Gene marker for determining symptoms of organism, construction method of detection model and detection device

Abstract

The invention relates to a gene marker for determining the symptom of an organism, a construction method of a detection model and a detection device, and the gene marker for determining the symptom of the organism is composed of k continuous base sequences on a reference genome, the base sequence is a 3 'end motif base sequence set obtained by comparing a 5'end breakpoint of free desoxyribonucleic acid (cfDNA) to a position on a reference genome and extending k bp from the position to the upstream at an interval of n bp bases. The inventor finds that the change of a fixed-length base sequence set at the 3'end of a free deoxyribonucleic acid (cfDNA) fragment, namely 3 'end motif, also reflects the change of different nuclease activities in long-term research, the free deoxyribonucleic acid (cfDNA) fragment can become a new biomarker for determining the symptoms of an organism, and a free deoxyribonucleic acid (cfDNA) fragment terminal sequence characteristic system is perfected together with 5' end motif and breakpoint motif. And compared with the 5 'end motif and the breakpoint motif, the 3' end motif has a more excellent classification effect when being used as the gene marker for determining the symptoms of the organism.

Description

Technical Field

[0001] The present invention belongs to the technical field of molecular biomedicine, and specifically relates to a gene marker for determining the pathological condition of an organism, a construction method of a detection model and a detection device. Background Art

[0002] Cell-free deoxyribonucleic acid (cfDNA), especially plasma free deoxyribonucleic acid, is composed of various tissues shed from the body and produced through non-random processes. It can be used as a non-invasive biomarker to conveniently and accurately diagnose the physiological and pathological conditions of organisms.

[0003] Specifically, mutations, copy number variations, methylation and fragment characteristics of cfDNA can be used to diagnose the physiology and pathology of organisms. The fragment characteristics mainly include the ratio of different lengths of each region of the whole genome and the end sequence preference of the fragments, and the end sequence preference of cfDNA fragments may be related to the activity of nucleases. Mechanistically, one of the reasons for the change in the end sequence preference of cfDNA fragments in patients with physiological changes or diseases (such as malignant tumors) is the change in the activity of different nucleases.

[0004] At present, in the end repair process of the library construction experiment for the analysis of the terminal sequence of free deoxyribonucleic acid (cfDNA) fragments, the protruding bases at the 3' end are removed and the missing bases at the 3' end are supplemented, so that the original 3' end information is modified, while the 5' end information is preserved. Therefore, the research on the terminal sequence of free deoxyribonucleic acid (cfDNA) fragments only has a fixed-length base sequence set at the 5' end called "5'endmotif" and a fixed-length base sequence set based on a number of bp upstream and downstream of the 5' end breakpoint called "breakpoint motif". Summary of the invention

[0005] In order to solve the above problems, the present invention provides a gene marker for determining the pathology of an organism, a method for constructing a detection model, and a detection device. The present invention discovered for the first time that the 3'end motif frequency is significantly different between benign plasma samples and plasma samples of different cancer types, and can be used as a gene marker for early screening and diagnosis of cancer. A classification model was constructed using the base sequence in the significantly different 3'end motif, and its AUC was 0.93, which is higher than the breakpoint model (AUC: 0.85) and the 5'end motif model (AUC: 0.89). As a gene marker for early screening and diagnosis of cancer, the 3'end motif has an excellent classification effect.

[0006] The technical solution of the present invention is:

[0007] In a first aspect of the present invention, a genetic marker for determining the pathology of an organism is provided, wherein the marker is composed of k consecutive base sequences on a reference genome, and the base sequence is a set of base sequences of a 3' end motif obtained by aligning the 5' end breakpoint of free deoxyribonucleic acid (cfDNA) to a position on the reference genome and extending it by k bp upstream with an interval of n bp.

[0008] In another preferred embodiment, n is any natural number, preferably, n=0.

[0009] In another preferred embodiment, n is any natural number selected from the following group: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10.

[0010] In another preferred embodiment, k is any positive integer, preferably, k=4.

[0011] In another preferred embodiment, k is any positive integer selected from the following group: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10.

[0012] In another preferred embodiment, when n=0 and k=4, the base sequence set of the 3'end motif includes the following base sequence set coded as D1 to D126:

[0013]

[0014]

[0015] In another preferred embodiment, when n=0 and k=4, the base sequence set of the 3'end motif is the set of base sequences coded as D1 to D126 in Table A below:

[0016] Table A

[0017]

[0018]

[0019] In the second aspect of the invention, the use of the gene marker in early cancer screening is provided.

[0020] In a preferred aspect, the present invention provides the use of the gene marker described in the first aspect of the present invention, or a detection reagent thereof, for preparing a diagnostic reagent or a diagnostic kit, wherein the diagnostic reagent or the diagnostic kit is used for early cancer screening.

[0021] In another preferred embodiment, the cancer is selected from the following group: gastric cancer, breast cancer, colorectal cancer, bile duct cancer, central or peripheral nervous system tissue cancer, endocrine or neuroendocrine system cancer or hematopoietic system cancer, glioma, sarcoma, epithelial cancer, lymphoma, melanoma, fibroma, meningioma, brain cancer, kidney cancer, thyroid cancer, parathyroid cancer, pituitary tumor, adrenal tumor, osteogenic sarcoma tumor, neuroendocrine system tumor, lung cancer, head and neck cancer, prostate cancer, esophageal cancer, tracheal cancer, liver cancer, bladder cancer, pancreatic cancer, ovarian cancer, uterine cancer, cervical cancer, testicular cancer, skin cancer, or a combination thereof.

[0022] In another preferred embodiment, the cancer is selected from the group consisting of gastric cancer, breast cancer, colorectal cancer, bile duct cancer, or a combination thereof.

[0023] In another preferred embodiment, the determination of early cancer screening is as follows:

[0024] When the gene marker is an up-regulated marker, and when it is up-regulated in the sample to be tested, it indicates that the subject to be tested corresponding to the sample to be tested has a high risk of suffering from cancer; and

[0025] When the gene marker is a down-regulated marker, and when it is down-regulated in the test sample, it indicates that the test subject corresponding to the test sample has a high risk of suffering from cancer.

[0026] During the process of nucleic acid hydrolysis by nuclease, a nucleic acid is hydrolyzed into two or more strands. Then the base sequence of the 5' end of a deoxyribonucleic acid (DNA) fragment extending upstream (5' end direction) corresponds to the 3' end base sequence of the upstream deoxyribonucleic acid DNA fragment, and the inventors of the present application define the base sequence set of the 5' end of free deoxyribonucleic acid (cfDNA) extending upstream by a certain length as the 3' end motif of free deoxyribonucleic acid (cfDNA), which also includes the base sequence set extending upstream by a certain length after an interval of n bp.

