0256] Add the appropriate volume of plasma sample.

[0257] 1.1.2.2. Mix the plasma sample and Binding Solution / Beads Mix thoroughly.

[0258] 1.1.2.3. Incubate for 10 min on a rotating mixer to allow binding of cfDNA to the magnetic beads.

[0259] 1.1.2.4. Place the binding tube on a magnetic stand for 5 min until the solution becomes clear and the magnetic beads are completely adsorbed on the magnetic stand.

[0260] 1.1.2.5. Carefully discard the supernatant with a pipette and keep the tube on the magnetic stand for a few more minutes and remove the remaining supernatant with a pipette.

[0261] 1.1.3 Washing

[0262] 1.1.3.1. Resuspend the beads in 1 ml Wash Solution.

[0263] 1.1.3.2. Transfer the resuspension to a new non-adsorbing 1.5 ml centrifuge tube. Keep the binding tube.

[0264] 1.1.3.3. Place the centrifuge tube containing the bead resuspension on the magnetic stand for 20 s.

[0265] 1.1.3.4. Remove the separated supernatant by pipetting and wash the combined tube. Collect the washed beads again in the resuspension and discard the lysis / binding tube.

[0266] 1.1.3.5. Place the tube on the magnetic stand for 2 min until the solution becomes clear and the beads are concentrated on the magnetic stand. Remove the supernatant with a 1 ml pipette.

[0267] 1.1.3.6. Leave the tube on the magnet and remove as much residual liquid as possible with a 200 μL pipette.

[0268] 1.1.3.7. Remove the tube from the magnet and add 1 ml Wash Solution and vortex for 30 s.

[0269] 1.1.3.8. Place on the magnet for 2 min until the solution is clear and the beads are aggregated on the magnet. Remove the supernatant with a 1 ml pipette.

[0270] 1.1.3.9. Leave the tube on the magnet and remove as much residual liquid as possible with a 200 μL pipette.

[0271] 1.1.3.10. Remove the tube from the magnet and add 1 ml 80% ethanol and vortex for 30 s.

[0272] 1.1.3.11. Place on the magnet for 2 min until the solution is clear and remove the supernatant with a 1 ml pipette.

[0273] 1.1.3.12. Leave the tube on the magnet and remove as much residual liquid as possible with a 200 μL pipette.

[0274] 1.1.3.13. Repeat steps 10-12 above once with 80% ethanol to remove as much supernatant as possible.

[0275] 1.1.3.14. Leave the tube on the magnet and dry the beads in air for 3-5 min.

[0276] 1.1.4 Elute cfDNA

[0277] 1.1.4.1. Add Elution Solution according to the table below.

[0278]

[0279] 1.1.4.2. Place on the magnet for 2 min until the solution is clear and pipette the cfDNA from the supernatant.

[0280] 1.1.4.3. Use the purified cfDNA immediately or transfer the supernatant to a new centrifuge tube and store at -20°C.

[0281] 1.2 gDNA fragmentation and purification:

[0282] 1.2.1. Take 2 μg DNA according to the Qubit concentration, add water to 125 μl and add to the covaris 130 μl fragmentation tube and set the program: 50 W, 20%, 200 cycles, 250 s.

[0283] 1.2.2 Take 1 μl sample after fragmentation for fragment detection using Agilent 2100, normal fragmentation sample detection main peak is about 150bp-200bp.

[0284] For cfDNA sample, Agilent 2100 for fragment detection, direct Qubit for subsequent experiments.

[0285] 1.3 End repair, 3 'end add "A":

[0286] 1.3.1. Take Xng fragmented gDNA or cfDNA into PCR tube, add to 50 μl with nuclease-free water, add the following reagents, vortex mix:

[0287] Component Volume gDNA / cfDNA 50 μl End Repair & A-Tailing Buffer 7 μl End Repair & A-Tailing Enzyme Mix 3 μl Total Volume 60 μl

[0288] 1.3.2. Set the following program on the PCR instrument for reaction:

[0289] Hot lid temperature 85℃.

[0290] Temperature Time 20℃ 30 min 65℃ 30 min 4℃ ∞

[0291] 1.4 Adapter ligation and purification:

[0292] 1.4.1. Refer to the following table to dilute the adapter in advance to the appropriate concentration:

[0293]

[0294]

[0295] 1.4.2. Prepare the following reagents according to the table below, gently pipette mix, centrifuge briefly:

[0296] Component Volume End Repair, A-Tailing reaction product 60 μl Adapters 5 μl Nuclease-free water 5 μl Ligation Buffer 30 μl DNA Ligase 10 μl Total Volume 110 μl

[0297] 1.4.3. Set the following program on the PCR instrument for reaction:

[0298] No hot lid.

