Method for scoring distribution deviation degree of tail end feature sequences of to-be-detected sample and contrast sample

Abstract

The invention relates to a method for scoring the distribution deviation degree of tail end feature sequences of a to-be-detected sample and a control sample. According to the method provided by the invention, the deviation of feature distribution of each terminal sequence of cfDNA of a sample to be detected relative to normal people is counted, the proportion of the deviation of each terminal sequence in the sum of the deviation of all the terminal sequences is calculated, and finally, the deviation proportions are accumulated (weighted summation) to obtain the overall deviation, so that the accuracy of the whole deviation is improved. Correcting the frequency of each tail end characteristic sequence in the cfDNA fragment of the sample to be detected by taking the average value of the frequency of each tail end characteristic sequence in the cfDNA fragment or all the cfDNA fragments in the preset length range of the healthy sample as a baseline to obtain a corrected frequency of each tail end characteristic sequence in the cfDNA fragment of the sample to be detected, the interference of different experimental conditions, experimental consumables, different sample collection, different transportation conditions and the like on the deviation of the terminal sequence feature distribution of the cfDNA of the whole to-be-detected sample relative to normal people is eliminated, so that the accuracy of determining the sample source is further improved, and the cancer screening quality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bioinformatics, and specifically, to a scoring method for the deviation degree of the end characteristic sequence distribution between a sample to be tested and a control sample. Background Art

[0002] Traditional cancer screening methods, such as imaging examinations, tumor marker screenings, and histopathological examinations, although they can achieve certain cancer screening effects, all have certain limitations in early cancer screening. For example, imaging examinations are affected by the development level of tumor lesions, some tumor markers are not abnormal in the early stage, and histopathological examination is an invasive examination method.

[0003] Detection based on cell-free DNA (cfDNA) in plasma, as a non-invasive liquid biopsy technology, has been widely used in cancer screening, efficacy evaluation, and prognosis evaluation in recent years. In recent studies, liquid biopsy based on the end sequence characteristics (End Motif) of cfDNA has shown great potential in cancer screening. Research shows that when cells apoptose, the caspase-mediated nucleic acid cleavage enzyme sites are usually at the junctions, so genomic DNA is cleaved into free DNA fragments and released into the blood for further cleavage to form cfDNA. Research shows that the cleavage sites of different nucleic acid cleavage enzymes have biases. For example, the DNASE1L3 cleavage enzyme tends to produce C-terminal sequence fragments; the DNASE1 cleavage enzyme tends to produce T-terminal sequence fragments; the DFFB cleavage enzyme tends to produce A-terminal sequence fragments. Research shows that there are different end sequence characteristics between cancer patients and normal people, and the end sequence characteristic diversity (Motifs diversity score, MDS) can be obtained based on the algorithm of information entropy (see Formula 1) to achieve the purpose of cancer screening and make up for the limitations of existing traditional screening methods. Formula 1:

[0004]

[0005] However, there are individual differences in actual operations. The differences in different experimental conditions and experimental consumables will affect the information entropy of samples. This may be because different sample collection and transportation conditions will affect the activity of nucleases (for example, EDTA blood collection tubes will inhibit the activity of DNases). Different enzyme cleavage sites have biases, thus affecting the distribution of the final cfDNA terminal characteristic sequences. At the same time, the research of Han DSC, et al. (The Biology of Cell-free DNA Fragmentation and the Roles of DNASE1, DNASE1L3, and DFFB) found that apoptosis is a conservative and orderly process. Different enzymes carry out DNA cleavage at different stages, generating DNA fragments of different lengths, and the base biases of these enzymes are different. Therefore, there are differences in the terminal characteristics of DNA fragments of different lengths. Moreover, information entropy only considers the frequency of individual terminal sequence characteristics in the terminal characteristic set, without considering the relationship between various terminal sequence characteristics in the terminal characteristic set. For example, for sets A, B, C, D and set B, C, A, D, although their information entropies are the same, there is still an order relationship between the two. Therefore, information entropy has certain limitations in cancer screening.

[0006] Therefore, there is an urgent need to develop a method that is less affected by experimental conditions and experimental consumables and can more accurately determine the degree of deviation of the terminal characteristic distribution of cfDNA between the test sample and the healthy sample, providing a reference for further cancer screening. Summary of the Invention

[0007] The present invention aims to solve at least one of the technical problems in the related art to some extent. For this reason, an object of the present invention is to provide a method for determining a score value of the overall deviation degree of the terminal sequence characteristic distribution of cfDNA of a test sample relative to a control sample. Using the frequency of the terminal sequences of normal samples as a baseline, a score value (Endmotif score, EMS) of the deviation degree relative to the terminal distribution of the control sample is calculated. And innovatively group the DNA fragments according to their lengths, and respectively count the characteristic differences and EMS under different DNA fragment lengths.