[0027] Considering that different nucleases show different preferences for the hydrolysis of nucleic acids, it is inferred that changes in the fixed-length base sequence set at the 3' end of free deoxyribonucleic acid (cfDNA) fragments, namely the 3'end motif, will also reflect changes in the activity of different nucleases, becoming a new biomarker for early cancer screening and diagnosis, and together with the 5'end motif and breakpoint motif, completing the terminal sequence feature system of free deoxyribonucleic acid (cfDNA) fragments.

[0028] In a third aspect of the present invention, a method for constructing a cancer early screening detection model is provided, comprising the following steps:

[0029] (1) Extraction and sequencing of cell-free deoxyribonucleic acid (cfDNA) were performed on samples from the positive group and the control group to obtain read segment data;

[0030] (2) aligning the read data to a reference genome to obtain the position of the 5' end of the read on the reference genome;

[0031] (3) according to the position of the 5' end of the read segment on the reference genome, the sequence data is extended by k bp after being spaced n bp upstream, and the obtained sequence data is used as the base sequence set of the 3' end motif;

[0032] (4) Counting the proportion of each base sequence in the base sequence set in all fragments, obtaining the frequency of each base sequence, and screening out base sequences with significant differences in frequency between the positive group and the control group samples;

[0033] (5) The proportion of base sequences with significant differences obtained in step (4) in all fragments is used as the input value of the classifier model, and the risk of having cancer is used as the output value. The model is trained using sample data from the positive group and the control group to obtain the cancer early screening model.

[0034] In a preferred aspect of the present invention, a method for constructing a cancer early screening detection model is provided, comprising the following steps:

[0035] (s1) providing a first data set and a second data set, wherein the first data set includes base sequence data of 3'end motifs in cancer-positive group samples and base sequence data of 3'end motifs in control group samples;

[0036] The second data set includes base sequence data of 3'end motifs in cancer positive group samples and base sequence data of 3'end motifs in control group samples; and

[0037] (s2) constructing a cancer early screening detection model based on the data in the first data set and verifying it on the second data set, thereby obtaining the cancer early screening detection model.

[0038] In another preferred embodiment, the sample is a plasma sample.

[0039] In another preferred embodiment, the base sequence data of the 3'end motif is obtained by the following steps:

[0040] (1) Extracting and sequencing free DNA from samples of the positive group and the control group, respectively, and obtaining read data;

[0041] (2) aligning the read data to a reference genome to obtain the position of the 5' end of the read on the reference genome;

[0042] (3) according to the position of the 5' end of the read segment on the reference genome, the sequence data obtained is extended by k bp after being spaced n bases upstream as the base sequence set of the 3' end motif; and

[0043] (4) Counting the proportion of each base sequence in the base sequence set in all fragments, obtaining the frequency of each base sequence, and thus obtaining the base sequence data of the 3' end motif.

[0044] In another preferred embodiment, step (s2) further includes the following sub-steps:

[0045] (s2a) screening out base sequences with significant differences in frequency between samples of the cancer-positive group and the control group in the first data set; and

[0046] (s2b) The proportion of base sequences with significant differences obtained in step (s2a) in all fragments is used as the input value of the classifier model, and the risk of having cancer is used as the output value. The model is trained and verified using sample data from the positive group and the control group to obtain the cancer early screening model.

[0047] In another preferred embodiment, the significant difference is manifested as a significant increase or decrease in the proportion (or content) of any base sequence in the base sequence set of the positive group in all fragments relative to the control group.

[0048] In another preferred embodiment, the significant upregulation refers to a significance test of the difference between the proportion (or content) C1 of the base sequence in all fragments and the control reference value C0, a multiple hypothesis verification of the difference p value, a multiple hypothesis test q value less than a threshold, and a ratio C1 / C0>1, preferably, a q value <0.05.

[0049] In another preferred embodiment, the significant down-regulation refers to a significance test of the difference between the ratio of the proportion (or content) C1 of the base sequence in all fragments and the control reference value C0, and a multiple hypothesis verification of the difference p value, the multiple hypothesis test q value is less than the threshold, and the ratio C0 / C1>1, preferably, the q value <0.05.

[0050] In another preferred embodiment, the extracted free deoxyribonucleic acid (cfDNA) is subjected to the following operations to obtain the position of the 5' end of the read segment on the reference genome, specifically comprising:

[0051] (S1) The free deoxyribonucleic acid (cfDNA) is sequenced to obtain raw sequencing sequences (rawreads);

[0052] (S2) The raw sequencing sequences (raw reads) are filtered out by software cutadptor to remove the adapters and poor quality sequencing sequences, thereby obtaining reliable sequencing data (clean reads);

[0053] (S3) Use the alignment software bwa to align the reliable sequencing data (clean reads) to the reference genome to obtain an alignment file;

[0054] (S4) using the software samtools to sort the alignment file to obtain a sorted alignment file;

[0055] (S5) using the software picard to remove duplicates from the sorted alignment file to obtain a deduplicated alignment file;

[0056] (S6) parsing the deduplication alignment file, and obtaining the position information of the 5' end of the read segment on the reference genome based on the alignment information.

[0057] Among them, the software cutadptor can be replaced by other filtering software. When fastp, soapnuke, trimmomatics, etc. are used to replace it, it can also play the role of data quality control filtering. The software bwa can be replaced by other comparison software. When soap, bowtie, etc. are used to replace it, it can also play the role of comparison; the software Picard can be replaced by other deduplication software. When samtools, sambamba, etc. are used to replace it, it can play the role of deduplication. The parameters of cutadptor and bwa can be changed according to the actual situation.

[0058] In another preferred embodiment, step (s2a) includes the following sub-steps: calculating the p-value of the difference in frequency of each base sequence in the base sequence set of the 3'end motif between the two groups, performing multiple hypothesis verification on the p-value, and selecting base sequences with q-values ​​less than a certain threshold.

[0059] In another preferred example, the threshold of the q value is 0.05.

[0060] Depending on the actual situation, student test, Mann-Whitney U test and other methods can be used to test whether there are significant differences between the two groups of data; according to the actual situation, "Benjamini&Hochberg", "bonferroni" and other methods can be used to correct the p value for multiple hypothesis testing.

[0061] In the study, the inventors of the present application evaluated the differences of 3'end motif between benign plasma samples and various malignant plasma samples by using 30 benign plasma samples and 81 cancer plasma samples (including 27 gastric cancer, 27 breast cancer and 27 colorectal cancer). The set of several sequences at the 3' end of the free deoxyribonucleic acid (cfDNA) fragments that were repaired was defined as modified 3'end motif, and the differences between the 3'end motif frequency distribution and the modified 3'end motif frequency distribution were analyzed using the above 30 benign plasma samples. The analysis results showed that: 1) There was a significant difference between the 3'end motif frequency distribution and the modified 3'end motif frequency distribution, indicating that the end repair had a significant modification on the 3'end sequence of the free deoxyribonucleic acid (cfDNA) fragment, and the 3'end sequence repaired at the end of the free deoxyribonucleic acid (cfDNA) fragment could not replace the 3'end motif. 2) There was a significant difference in the frequency of 3'end motif between benign and various cancer samples, which can be used as a biomarker for early screening and diagnosis of cancer. 3) Compared with the 5'end motif and breakpoint motif, the 3'end motif has the greatest difference between benign and malignant samples.