[0299] Temperature Time 20℃ 30 min 4℃ ∞

[0300] 1.4.4. According to the following system, add purified magnetic beads for experiment (Agencourt AMPure XP magnetic beads are taken to room temperature in advance and mixed well for standby):

[0301] Component Volume Adapter ligation product 110 μl Agencourt AMPure XP beads 110 μl Total Volume 220 μl

[0302] 1.4.4.1. Gently pipette mix 6 times.

[0303] 1.4.4.2. Incubate at room temperature for 5-15 min. Place the PCR tube on the magnetic stand for 3 min to clear the solution.

[0304] 1.4.4.3. Remove the supernatant. Continue to place the PCR tube on the magnetic stand. Add 200 μΐ of 80% ethanol solution to the PCR tube. Incubate for 30 s.

[0305] 1.4.4.4. Remove the supernatant. Add another 200 μΐ of 80% ethanol solution to the PCR tube. Incubate for 30 s and remove the supernatant completely (it is recommended to use a 10 μΐ pipette to remove the residual ethanol solution at the bottom).

[0306] 1.4.4.5. Incubate at room temperature for 3-5 min to allow the residual ethanol to evaporate completely.

[0307] 1.4.4.6. Add 22 μΐ of Nuclease-free water. Take the PCR tube off the magnetic stand. Gently pipette to resuspend the magnetic beads, avoiding the generation of bubbles. Incubate at room temperature for 2 min.

[0308] 1.4.4.7. Place the PCR tube on the magnetic stand for 2 min to clear the solution.

[0309] 1.4.4.8. Use a pipette to pipette 20 μΐ of the supernatant and transfer it to a new PCR tube.

[0310] 1.5 Bisulfite treatment and purification:

[0311] 1.5.1. Take out the required reagents in advance and dissolve them. Add each reagent according to the following table:

[0312] Component High concentration samples (1 ng - 2 μg) body Low concentration samples (1 - 500 ng) Adapter ligation purification product 20 μl 40 μl Bisulfite solution 85 μl 85 μl DNA protect buffer 35 μl 15 μl Total Volume 140 μl 140 μl

[0313] 1.5.2. The liquid DNA Protect buffer turns blue. Gently pipette to mix, and then divide into two tubes and place them on the PCR instrument.

[0314] 1.5.3. Set the following program and run:

[0315] Hot lid 105 °C.

[0316] Temperature Time 95℃ 5 min 60℃ 10 min 95℃ 5 min 60℃ 10 min 4℃ ∞

[0317] 1.5.4. Briefly centrifuge to combine the same samples in two tubes into a clean 1.5 ml centrifuge tube.

[0318] 1.5.5. Add 310 μΐ of Buffer BL to each sample (add 1 μΐ of Carrier RNA (1 μg / μΐ) if the sample amount is less than 100 ng). Vortex to mix and briefly centrifuge.

[0319] 1.5.6. Add 250 μl absolute ethanol to each sample, vortex mix for 15 s, spin briefly, and add the mix to the prepared corresponding spin column.

[0320] 1.5.7. Let stand for 1 min, spin for 1 min, and transfer the liquid from the collection tube back to the spin column, spin for 1 min, and discard the liquid from the spin column.

[0321] 1.5.8. Add 500 μl buffer BW (note if absolute ethanol is added), spin for 1 min, and discard the waste.

[0322] 1.5.9. Add 500 μl buffer BD (note if absolute ethanol is added), cap the tube, and let stand at room temperature for 15 min. Spin for 1 min, and discard the liquid under the spin column.

[0323] 1.5.10. Add 500 μl buffer BW (note if absolute ethanol is added), spin for 1 min, and discard the liquid under the spin column. Repeat once more for a total of 2 times.

[0324] 1.5.11. Add 250 μl absolute ethanol, spin for 1 min, and place the spin column in a new 2 ml collection tube, discarding all remaining liquid.

[0325] 1.5.12. Place the spin column in a clean 1.5 ml centrifuge tube, add 20 μl nuclease-free water to the center of the spin column membrane, gently cap the tube, let stand at room temperature for 1 min, and spin for 1 min.

[0326] 1.5.13. Transfer the liquid from the collection tube back to the spin column, let stand at room temperature for 1 min, and spin for 1 min.