[0008] The first aspect of the present invention provides a method for determining a score value of the overall deviation degree of the terminal sequence characteristic distribution of cfDNA of a test sample relative to a control sample. According to an embodiment of the present invention, the method includes:

[0009] (1) Obtain the cfDNA of the test sample and the control sample, and perform sequencing processing on the cfDNA of the test sample and the control sample respectively to obtain sequencing reads. Among them, the control sample is a blood / serum sample derived from multiple healthy people;

[0010] (2) Obtain the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp, or 7 bp, or 8 bp at the 5'-end of the sequencing reads respectively; and / or the base information of 1 bp each, or 2 bp each, or 3 bp each, or 4 bp each, or 5 bp each, or 6 bp each at the upstream and downstream of the 5'-end of the breakpoint of the sequencing reads of each cfDNA, as the terminal characteristic sequence of cfDNA, wherein the upstream and downstream of the 5'-end of the obtained breakpoint of the sequencing reads are determined according to the position of the breakpoint on the reference genome;

[0011] (3) For the sequencing data of cfDNA of the test sample and the control sample, obtain the length of each cfDNA fragment and the corresponding terminal characteristic sequence, and count the number and proportion of each terminal characteristic sequence under different cfDNA fragment lengths, to obtain the terminal sequence characteristic distribution of cfDNA of the test sample and the control sample;

[0012] (4) Based on the terminal sequence characteristic distribution of cfDNA of the test sample and the control sample, obtain the frequency P of each terminal characteristic sequence in cfDNA fragments or all cfDNA fragments within a predetermined length range of the test sample, and the average value of the frequencies of each terminal characteristic sequence in cfDNA fragments or all cfDNA fragments within a predetermined length range of multiple control samples

[0013] (5) Using the average value as the baseline, correct the frequency P of each terminal characteristic sequence in the cfDNA fragments of the test sample, obtain the deviation degree of each terminal characteristic sequence in the cfDNA fragments of the test sample from the corresponding terminal characteristic sequence of the cfDNA fragments of the control sample, calculate the proportion of the deviation degree of each terminal characteristic sequence in the sum of the deviation degrees of all terminal characteristic sequences, and accumulate the proportion of the deviation degree, so as to obtain the scoring value of the overall deviation degree of the terminal sequence characteristic distribution of cfDNA of the test sample relative to the control sample.

[0014] The existing algorithm based on information entropy can obtain the diversity of terminal sequence features (MDS) to achieve the purpose of cancer screening. This algorithm is to statistically analyze the diversity of each terminal sequence feature of cfDNA in the sample and determine whether each feature is biased or evenly distributed. However, in actual operation, there are individual differences, and the differences in different experimental conditions and experimental consumables will affect the information entropy of the sample. Moreover, because different sample collection and transportation conditions will affect the activity of nucleases. And different enzyme cleavage sites have biases, thus affecting the distribution of the final cfDNA terminal characteristic sequences. In addition, information entropy only considers the frequency of individual terminal sequence features in the terminal feature set, without considering the relationship between various terminal sequence features in the terminal feature set. For example, for the sets A, B, C, D and the set B, C, A, D, although their information entropies are the same, there is still an order relationship between the two. Therefore, information entropy has certain limitations in cancer screening.

[0015] The inventors creatively discovered a new method for determining the score value of the overall deviation degree of the terminal sequence feature distribution of cfDNA in a test sample relative to a control sample. By statistically analyzing the deviation of the distribution of each terminal sequence feature of cfDNA in the test sample relative to the normal population, calculating the proportion of the deviation degree of each terminal characteristic sequence in the sum of the deviation degrees of all terminal characteristic sequences, and accumulating (weighted summing) these deviation proportions to obtain the overall deviation. Using the average value of the frequency of each terminal characteristic sequence in the cfDNA fragments within the predetermined length range of healthy samples or all cfDNA fragments as the baseline, the frequency of the corresponding terminal characteristic sequence in the cfDNA fragments of the test sample is corrected to exclude the interference of different experimental conditions, experimental consumables, different sample collections, different transportation conditions, etc. on the deviation of the terminal sequence feature distribution of the entire test sample cfDNA relative to the normal population, thereby further improving the accuracy of determining the sample source and the quality of cancer screening. Moreover, the screening performance of the method provided by the present invention is superior to the algorithm of information entropy. It should be noted that the method provided by the present invention for determining the score value of the overall deviation degree of the terminal sequence feature distribution of cfDNA in a test sample relative to a control sample, and the obtained EMS values of these test samples can also be used as experimental data for further scientific research.