[0062] 30 benign plasma samples and 40 malignant tumor plasma samples (including 10 gastric cancer cases, 10 breast cancer cases, 10 colorectal cancer cases and 10 bile duct cancer cases) were used as training sets to construct the classifier model. Another 30 benign plasma samples and 61 malignant tumor plasma samples (including 17 gastric cancer cases, 17 breast cancer cases, 17 colorectal cancer cases and 10 bile duct cancer cases) were used as test sets to evaluate the performance of the classifier model. The evaluation content includes: the performance of the classifier model of base sequences of different lengths in the 3'end motif, the classifier model of different interval distances in the 3'end motif, the breakpoint motif classifier model and the 5'end motif classifier model. The results showed that the 3'end motif classifier model with a length of 4bp and an interval of 0bp was the best, with an AUC of 0.93. Compared with the 5'end motif and the breakpoint motif, the 3'end motif has a better classification effect as a biomarker for early screening and diagnosis of cancer.

[0063] The plasma sample data in the training set and the test set are independent of each other.

[0064] In the fourth aspect of the present invention, a cancer early screening detection model constructed by the method described in the third aspect of the present invention is provided.

[0065] In a preferred aspect of the present invention, a computer storage device is provided, in which a computer program corresponding to the algorithm of the early cancer screening detection model constructed according to the method described in the second aspect of the present invention is stored.

[0066] In a fifth aspect of the present invention, a cancer early screening detection device is provided, the device comprising:

[0067] (1) Input module: The input module is configured to input a base sequence set of the 3' end motif of the plasma sample of the test subject; the base sequence set includes the following base sequence set coded as D1 to D126:

[0068]

[0069]

[0070] (2) a processing module, wherein the processing module is configured to input the input base sequence data into the cancer early screening detection model established in the third aspect of the present invention for processing, or to perform the following processing: compare the data with a control reference value, and when the base sequence in the data is an up-regulated base sequence in Table A, the content of the base sequence or the proportion of the base sequence in all fragments significantly increases, and when the base sequence in the data is a down-regulated base sequence in Table A, the content of the base sequence or the proportion of the base sequence in all fragments significantly decreases, it indicates a high risk of cancer; otherwise, it indicates a low risk of cancer; and

[0071] (3) An output module, wherein the output module is configured to output the cancer early screening test result.

[0072] In another preferred embodiment, the device further comprises a sequencing module, and the sequencing module is configured to extract free deoxyribonucleic acid from the sample, construct a library and sequence it to obtain read segment data.

[0073] In another preferred example, the device also includes a comparison module, which is configured to compare the read segment data to a reference genome, obtain the position of the 5' end of the read segment on the reference genome, and obtain sequence data on the reference genome that is n bp bases upstream of the position and then extended by k bp bases as a base sequence set of the 3'end motif.

[0074] In another preferred example, n=0 and k=4.

[0075] In another preferred example, the device further comprises a screening module, wherein the screening module is configured to screen out base sequences in the 3' end motif base sequence set that have significant differences between the positive group and the control group.

[0076] In another preferred example, the device also includes a model building module, which is configured to use the proportion of each base sequence with significant differences in all fragments as the input value of the classifier model, and the risk of cancer as the output value, and use the sample data of the positive group and the control group to train and verify the model to obtain a cancer early screening detection model.

[0077] In another preferred example, the device also includes an analysis module, which is configured to use the proportion of each base sequence with significant differences in all fragments as the input value of the cancer early screening detection model, and the risk of having cancer as the output value; if the cancer risk is higher than the threshold, the person is judged to be a high-risk cancer population, otherwise, the person is judged to be a low-risk cancer population.

[0078] The beneficial effects of the present invention are:

[0079] The present invention provides a gene marker for determining the pathological condition of an organism, wherein the marker is composed of k continuous base sequences on a reference genome, and the base sequence is a set of base sequences of a 3'end motif obtained by aligning the 5' end breakpoint of free deoxyribonucleic acid (cfDNA) to the position on the reference genome and extending it by k bp upstream. The inventors of the present application have found in long-term research that the fixed-length base sequence set at the 3' end of the free deoxyribonucleic acid (cfDNA) fragment, i.e., the change of the 3'end motif, will also reflect the change of different nuclease activities, and can become a new biomarker for early screening and diagnosis of cancer, and together with the 5'end motif and the breakpoint motif, improve the terminal sequence feature system of the free deoxyribonucleic acid (cfDNA) fragment. And compared with the 5'end motif and the breakpoint motif, the 3'end motif has an excellent classification effect as a gene marker for early screening and diagnosis of cancer.

[0080] It should be understood that within the scope of the present invention, the above-mentioned technical features of the present invention and the technical features specifically described below (such as embodiments) can be combined with each other to form a new or preferred technical solution. Due to space limitations, they will not be described one by one here. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0082] Figure 1 A comparison chart showing the frequency distribution of the 3'end motif and the 3'end sequence of the end-repaired read (modified 3'end motif).

[0083] Figure 2 Shows the comparison of the differential base sequences of the 3'end motif, breakpoint motif, and 5'end motif.

[0084] Figure 3 Shows the comparison of base sequence classifier models with different 3'end motif lengths.

[0085] Figure 4 Shows the comparison of the sequence classifier models for 4 bp base sequence with different spacing distances of 3'end motif.

[0086] Figure 5 Shown are the performances of the 3'end motif classifier model, the breakpoint motif classifier model, and the 5'end motif classifier model. DETAILED DESCRIPTION

[0087] After extensive and in-depth research, the inventor unexpectedly discovered that the 3'end motifs in benign plasma samples and malignant plasma samples (or plasma samples of tumor patients) are significantly different, and the base sequence set of the 3'end motifs with significant differences can be used to effectively distinguish benign plasma samples from malignant plasma samples, thereby effectively distinguishing high-risk cancer subjects from low-risk cancer subjects. On this basis, the present invention was completed.

[0088] Specifically, the inventors extracted 30 benign plasma samples and 81 cancer (including 27 cases each of gastric cancer, breast cancer and colorectal cancer) plasma samples and obtained their sample data, and found that the significant difference of 3'end motif in benign plasma samples and cancer (gastric cancer, breast cancer and colorectal cancer) patient plasma samples was greater than breakpoint and 5end motif.