[0327] 1.6 Pre-amplification and purification prior to hybridization:

[0328] 1.6.1. Prepare the reaction according to the table below, mix by pipetting, and spin briefly:

[0329]

[0330]

[0331] 1.6.2. Set the following program and start the PCR program:

[0332] Hot lid 105 °C

[0333]

[0334] 1.6.3. The number of PCR cycles is adjusted according to the amount of DNA input. The reference data is shown below:

[0335]

[0336]

[0337] 1.6.4. Add 50 μΐ of Agencourt AMPure XP magnetic beads to the PCR tube after the reaction is completed, mix well with a pipette to avoid air bubbles, and mix well in advance at room temperature (Agencourt AMPure XP is mixed and balanced at room temperature in advance).

[0338] 1.6.5. Incubate at room temperature for 5-15 min, and place the PCR tube on the magnetic stand for 3 min to clarify the solution.

[0339] 1.6.6. Remove the supernatant, and continue to place the PCR tube on the magnetic stand, and add 200 μΐ of 80% ethanol solution to the PCR tube, and stand for 30 s.

[0340] 1.6.7. Remove the supernatant, and add 200 μΐ of 80% ethanol solution to the PCR tube, and stand for 30 s, and then completely remove the supernatant (it is recommended to use a 10 μΐ pipette to remove the residual ethanol solution at the bottom).

[0341] 1.6.8. Stand at room temperature for 5 min to completely volatilize the residual ethanol.

[0342] 1.6.9. Add 30 μΐ of nuclease-free water, remove the centrifuge tube from the magnetic stand, and resuspend the magnetic beads by gently pipetting with a pipette.

[0343] 1.6.10. Stand at room temperature for 2 min, and place the 200 μΐ PCR tube on the magnetic stand for 2 min to clarify the solution.

[0344] 1.6.11. Use a pipette to transfer the supernatant to a new 200 μΐ PCR tube (place it in an ice box), label the sample number on the reaction tube, and prepare for the next reaction.

[0345] 1.6.12. Take 1 μΐ of sample to determine the library concentration using Qubit, and record the library concentration.

[0346] 1.6.13. Take 1 μΐ of sample to determine the library fragment length using Agilent 2100, and the library length is about 270-320 bp.

[0347] 1.6.14. Use the Illumina high-throughput sequencing platform for sequencing.

[0348] 1.6.15. Methylation bioinformatics analysis process. The process is as follows: use fastp quality control software to view the quality of raw sequencing data, and filter, cut or remove low-quality reads to obtain corresponding clean data; use Bismark bowtie2 alignment software to align the quality-controlled clean data data to the reference genome (hg19); use deduplicate_bismark to remove duplicates from the initial alignment bam file; use Bismark_methylation_extractor to extract the corresponding methylation site information to obtain the final methylation CG file (including all single CG site information files); finally, use the sliding window method to divide the window on the reference genome, and calculate the overall methylation level of the CG site in each window interval; for each sample, the methylation level of the corresponding window is calculated, and the differentially methylated window is found according to the group of different samples.

[0349] 1.6.16. cfDNA fragment feature extraction process.

[0350] Use fastp quality control software to view the sequencing quality, remove low-quality reads, and then use Bismarker alignment software to align the quality-controlled clean data data to the reference genome,

[0351] Obtain the aligned bam file, filter out reads with MAPQ <30, and use the R package GCcontent to extract whole genome fragment information.

[0352] The extracted fragment information is tiled into adjacent, non-overlapping 100kb intervals according to the autosomes of the hg19 reference genome, and intervals with low alignment rates are excluded based on previous research work. Define the short fragment length to be between 150-200bp, and the long fragment length to be between 201-320bp, and calculate the short fragment coverage, long fragment coverage, and total coverage of short fragments + long fragments in each interval.

[0353] Use local weighted regression (LOWESS) to correct short fragments, long fragments, and total coverage, thereby removing coverage bias caused by GC bias.

[0354] Merge the 100kb intervals into 5MB intervals in turn to obtain 499 non-overlapping intervals, and then calculate the corrected short fragment coverage, long fragment coverage, short-long fragment ratio, and total coverage of each interval, thereby obtaining multiple preliminary features of the sample.