[0016] According to an embodiment of the present invention, the predetermined length range includes at least one of a first nucleosome fragment size range, a fragment length range of a part of the first nucleosome fragment size, a second nucleosome fragment size range, and a fragment length range of a part of the second nucleosome fragment size. When there are multiple predetermined length ranges, there is no overlap between each predetermined length range. The definitions of the first nucleosome and the second nucleosome are from a published article (Circulating cell-free DNA fragmentation is a stepwise and conserved process linked to apoptosis. BMC Biology, (2023) 21(1), 253), specifically see Figure 2 。

[0017] According to an embodiment of the present invention, the first nucleosome fragment size range is 50 - 250 bp in length.

[0018] According to an embodiment of the present invention, the fragment length range of a part of the first nucleosome fragment size is 50 - 150 bp in length.

[0019] According to an embodiment of the present invention, the second nucleosome fragment size range is greater than 250 bp and less than or equal to 450 bp in length.

[0020] According to an embodiment of the present invention, the fragment length range of a part of the second nucleosome fragment size is 259 bp - 439 bp in length.

[0021] According to an embodiment of the present invention, when calculating the deviation degree between each end characteristic sequence in the cfDNA fragments of the test sample and each end characteristic sequence in the cfDNA fragments of the control sample, the selected length ranges of the cfDNA fragments of the test sample and the control sample are the same.

[0022] According to an embodiment of the present invention, the calculation formula for the scoring value EMS of the overall deviation degree of the end sequence characteristics distribution of the cfDNA of the test sample relative to the control sample is:

[0023]

[0024] where i represents a certain end characteristic sequence;

[0025] Pi is the proportion of the end characteristic sequence i in the cfDNA fragments of the test sample among all end characteristic sequences;

[0026] is the average value or median value of the proportion of the end characteristic sequence i in all end characteristic sequences among multiple control samples;

[0027] m is the nth power of 4, where n is the number of bases in the terminal characteristic sequence of cfDNA.

[0028] According to an embodiment of the present invention, the method further includes:

[0029] Based on the terminal sequence characteristic distributions of cfDNA of the sample to be tested and the control sample, obtaining the frequency P of each terminal characteristic sequence in the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size of the sample to be tested, and the average value of the frequencies of each terminal characteristic sequence in the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size of multiple control samples

[0030] Using step (5) to obtain the scoring value EMS of the deviation degree of the terminal sequence characteristic distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size of the sample to be tested.

[0031] According to an embodiment of the present invention, the method further includes:

[0032] Based on the terminal sequence characteristic distributions of cfDNA of the sample to be tested and the control sample, obtaining the frequency P of each terminal characteristic sequence in the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of the sample to be tested, and the average value of the frequencies of each terminal characteristic sequence in the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of multiple control samples

[0033] Using step (5) to obtain the scoring value EMS of the deviation degree of the terminal sequence characteristic distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of the sample to be tested.

[0034] According to an embodiment of the present invention, step (5) further includes:

[0035] ⅰ. Obtaining the scoring value EMS of the deviation degree of the terminal sequence characteristic distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size of the sample to be tested;

[0036] ⅱ. Obtaining the scoring value EMS of the deviation degree of the terminal sequence characteristic distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of the sample to be tested;

[0037] iii. Based on the EMS values obtained in i and ii, through a linear regression model, in order to obtain the deviation degree score value integration value of the end sequence characteristics distribution of cfDNA fragments in a part of the fragment length range of the first nucleosome fragment size range or the first nucleosome fragment size, and the second nucleosome fragment size range or a part of the fragment length range of the second nucleosome fragment size.

[0038] According to an embodiment of the present invention, the linear regression model is:

[0039] EMS_Merge = a * EMS_mono + b * EMS_di

[0040] Wherein, EMS_mono is the deviation degree score value of the end sequence characteristics distribution of cfDNA fragments in a part of the fragment length range of the first nucleosome fragment size range or the first nucleosome fragment size, and a is the weight coefficient of EMS_mono;

[0041] EMS_di is the deviation degree score value of the end sequence characteristics distribution of cfDNA fragments in a part of the fragment length range of the second nucleosome fragment size range or the second nucleosome fragment size, and b is the weight coefficient of EMS_di;

[0042] EMS_Merge is the integration value of the deviation degree score value EMS_mono of the end sequence characteristics distribution of cfDNA fragments in a part of the fragment length range of the first nucleosome fragment size range or the first nucleosome fragment size and the deviation degree score value EMS_di of the end sequence characteristics distribution of cfDNA fragments in a part of the fragment length range of the second nucleosome fragment size range or the second nucleosome fragment size.