[0089] Subsequently, the inventor randomly selected 10 gastric cancers, 10 breast cancers, 10 colorectal cancers and 10 newly added bile duct cancer samples from 81 cancer samples, observed the differences between the 40 plasma samples and the 30 benign plasma samples, and used the 70 sample data as a training set to construct a cancer early screening detection model (or classifier model). The results showed that 126 of the 256 base sequences of the 3'end motif in the cancer samples (D1 to D126) were significantly different from those in the benign samples, which can be used to effectively distinguish high-risk cancer subjects from low-risk cancer subjects. The remaining 51 samples (including 17 gastric cancers, 17 breast cancers, and 17 colorectal cancers) and the corresponding data of the newly added 10 bile duct cancers and 30 benign plasma sample data were used as test sets to test (or verify) the model effect.

[0090] The test results show that the 3'end motif base sequence set shown in D1 to D126 of the present invention (e.g., when n=0 and k=4) can effectively distinguish cancer (e.g., gastric cancer, breast cancer, colorectal cancer, bile duct cancer) plasma samples from benign plasma samples, and the highest AUC value obtained by verification is 0.93. Therefore, the 3'end motif base sequence set of the present invention can effectively determine whether the subject to be tested is a high-risk subject for cancer.

[0091] To make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described in detail below. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other implementation methods obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0092] The above technical solution is described in detail below in conjunction with specific embodiments.

[0093] The experimental methods used in the following examples are conventional methods unless otherwise specified. The materials, reagents, etc. used in the following examples are all commercially available unless otherwise specified.

[0094] The main advantages of the present invention include:

[0095] (a) The gene markers of the present invention (or the base sequence set of the 3' end motif of D1 to D126) can effectively determine whether the subject to be tested is a high-risk subject for cancer, thereby effectively assisting clinicians in formulating appropriate treatment strategies.

[0096] The present invention will be further described below in conjunction with specific examples. It should be understood that these examples are intended to illustrate the present invention only and are not intended to limit the scope of the present invention. The experimental methods in the following examples where specific conditions are not specified are usually performed under conventional conditions, such as those described in Sambrook et al., Molecular Cloning: A Laboratory Manual (New York: Cold Spring Harbor Laboratory Press, 1989), or under conditions recommended by the manufacturer. Unless otherwise indicated, percentages and parts are weight percentages and weight parts.

[0097] Example 1

[0098] This example evaluates the differences in 3'end motifs between benign plasma samples and different types of malignant plasma samples, and the specific steps are as follows:

[0099] (1) Cell-free deoxyribonucleic acid (cfDNA) was extracted from 30 benign plasma samples and 81 cancer plasma samples, including 27 cases of gastric cancer, breast cancer, and colorectal cancer. The extracted cell-free deoxyribonucleic acid (cfDNA) was subjected to the following operations to obtain the position of the 5' end of the read segment on the reference genome, including:

[0100] (S1) Free deoxyribonucleic acid (cfDNA) is sequenced to obtain raw sequencing sequences (raw reads);

[0101] (S2) The raw sequencing sequences (raw reads) are filtered out by software cutadptor to remove the adapters and poor quality sequencing sequences to obtain reliable sequencing data (clean reads). The parameters of the software cutadptor are set to '-n10--max-n 0.05-q 30,30-m 36-e 0.1';

[0102] (S3) Use the alignment software bwa to align the obtained reliable sequencing data (clean reads) to the reference genome to obtain an alignment file with default parameters;

[0103] (S4) using the software samtools to sort the alignment file to obtain a sorted alignment file;

[0104] (S5) using the software picard to remove duplicates from the sorted alignment file to obtain a deduplicated alignment file;

[0105] (S6) Parse the deduplication alignment file, remove the alignment results on chromosomes chrX and chrY, remove "not the best alignment sequence (not primary alignment), filter the alignment sequence that fails platform / vendor quality checks, duplicate sequences caused by PCR or instrument sequencing (read is PCR or optical duplicate), and possible chimeric alignment sequences (supplementary alignment)", remove the alignment sequence with a mapping quality value (MAPQ) <1, remove the alignment sequence with soft clip (softclip) or hard clip (hardclip) at the 3' end of the read segment, and remove the alignment sequence with soft clip (softclip) or hard clip (hardclip) at the 5' end of the read segment. Only the alignment sequence with the correct length and direction (read mapped inproper pair) is retained; based on the alignment information, obtain the position information of the 5' end of the read segment on the reference genome.

[0106] (2) according to the position of the 5' end of the read segment on the reference genome, the 5' end of the read segment is extended by 4 bp after a gap of 0 bp upstream to obtain a 4-mer 3 end motif;

[0107] (3) The proportion of each base sequence in the 3'end motif base sequence set in all fragments was calculated to obtain the frequency of each base sequence. The Mann-Whitney U test was used to calculate the difference in the frequency of each base sequence in the 3'end motif base sequence set between benign plasma samples and gastric cancer plasma samples, between benign plasma samples and breast cancer plasma samples, and between benign plasma samples and colorectal cancer plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with q values ​​< 0.05 were selected as base sequences with significant differences.

[0108] In this example, the 3'end motifs in the benign plasma and gastric cancer plasma samples were compared. The results are shown in Table 1, indicating that 174 out of 256 motifs were significantly different.

[0109] Table 1-Differences in 3'end motifs between benign plasma samples and gastric cancer plasma samples

[0110]

[0111]

[0112]

[0113]

[0114] The 3'end motifs in the benign plasma and breast cancer plasma samples were compared, and the results are shown in Table 2, indicating that 132 out of 256 were significantly different.

[0115] Table 2-3'end motif differences between benign plasma samples and breast cancer plasma samples

[0116]

[0117]

[0118]

[0119]

[0120] The 3'end motifs in the benign plasma and colorectal cancer plasma samples were compared, and the results are shown in Table 3, indicating that 212 out of 256 were significantly different.

[0121] Table 3-Differences in 3'end motifs between benign plasma samples and colorectal cancer plasma samples

[0122]

[0123]

[0124]

[0125]

[0126]

[0127] The above results show that there are significant differences in the frequency of 3'end motifs between benign plasma samples and plasma samples of three different types of cancer, indicating that the 3'end motifs differ in multiple cancer types.

[0128] Example 2

[0129] This embodiment provides a cancer early screening detection device, including a sequencing module, an alignment module, a screening module, a model building module and an analysis module;

[0130] The sequencing module is used to extract and sequence blood free deoxyribonucleic acid (cfDNA) from the sample to obtain read segment data;

[0131] The alignment module is used to align the read segment data to the reference genome, obtain the position of the 5' end of the read segment on the reference genome, and obtain sequence data of the reference genome extending k bp bases upstream of the position by n bases as the base sequence set of the 3'end motif;

[0132] The screening module is used to screen out base sequences with significant differences between the positive group and the control group in the base sequence set of the 3'end motif;

[0133] The model building module is used to use the proportion of each base sequence with significant difference in all fragments as the input value of the classifier model, and the risk of having cancer as the output value, and the model is trained and verified using the sample data of the positive group and the control group to obtain the cancer early screening detection model;

[0134] The analysis module is used to use the proportion of each base sequence with significant differences in all fragments as the input value of the cancer early screening model (or classifier model), and the risk of having cancer as the output value; if the cancer risk is higher than the threshold, it is determined to be a high-risk cancer population, otherwise it is a low-risk cancer population.