[0355] Example 2

[0356] Based on the 14 lung cancer patient cfDNA and 22 healthy person cfDNA training set, the methylation levels of 1583 initial markers of 14 lung cancer patients and 22 healthy persons were detected by the method described in Example 1, and 5 methylation regions most significantly distinguishing lung cancer and healthy person cfDNA were screened out as specific regions for lung cancer detection. As shown in Table 1, the corresponding region information is as follows: the first region, 151445000-151450000 of chromosome 1; the second region, 191183500-191188500 of chromosome 2; the third region, 191184000-191189000 of chromosome 2; the fourth region, 68566500-68571500 of chromosome 4; and the fifth region, 30601500-30606500 of chromosome 11.

[0357] The methylation level data of each of the above markers detected in each region of the 14 lung cancer patients and 22 healthy persons detected based on the method of Example 1 were input into R software, and a random forest model was constructed by using the randomForest package of R software for model regression. The regression results showed that in the training set, the cutoff value of the comprehensive methylation level based on 5 markers that can be used to predict lung cancer results was 0.442, i.e., the specified threshold was 0.442 (greater than 0.442 was judged as lung cancer patient), the AUC of the final model reached 1, the accuracy was 100%, the sensitivity was 100%, the specificity was 100%, the PPV was 100%, and the NPV was 100%, see Table 1 and Figure 1 .

[0358] Table 1

[0359]

[0360] Example 3

[0361] Based on the 5 methylation markers of Example 2, the pROC in R software was used to calculate the cutoff and AUC values of the comprehensive methylation level of each methylation marker that can be used to predict lung cancer results in the test set (10 lung cancer patient cfDNA not used in the training set and 16 healthy person cfDNA not used in the training set), see Table 2.

[0362] Table 2

[0363]

[0364] Example 4

[0365] Based on the model constructed in Example 2, in the test set of 10 lung cancer patient cfDNA and 16 healthy person cfDNA, the cutoff of the comprehensive methylation level based on 5 markers for predicting lung cancer results is 0.442, that is, the specified threshold is 0.442 (greater than 0.442 is judged as lung cancer patient), the AUC reaches 0.919, the accuracy is 84.62%, the sensitivity is 90%, the specificity is 81.25%, the PPV is 75%, and the NPV is 92.86%, and the specific information is shown in Table 3 and Figure 2 .

[0366] Table 3

[0367]

[0368] Example 5

[0369] Select another 42 lung cancer samples different from Examples 2-4, another 64 healthy people different from Examples 2-4, a total of 106 samples, collect peripheral blood according to the method of Example 1; build a library, and sequence through the Illumina platform; the methylation level of the 5 differential methylation regions obtained by the above methylation analysis process; the above 106 samples are divided into a 70% training set and a 30% test set; on all 106 samples, use multiple machine learning models (logistic regression, support vector machine, random forest, gradient boosting tree, etc.) to evaluate the training results of each model using 5-fold cross-validation, and the gradient boosting tree model is the best on this data set. Therefore, the gradient boosting tree model is selected as the final model for modeling.

[0370] Further, for the training set (selecting 28 lung cancer samples from 42 lung cancer samples and 46 healthy samples from 64 healthy people), the gradient boosting tree model is used, and the 5-fold cross-validation method is used to obtain the optimal model; the results of the model on the test set are as Figure 3 and Table 4, the AUC value is 0.905, the sensitivity is 85.7%, the specificity is 88.9%, the positive predictive value PPV is 85.7%, and the negative predictive value NPV is 88.9%.

[0371] Table 4

[0372]

[0373] Example 6

[0374] The same 42 lung cancer samples and 64 healthy people as in Example 5 were selected, and peripheral blood was collected according to the method of Example 1; the library was constructed and sequenced by the Illumina platform; according to the fragment group bioinformatics analysis process, multiple preliminary characteristics of the sample about the fragment group were obtained: 499 short fragment coverage characteristics, 499 long fragment coverage characteristics and 499 total coverage characteristics. Four comprehensive indicators were obtained by calculating the above three types of characteristics: cov (coverage, 499 intervals combined into one interval, calculating the average value of 499 total fragment coverage), short.cor (the correlation coefficient of the average value vector of the short fragment coverage of the corresponding interval of all healthy people and the short fragment coverage of the sample), long.cor (the correlation coefficient of the average value vector of the long fragment coverage of the corresponding interval of all healthy people and the long fragment coverage of the sample), ratio.cor (the correlation coefficient of the average value vector of the short-long fragment ratio of all healthy people and the short-long fragment ratio of the sample). Using the above four characteristics combined with mean_size (fragment size average) and mode_size (fragment size mode), 106 samples were divided into a 70% training set and a 30% test set,

[0375] Further, the training set (28 lung cancer samples selected from the 42 lung cancer samples and 46 healthy samples selected from the 64 healthy people) was used; the gradient boosting tree model was constructed by using the R language tool and adopting 5-fold cross-validation, and then the effect of the gradient boosting tree model was verified on the test set.

[0376] The results of the model of the present application on the test set are as follows: Figure 4 and Table 5, the AUC value is 0.937, the sensitivity is 85.7%, and the specificity is 88.9%. The positive predictive value PPV is 85.7%, and the negative predictive value NPV is 88.9%.

[0377] Table 5

[0378]

[0379]

[0380] Example 7

[0381] The same 42 lung cancer samples and 64 healthy people as in Example 5 were selected, and peripheral blood was collected according to the method of Example 1; the library was constructed and sequenced by the Illumina platform; according to the fragment group bioinformatics analysis process, multiple preliminary characteristics of the sample about the fragment group were obtained: 499 short fragment coverage characteristics, 499 long fragment coverage characteristics and 499 total coverage characteristics. Four comprehensive indicators were obtained by calculating the above three types of characteristics: cov (coverage, 499 intervals combined into one interval, calculating the average value of 499 total fragment coverage), short.cor (the correlation coefficient of the average value vector of the short fragment coverage of the corresponding interval of all healthy people and the short fragment coverage of the sample), long.cor (the correlation coefficient of the average value vector of the long fragment coverage of the corresponding interval of all healthy people and the long fragment coverage of the sample), ratio.cor (the correlation coefficient of the average value vector of the short-long fragment ratio of all healthy people and the short-long fragment ratio of the sample). Using the above four characteristics combined with mean_size (fragment size average) and mode_size (fragment size mode), 106 samples were divided into a 70% training set and a 30% test set,

[0375] Further, the training set (28 lung cancer samples selected from the 42 lung cancer samples and 46 healthy samples selected from the 64 healthy people) was used; the gradient boosting tree model was constructed by using the R language tool and adopting 5-fold cross-validation, and then the effect of the gradient boosting tree model was verified on the test set.

[0376] The results of the model of the present application on the test set are as follows: Figure 4 and Table 5, the AUC value is 0.937, the sensitivity is 85.7%, and the specificity is 88.9%. The positive predictive value PPV is 85.7%, and the negative predictive value NPV is 88.9%.

[0377] Table 5

[0378]

[0379]

[0380] Example 7

[0381] The same 42 lung cancer samples and 64 healthy people as in Example 5 were selected, and peripheral blood was collected according to the method of Example 1; the library was constructed and sequenced by the Illumina platform; according to the fragment group bioinformatics analysis process, multiple preliminary characteristics of the sample about the fragment group were obtained: 499 short fragment coverage characteristics, 499 long fragment coverage characteristics and 499 total coverage characteristics. Four comprehensive indicators were obtained by calculating the above three types of characteristics: cov (coverage, 499 intervals combined into one interval, calculating the average value of 499 total fragment coverage), short.cor (the correlation coefficient of the average value vector of the short fragment coverage of the corresponding interval of all healthy people and the short fragment coverage of the sample), long.cor (the correlation coefficient of the average value vector of the long fragment coverage of the corresponding interval of all healthy people and the long fragment coverage of the sample), ratio.cor (the correlation coefficient of the average value vector of the short-long fragment ratio of all healthy people and the short-long fragment ratio of the sample). Using the above four characteristics combined with mean_size (fragment size average) and mode_size (fragment size mode), 106 samples were divided into a 70% training set and a 30% test set,

[0375] Further, the training set (28 lung cancer samples selected from the 42 lung cancer samples and 46 healthy samples selected from the 64 healthy people) was used; the gradient boosting tree model was constructed by using the R language tool and adopting 5-fold cross-validation, and then the effect of the gradient boosting tree model was verified on the test set.Figure 5 and Table 3, the AUC value is 0.978, the sensitivity is 92.9%, and the specificity is 94.4%. The positive predictive value PPV is 92.9%, and the negative predictive value NPV is 94.4%.

[0382] Table 6

[0383]

[0384] The method of the present application using WGBS data can be used for cancer screening, the fragment group information obtained by WGBS data is reliable, and WGBS can also calculate the methylation level, so that the two aspects can be combined at the same time to do early screening. The previous study used WGS data to obtain fragment group information, and then combined WGS data to obtain CNV copy number variation information to do tumor early screening.