[0135] Among them, the data sets for model training and model verification (or testing) are data sets with different data sources.

[0136] Experimental Example 1

[0137] This experimental example verifies that there are significant differences in the frequency distribution of the 3'end motif and the modified 3'end motif, as follows:

[0138] Free deoxyribonucleic acid (cfDNA) was extracted from 30 benign plasma samples, and steps (S1) to (S6) in Example 1 were repeated for the extracted free deoxyribonucleic acid (cfDNA) to obtain the position of the 5' end of the read segment on the reference genome, and the following operations were performed:

[0139] (1) according to the position of the 5' end breakpoint of the read segment on the reference genome, the 4-mer 3' end motif is obtained by extending the 5' end breakpoint upstream by 0 bp and then by 4 bp, and at the same time, a modified 3' end motif of the 4-mer position of the 3' end of the read segment on the reference genome is extracted;

[0140] (2) Calculate the sum of the number of each base sequence in the 3'end motif in all samples, and calculate the sum of the number of each base sequence in the modified3'end motif in all samples;

[0141] (3) Count the proportion of each base sequence in the 3'end motif in all fragments to obtain the frequency of each base sequence; count the proportion of each base sequence in the modified 3'end motif in all fragments to obtain the frequency of each base sequence.

[0142] (4) The chi-square goodness of fit test was used to evaluate the difference in frequency distribution of 3'end motif and modified 3'end motif.

[0143] (5) Calculate the Pearson correlation coefficient C based on the sum of the X-squared value and the number of base sequences in the 3'end motif.

[0144] like Figure 1 As shown, the pvalue in the chi-square goodness of fit test is 0, and the Pearson series connection coefficient C is 0.64, indicating that the 3'end motif and the 3'end sequence of the read segment after end repair have different frequency distributions. The end repair has a significant modification on the 3'end sequence of the read segment, and the 3'end motif cannot be replaced by the 3'end sequence of the free deoxyribonucleic acid (cfDNA) fragment end repair.

[0145] Experimental Example 2

[0146] This experimental example compares the 3'end motif with the breakpoint motif and the 5'end motif as follows:

[0147] Free deoxyribonucleic acid (cfDNA) was extracted from 30 benign plasma samples and 81 malignant tumor plasma samples, including 27 gastric cancer cases, 27 breast cancer cases and 27 colorectal cancer cases. Steps (S1) to (S6) in Example 1 were repeated for the extracted free deoxyribonucleic acid (cfDNA) to obtain the position of the 5' end of the read segment on the reference genome, and the following operations were performed:

[0148] (1) According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 4 bp upstream after a gap of 0 bp to obtain a 4-mer 3' end motif; According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 2 bp upstream and downstream respectively to obtain a 4-mer breakpoint motif; According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 4 bp downstream to obtain a 4-mer 5' end motif;

[0149] (2) Count the frequencies of each base sequence in the 3'end motif, breakpoint motif, and 5'end motif respectively;

[0150] (3) The Mann-Whitney U test was used to statistically analyze the differences in the frequencies of each base sequence in the 3'end motif, breakpoint motif, and 5'end motif between benign plasma samples and gastric cancer plasma samples. The Mann-Whitney U test was used to statistically analyze the differences in the frequencies of each base sequence in the 3'end motif, breakpoint motif, and 5'end motif between benign plasma samples and breast cancer plasma samples. The Mann-Whitney U test was used to statistically analyze the differences in the frequencies of each base sequence in the 3'end motif, breakpoint motif, and 5'end motif between benign plasma samples and colorectal cancer plasma samples. The Benjamini&Hochberg method was used for multiple hypothesis verification, and base sequences with q values ​​<0.05 were selected and their effect size Cohen's distance between the two groups was calculated;

[0151] (4) Compare the Cohen’s d values ​​of the three markers: 3’end motif, breakpoint motif and 5’end motif.

[0152] like Figure 2 As shown, the median Cohen's d value of the 3'end motif is 1.07, the median Cohen's d value of the breakpoint motif is 1.04, and the median Cohen's d value of the 5'end motif is 1.05. The 3'end motif has the greatest difference between benign and malignant samples.

[0153] Experimental Example 3

[0154] This experimental example selects 3'end motif base sequences with significant differences between benign samples and malignant samples, constructs a cancer early screening detection model (or classifier model) and verifies it, as follows:

[0155] Thirty benign plasma samples and 40 malignant tumor plasma samples (including 10 gastric cancer cases, 10 breast cancer cases, 10 colorectal cancer cases and 10 bile duct cancer cases) were used as training sets, and free deoxyribonucleic acid (cfDNA) of the training set plasma samples was extracted. Steps (S1) to (S6) in Example 1 were repeated for the extracted free deoxyribonucleic acid (cfDNA) to obtain the position of the 5' end of the read segment on the reference genome, and the following operations were performed:

[0156] (1) According to the position of the 5' end breakpoint of the read segment on the reference genome, extend 4 bp upstream after 0 bp to obtain the 3' end motif of the 4-mer;

[0157] (2) Count the frequency of each base sequence in the 3'end motif separately;

[0158] (3) The Mann-Whitney U test was used to statistically analyze the difference in the frequency of each base sequence in the 3'end motif between benign plasma samples and malignant plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with a q value < 0.05 were selected;

[0159] (4) The ratio of the base sequences in the 3'end motif with significant differences obtained in step (3) to the total fragments is used as the input value of the 3'end motif classifier model, and the model is trained using the R software package caret and the gbm algorithm. The parameters of the trainControl function are: "method =

[0160] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0161] The parameters of the training model train function are:

[0162] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0163] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0164] trControl=ctrl)"

[0165] (5) Using 30 benign plasma samples and 61 malignant tumor plasma samples (including 17 gastric cancer cases, 17 breast cancer cases, 17 colorectal cancer cases and 10 bile duct cancer cases) as a test set, extracting free deoxyribonucleic acid (cfDNA) from the test set samples, repeating the operations (S1)-(S6) in step (1) of Example 1 for the extracted free deoxyribonucleic acid (cfDNA), repeating the operations of steps (1) and (2) of this experimental example, and obtaining the frequency of each base sequence in the 3'end motif;

[0166] (6) Select the frequencies of the 3'end motif base sequences with significant differences obtained in step (3) and input them into the 3'end motif classifier model obtained in step (4) to evaluate the performance of the 3'end motif classifier model.