[0385] Although the embodiments of the present application are described above in conjunction with the drawings, the present application is not limited to the specific embodiments and application fields described above, and the specific embodiments described above are only illustrative and guiding, but not limiting. A person of ordinary skill in the art can make many forms under the guidance of the present specification and without departing from the scope protected by the claims of the present application, and these all belong to the protection of the present application.

Claims

1. A system for cancer screening, comprising: a data acquisition module for obtaining methylation levels of target regions and cfDNA related features of a subject; and a cancer computation module for predicting whether the subject suffers from cancer based on the methylation levels and the cfDNA related features obtained by the data acquisition module; wherein the target regions comprise the following regions with reference to the hg19 reference genome: chromosome 1: 151445000-151450000, chromosome 2: 191183500-191188500, chromosome 2: 191184000-191189000, chromosome 4: 68566500-68571500, and chromosome 11: 30601500-30606500; wherein the cancer is lung cancer; the methylation levels of the target regions are calculated based on the methylation levels of each CG site in the target regions, wherein the methylation level of the CG site is the ratio of the number of methylated cytosines to the sum of the number of methylated cytosines and the number of unmethylated cytosines in all sequence results of the site detected; the cfDNA related features comprise: the average of all fragment sizes obtained from the cfDNA sequencing data, the mode of all fragment sizes obtained from the cfDNA sequencing data, the average coverage obtained from the cfDNA sequencing data, the correlation coefficient between the pre-stored average vector of short fragment coverage of corresponding intervals of healthy people and the short fragment coverage obtained from the cfDNA sequencing data of the subject, the correlation coefficient between the pre-stored average vector of long fragment coverage of corresponding intervals of healthy people and the long fragment coverage obtained from the cfDNA sequencing data of the subject, and the correlation coefficient between the pre-stored average vector of short-long fragment ratio of corresponding intervals of healthy people and the short-long fragment ratio obtained from the cfDNA sequencing data of the subject. 2.The system of claim 1, wherein the data acquisition module comprises a sequencing module, a methylation level analysis module and a cfDNA related feature extraction module, the sequencing module is used for whole genome sequencing of cfDNA of the subject, the methylation level analysis module is used for analyzing the methylation levels of the target regions from the sequencing data obtained by the sequencing module, and the cfDNA related feature extraction module is used for extracting the cfDNA related features from the sequencing data obtained by the sequencing module. 3.The system of claim 1, wherein the pre-stored average vector of short fragment coverage of corresponding intervals of healthy people refers to the average of short fragment coverage of corresponding intervals of healthy people based on the sequencing data of cfDNA of known healthy people, which is provided to the data acquisition module for calculation; the pre-stored average vector of long fragment coverage of corresponding intervals of healthy people refers to the average of long fragment coverage of corresponding intervals of healthy people based on the sequencing data of cfDNA of known healthy people, which is provided to the data acquisition module for calculation; The pre-stored average vector of the interval short-long fragment ratio of healthy people is the average of the interval short-long fragment ratio of known healthy people based on the sequencing data of the cfDNA of the known healthy people, which is provided to the data acquisition module for calculation.

4. The system of claim 1, wherein, The average of all fragment sizes obtained from the cfDNA sequencing data is the ratio of the sum of the sizes of all fragments obtained from the cfDNA sequencing data of the subject to the number of all fragments. The average coverage obtained from the cfDNA sequencing data is the number of all fragments obtained from the cfDNA sequencing data of the subject. The short fragment coverage obtained from the cfDNA sequencing data is the number of short fragments obtained from the cfDNA sequencing data of the subject. The long fragment coverage obtained from the cfDNA sequencing data is the number of long fragments obtained from the cfDNA sequencing data of the subject.

5. The system of claim 1, wherein, The length of the long fragment is 201-320 bp, and the length of the short fragment is 150-200 bp.

6. The system of claim 2, wherein, The cfDNA sequencing data is the cfDNA sequencing data after removing low-quality sequencing fragments.

7. The system of claim 6, wherein, The cfDNA sequencing data is the sequencing data after removing low-quality sequencing fragments and further removing sequencing data in the low mapping rate interval.

8. The system of claim 1, wherein, In the cancer calculation module, a model based on the methylation level of known samples and the data fitting of cfDNA-related features is pre-stored for predicting whether the subject is suffering from cancer. The model is based on the methylation level of known samples and the data of cfDNA-related features by gradient boosting tree model, and is obtained by 5-fold cross-validation selection.

9. The system of claim 1, wherein, The system further comprises a bisulfite treatment module for bisulfite treatment of the cfDNA of the subject.

Images