[0167] The results showed that 126 of the 256 base sequences of the 3'end motif were significantly different between good and bad, as shown in Table 4:

[0168] Table 4 Differences in 3'end motifs between benign and malignant plasma samples

[0169]

[0170]

[0171]

[0172] The frequencies of the 126 bases described in Table 4 in the training set plasma samples were selected as input values ​​to build a cancer early screening detection model (or classifier model), and the frequencies of the 126 bases described in Table 4 in the test set plasma samples were input into the cancer early screening detection model to verify the performance of the model. The AUC was 0.93, as shown in Figure 2. Figure 3 shown.

[0173] Experimental Example 4

[0174] This experimental example uses 3'end motif base sequences of different lengths to construct a cancer early screening detection model (or classifier model) and compares them as follows:

[0175] Thirty benign plasma samples and 40 malignant tumor plasma samples (including 10 gastric cancer cases, 10 breast cancer cases, 10 colorectal cancer cases and 10 bile duct cancer cases) were used as training sets, and free deoxyribonucleic acid (cfDNA) of the training set plasma samples was extracted. Steps (S1) to (S6) in Example 1 were repeated for the extracted free deoxyribonucleic acid (cfDNA) to obtain the position of the 5' end of the read segment on the reference genome, and the following operations were performed:

[0176] (1) According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 2 bp, 4 bp and 6 bp respectively after a 0 bp interval upstream to obtain a 2 bp base sequence, a 4 bp base sequence and a 6 bp base sequence;

[0177] (2) Count the frequencies of 2 bp base sequences, 4 bp base sequences, and 6 bp base sequences respectively;

[0178] (3) The Mann-Whitney U test was used to statistically analyze the difference in the frequency of 2bp base sequences between benign plasma samples and malignant plasma samples, and the Benjamini & Hochberg method was used to perform multiple hypothesis verification, and base sequences with a q value of <0.05 were selected; the Mann-Whitney U test was used to statistically analyze the difference in the frequency of 4bp base sequences between benign plasma samples and malignant plasma samples, and the Benjamini & Hochberg method was used to perform multiple hypothesis verification, and base sequences with a q value of <0.05 were selected; the Mann-Whitney U test was used to statistically analyze the difference in the frequency of 6bp base sequences between benign plasma samples and malignant plasma samples, and the Benjamini & Hochberg method was used to perform multiple hypothesis verification, and base sequences with a q value of <0.05 were selected;

[0179] (4) The proportion of 2bp base sequences with significant differences in all fragments obtained in step (3) was used as the input value of the 2bp classifier model, and the model was trained using the gbm algorithm using the R software package caret. The parameters of the trainControl function are: "method = "repeatedcv", number = 10, repeats = 10, verboseIter = FALSE, savePredictions = TRUE, classProbs = TRUE, summaryFunction = twoClassSummary";

[0180] The parameters of the training model train function are:

[0181] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0182] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0183] trControl=ctrl)"

[0184] (5) The proportion of 4bp base sequences with significant differences in all fragments obtained in step (3) was used as the input value of the 4bp classifier model, and the model was trained using the gbm algorithm using the R software package caret. The parameters of the trainControl function are: "method = "repeatedcv", number = 10, repeats = 10, verboseIter = FALSE, savePredictions = TRUE, classProbs = TRUE, summaryFunction = twoClassSummary";

[0185] The parameters of the training model train function are:

[0186] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0187] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0188] trControl=ctrl)"

[0189] (6) The proportion of 6bp base sequences with significant differences in all fragments obtained in step (3) was used as the input value of the 6bp classifier model, and the model was trained using the gbm algorithm using the R software package caret. The parameters of the trainControl function are: "method = "repeatedcv", number = 10, repeats = 10, verboseIter = FALSE, savePredictions = TRUE, classProbs = TRUE, summaryFunction = twoClassSummary";

[0190] The parameters of the training model train function are:

[0191] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0192] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0193] trControl=ctrl)"

[0194] (7) Using 30 benign plasma samples and 61 malignant tumor plasma samples (including 17 gastric cancer cases, 17 breast cancer cases, 17 colorectal cancer cases and 10 bile duct cancer cases) as a test set, extracting free deoxyribonucleic acid (cfDNA) from the test set samples, repeating the operations (S1)-(S6) in step (1) of Example 1 for the extracted free deoxyribonucleic acid (cfDNA), and repeating the operations (1)-(2) of this experimental example to obtain the frequencies of 2bp, 4bp and 6bp base sequences;

[0195] (8) Select the frequencies of the 2bp base sequences with significant differences obtained in step (3) in the test set and input them into the 2bp base sequence classifier model obtained in step (4) to evaluate the performance of the 2bp sequence classifier model.

[0196] (9) Select the frequencies of the 4 bp base sequences with significant differences obtained in step (3) in the test set and input them into the 4 bp base sequence classifier model obtained in step (5) to evaluate the performance of the 4 bp base sequence classifier model.

[0197] (10) Select the frequencies of the 6 bp base sequences with significant differences obtained in step (3) in the test set and input them into the 6 bp base sequence classifier model obtained in step (6) to evaluate the performance of the 6 bp base sequence classifier model.

[0198] (11) Compare the performance of the 2bp base sequence classifier model, the 4bp base sequence classifier model and the 6bp base sequence classifier model in predicting benign or malignant tumors.

[0199] The results showed that 10 of the 16 base sequences of 2bp in the 3'end motif had significant differences between good and malignant tumors; 126 of the 256 base sequences of 4bp in length had significant differences between good and malignant tumors; 1699 of the 4096 base sequences of 6bp in length had significant differences between good and malignant tumors; the AUC of the 2bp base sequence classifier model was 0.86, the AUC of the 4bp base sequence classifier model was 0.93, and the AUC of the 6bp base sequence classifier model was 0.89. The classifier model constructed with the 4bp base sequence had the best performance. Figure 3 as shown in .

[0200] Experimental Example 5

[0201] This experimental example uses 3'end motif base sequences with a length of 4 bp and different spacing distances to construct a cancer early screening detection model (or classifier model) and compare them as follows:

[0202] Thirty benign plasma samples and 40 malignant tumor plasma samples (including 10 gastric cancer cases, 10 breast cancer cases, 10 colorectal cancer cases and 10 bile duct cancer cases) were used as training sets, and free deoxyribonucleic acid (cfDNA) of the training set plasma samples was extracted. Steps (S1) to (S6) in Example 1 were repeated for the extracted free deoxyribonucleic acid (cfDNA) to obtain the position of the 5' end of the read segment on the reference genome, and the following operations were performed:

[0203] (1) according to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 4 bp after being spaced upstream by 0 bp, 1 bp, 2 bp and 3 bp, respectively, to obtain a base sequence with a length of 4 bp and a spacing distance of 0 bp, a base sequence with a length of 4 bp and a spacing distance of 1 bp, a base sequence with a length of 4 bp and a spacing distance of 2 bp, and a base sequence with a length of 4 bp and a spacing distance of 3 bp;

[0204] (2) Count the frequencies of base sequences with a length of 4 bp at intervals of 0 bp, 1 bp, 2 bp, and 3 bp respectively;

[0205] (3) The Mann-Whitney U test was used to statistically analyze the difference in the frequency of base sequences with an interval distance of 0 bp and a length of 4 bp between benign plasma samples and malignant plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with a q value of < 0.05 were selected. The Mann-Whitney U test was used to statistically analyze the difference in the frequency of base sequences with an interval distance of 1 bp and a length of 4 bp between benign plasma samples and malignant plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with a q value of < 0.05 were selected. The Mann-Whitney U test was used to statistically analyze the difference in the frequency of base sequences with an interval distance of 0 bp and a length of 2 bp between benign plasma samples and malignant plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with a q value of < 0.05 were selected. The test statistically analyzed the difference in the frequency of 4-bp base sequences with an interval distance of 3 bp between benign plasma samples and malignant plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with q values ​​< 0.05 were selected.

[0206] (4) The proportion of the base sequences with a significant difference of 0 bp interval and a length of 4 bp in all fragments obtained in step (3) is used as the input value of the 0 bp interval classifier model, and the model is trained using the R software package caret and the gbm algorithm. The parameters of the trainControl function are: "method =

[0207] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0208] The parameters of the training model train function are:

[0209] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0210] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0211] trControl=ctrl)"

[0212] (5) The proportion of 4 bp base sequences with a significant difference of 1 bp interval obtained in step (3) in all fragments is used as the input value of the 1 bp interval classifier model, and the model is trained using the R software package caret and the gbm algorithm. The parameters of the trainControl function are: "method =

[0213] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0214] The parameters of the training model train function are:

[0215] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0216] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0217] trControl=ctrl)"

[0218] (6) The proportion of 4 bp base sequences with a significant difference of 2 bp interval obtained in step (3) in all fragments is used as the input value of the 2 bp interval classifier model, and the model is trained using the R software package caret and the gbm algorithm. The parameters of the trainControl function are: "method =

[0219] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0220] The parameters of the training model train function are:

[0221] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0222] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0223] trControl=ctrl)"

[0224] (7) The proportion of 4 bp base sequences with a significant difference of 3 bp interval obtained in step (3) in all fragments is used as the input value of the 3 bp interval classifier model, and the model is trained using the R software package caret and the gbm algorithm. The parameters of the trainControl function are: "method =

[0225] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0226] The parameters of the training model train function are:

[0227] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0228] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0229] trControl=ctrl)"

[0230] (8) Using 30 benign plasma samples and 61 malignant tumor plasma samples (including 17 gastric cancer cases, 17 breast cancer cases, 17 colorectal cancer cases and 10 bile duct cancer cases) as a test set, extracting free deoxyribonucleic acid (cfDNA) from the test set samples, repeating the operations (S1)-(S6) in step (1) of Example 1 for the extracted free deoxyribonucleic acid (cfDNA), and repeating the operations (1)-(2) of this experimental example to obtain the frequencies of 4 bp base sequences with intervals of 0 bp, 1 bp, 2 bp and 3 bp;

[0231] (9) Select the frequencies of base sequences with a significant difference of 0 bp interval and a length of 4 bp obtained in step (3) in the test set and input them into the 0 bp interval classifier model obtained in step (4) to evaluate the performance of the 0 bp interval classifier model.

[0232] (10) Select the frequencies of 4 bp base sequences with a significant difference of 1 bp interval obtained in step (3) in the test set and input them into the 1 bp interval classifier model obtained in step (4) to evaluate the performance of the 1 bp interval classifier model.

[0233] (11) Select the frequencies of 4 bp base sequences with a significant difference of 2 bp interval obtained in step (3) in the test set and input them into the 2 bp interval classifier model obtained in step (4) to evaluate the performance of the 2 bp interval classifier model.

[0234] (12) Select the frequencies of 4 bp base sequences with a significant difference of 3 bp interval obtained in step (3) in the test set and input them into the 3 bp interval classifier model obtained in step (4) to evaluate the performance of the 3 bp interval classifier model.

[0235] (13) Compare the performance of the 0 bp interval classifier model, 1 bp interval classifier model, 2 bp interval classifier model and 3 bp interval classifier model in predicting benign and malignant tumors.

[0236] The results showed that 126 of the 256 base sequences with a 0bp interval and a length of 4bp in the 3'end motif showed significant differences between good and malignant, 137 of the 256 base sequences with a 1bp interval and a length of 4bp showed significant differences between good and malignant, 129 of the 256 base sequences with a 2bp interval and a length of 4bp showed significant differences between good and malignant, and 114 of the 256 base sequences with a 3bp interval and a length of 4bp showed significant differences between good and malignant. The AUC of the 0bp classifier model was 0.93, the AUC of the 1bp classifier model was 0.92, the AUC of the 2bp classifier model was 0.90, and the AUC of the 3bp classifier model was 0.89. The classifier model constructed with a 0bp interval performed best, as shown in Figure 2. Figure 4 shown.

[0237] Experimental Example 6

[0238] This experimental example compares the cancer early screening detection model (classifier model) constructed by 3'end motif, breakpoint motif, and 5'end motif, as follows:

[0239] Thirty benign plasma samples and 40 malignant tumor plasma samples (including 10 gastric cancer cases, 10 breast cancer cases, 10 colorectal cancer cases and 10 bile duct cancer cases) were used as training sets, and free deoxyribonucleic acid (cfDNA) of the training set plasma samples was extracted. Steps (S1) to (S6) in Example 1 were repeated for the extracted free deoxyribonucleic acid (cfDNA) to obtain the position of the 5' end of the read segment on the reference genome, and the following operations were performed:

[0240] (1) According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 4 bp upstream after a gap of 0 bp to obtain the 3' end motif of the 4-mer; According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 2 bp upstream and downstream respectively to obtain the breakpoint motif of the 4-mer; According to the position of the 5' end breakpoint of the read segment on the reference genome, the read segment is extended by 4 bp downstream respectively to obtain the 3' end motif of the 4-mer;

[0241] (2) Count the frequencies of each base sequence in the 3'end motif, breakpoint motif, and 5'end motif respectively;

[0242] (3) The Mann-Whitney U test was used to statistically analyze the differences in the frequencies of each base sequence in the 3'end motif, breakpoint motif, and 5'end motif between benign plasma samples and malignant plasma samples. The Benjamini & Hochberg method was used for multiple hypothesis verification, and base sequences with q values ​​< 0.05 were selected;

[0243] (4) The ratio of the base sequences in the 3'end motif with significant differences obtained in step (3) to the total fragments is used as the input value of the 3'end motif classifier model, and the model is trained using the SVM algorithm using the R software package caret. The parameters of the trainControl function are: "method =

[0244] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0245] The parameters of the training model train function are:

[0246] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0247] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0248] trControl=ctrl)"

[0249] (5) The ratio of the base sequences in the breakpoint motif with significant differences obtained in step (3) to the total number of fragments is used as the input value of the breakpoint motif classifier model, and the model is trained using the R software package caret and the gbm algorithm. The parameters of the trainControl function are: "method =

[0250] "repeatedcv",number=10,repeats=10,verboseIter=FALSE,savePredictions=TRUE,classProbs=TRUE,summaryFunction=twoClassSummary";

[0251] The parameters of the training model train function are:

[0252] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0253] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0254] trControl=ctrl)"

[0255] (6) The proportion of base sequences in the 5'end motif with significant differences obtained in step (3) in all fragments is used as the input value of the 5'end motif classifier model, and the model is trained using the gbm algorithm using the R software package caret. The parameters of the trainControl function are: "method = "repeatedcv", number = 10, repeats = 10, verboseIter = FALSE, savePredictions = TRUE, classProbs = TRUE, summaryFunction = twoClassSummary";

[0256] The parameters of the training model train function are:

[0257] "method='gbm',tuneGrid=data.frame(n.trees=150,interaction.depth=3,

[0258] shrinkage=0.1n.minobsinnode=10),preProcess=c("corr","nzv"),

[0259] trControl=ctrl)"

[0260] (7) Using 30 benign plasma samples and 61 malignant tumor plasma samples (including 17 gastric cancer cases, 17 breast cancer cases, 17 colorectal cancer cases and 10 bile duct cancer cases) as a test set, extracting free deoxyribonucleic acid (cfDNA) from the test set samples, repeating the operations (S1)-(S6) in step (1) of Example 1 for the extracted free deoxyribonucleic acid (cfDNA), repeating the operations of steps (1) and (2) of this experimental example, and obtaining the frequency of each base sequence in the 3'end motif, breakpoint motif and 5'end motif;

[0261] (8) Select the frequencies of the 3'end motif base sequences with significant differences obtained in step (3) and input them into the 3'end motif classifier model obtained in step (4) to evaluate the performance of the 3'end motif classifier model.

[0262] (9) Select the frequencies of the breakpoint motif base sequences with significant differences obtained in step (3) and input them into the breakpoint motif classifier model obtained in step (5) to evaluate the performance of the breakpoint motif classifier model.

[0263] (10) Select the frequencies of the 5'end motif base sequences with significant differences obtained in step (3) and input them into the 5'end motif classifier model obtained in step (6) to evaluate the performance of the 5'end motif classifier model.

[0264] (11) Compare the performance of the 3'end motif classifier model, the breakpoint motif classifier model, and the 5'end motif classifier model in predicting benign or malignant status.

[0265] The results showed that 126 of the 256 base sequences of the 3'end motif were significantly different between good and malignant, 95 of the 256 base sequences of the breakpoint motif were significantly different between good and malignant, and 85 of the 256 base sequences of the 5'end motif were significantly different between good and malignant. The AUC of the 3'end motif classifier model was 0.93, the AUC of the breakpoint motif classifier model was 0.85, and the AUC of the 5'end motif classifier model was 0.89. The 3'end motif classifier model had the best classification effect. Figure 5 shown.

[0266] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A gene marker for determining a disease state of an organism, characterized in that: The gene marker is composed of k consecutive base sequences on the reference genome, which are a set of base sequences of the 3'end motif obtained by aligning the 5' end breakpoint of the free deoxyribonucleic acid to the position on the reference genome and extending it by k bp upstream with an interval of n bp bases.

2. The gene marker according to claim 1, characterized in that n is any natural number, preferably, n=0.

3. The gene marker according to claim 1, characterized in that k is any positive integer, preferably, k=4.

4. The gene marker according to claim 1, characterized in that When n=0 and k=4, the base sequence set of the 3'end motif includes the following base sequence set coded as D1 to D126:

5. The use of the gene marker or the detection reagent thereof according to any one of claims 1 to 4, characterized in that: Used for preparing diagnostic reagents or diagnostic kits, wherein the diagnostic reagents or diagnostic kits are used for early cancer screening.

6. A method for constructing a cancer early screening detection model, characterized in that: The steps include: (s1) providing a first data set and a second data set, wherein the first data set includes base sequence data of 3'end motifs in cancer-positive group samples and base sequence data of 3'end motifs in control group samples; The second data set includes base sequence data of 3'end motifs in cancer positive group samples and base sequence data of 3'end motifs in control group samples; and (s2) constructing a cancer early screening detection model based on the data in the first data set and verifying it on the second data set, thereby obtaining the cancer early screening detection model.

7. The method according to claim 6, characterized in that The base sequence data of the 3'end motif is obtained by the following steps: (1) Extract free DNA from samples of the positive group and the control group, build libraries and sequence them to obtain read data; (2) aligning the read data to a reference genome to obtain the position of the 5' end of the read on the reference genome; (3) according to the position of the 5' end of the read segment on the reference genome, the sequence data is extended by k bp after being spaced n bases upstream as the base sequence set of the 3' end motif; and (4) Counting the proportion of each base sequence in the base sequence set in all fragments, obtaining the frequency of each base sequence, and thus obtaining the base sequence data of the 3' end motif.

8. The method according to claim 6, characterized in that Step (s2) also includes the following sub-steps: (s2a) screening out base sequences with significant differences in frequency between samples of the cancer-positive group and the control group in the first data set; and (s2b) The proportion of base sequences with significant differences obtained in step (s2a) in all fragments is used as the input value of the classifier model, and the risk of having cancer is used as the output value. The model is trained and verified using sample data from the positive group and the control group to obtain the cancer early screening model.

9. A computer storage device, characterized in that: The device stores a computer program corresponding to the algorithm of the cancer early screening detection model constructed according to the method according to any one of claims 6-8.

10. A cancer early screening detection device, characterized in that: The device comprises: (1) Input module: The input module is configured to input the base sequence set data of the 3' end motif of the plasma sample of the test subject; the base sequence set includes the following base sequence set coded as D1 to D126: (2) a processing module, wherein the processing module is configured to input the input base sequence set data into the cancer early screening detection model established in claim 6 for processing, or to perform the following processing: compare the data with a control reference value, and when the base sequence in the data is an up-regulated base sequence in Table A, the content of the base sequence or the proportion of the base sequence in all fragments significantly increases, and when the base sequence in the data is a down-regulated base sequence in Table A, the content of the base sequence or the proportion of the base sequence in all fragments significantly decreases, it indicates a high risk of cancer; otherwise, it indicates a low risk of cancer; and (3) An output module, wherein the output module is configured to output the cancer early screening test result.

Images