[0043] According to an embodiment of the present invention, the weight coefficients a and b are the optimal weight coefficients determined by using a machine learning algorithm.

[0044] According to an embodiment of the present invention, the machine learning algorithm includes the exhaustive method and linear fitting.

[0045] According to an embodiment of the present invention, in step (2), the base information of 1bp, or 2bp, or 3bp, or 4bp, or 5bp, or 6bp, or 7bp, or 8bp at the 5' end of the sequencing read is obtained as the end characteristic sequence of cfDNA.

[0046] According to an embodiment of the present invention, the test sample and the control sample adopt the same sample collection and preservation method and the same sequencing platform.

[0047] According to an embodiment of the present invention, the sample to be tested is derived from a suspected cancer patient.

[0048] According to an embodiment of the present invention, the number of healthy people from which the control sample is derived is not less than 50. The control sample meets this condition, which can further improve the credibility of the value.

[0049] The second aspect of the present invention provides a method for determining the source of a sample to be tested. According to an embodiment of the present invention, the method includes:

[0050] Using the method described in the first aspect, obtaining a scoring value EMS of the overall deviation degree of the terminal sequence characteristics of cfDNA of the sample to be tested relative to the control sample, and determining the source of the sample to be tested based on the scoring value EMS,

[0051] wherein, the closer the scoring value EMS is to 1, it indicates that the overall deviation of the terminal sequence characteristics of cfDNA of the sample to be tested from the terminal sequence characteristics of cfDNA of the control sample is smaller, and the closer the scoring value EMS is to 0, it indicates that the overall deviation of the terminal sequence characteristics of cfDNA of the sample to be tested from the terminal sequence characteristics of cfDNA of the control sample is larger.

[0052] According to an embodiment of the present invention, the threshold of the scoring value EMS for the sample to be tested being a cancer sample is determined by a predetermined specificity.

[0053] The third aspect of the present invention provides a method for determining the source of a sample to be tested. According to an embodiment of the present invention, the method includes:

[0054] S1. Using steps (1)-(3) in the method described in the first aspect to obtain the terminal sequence characteristics distributions of cfDNA of the sample to be tested and the control sample;

[0055] S2. Based on the terminal sequence characteristics distributions of cfDNA of the sample to be tested and the control sample, performing a difference test, and selecting the significantly different terminal sequences as features;

[0056] S3. Constructing a prediction model based on the features screened in step S2 to determine the probability that the sample to be tested is a cancer sample,

[0057] The prediction model is obtained by constructing a machine learning model based on the terminal sequence characteristics distributions of cfDNA of known cancer samples and known healthy samples.

[0058] The fourth aspect of the present invention provides a method for determining the source of a sample to be tested. According to an embodiment of the present invention, the method includes:

[0059] 1) Obtain the end sequence feature distributions of cfDNA of the sample to be tested and the control sample using steps (1)-(3) in the method described in the first aspect;

[0060] 2) Based on the end sequence feature distributions of cfDNA of the sample to be tested and the control sample, perform a difference test, and select the significantly different end sequences as features.

[0061] 3) Based on the features screened in step 2) and the scoring value EMS obtained by the method described in the first aspect, construct a prediction model to determine the probability that the sample to be tested is a cancer sample.

[0062] The prediction model is obtained by constructing a machine learning model based on the end sequence feature distributions of cfDNA of known cancer samples and known healthy samples.

[0063] The fifth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored. According to the embodiments of the present invention, when the computer program is executed by a processor, it implements the method for determining the scoring value of the overall deviation degree of the end sequence feature distribution of cfDNA of the sample to be tested relative to the control sample described in the first aspect or the steps in the method for determining the source of the sample to be tested described in the second, third, or fourth aspect.

[0064] The sixth aspect of the present invention provides an electronic device. According to the embodiments of the present invention, the electronic device includes the computer-readable storage medium described in the fourth aspect; and

[0065] One or more processors for executing the program in the computer-readable storage medium.

[0066] The present invention provides a method for determining a score value representing the overall deviation degree of the terminal sequence feature distribution of cfDNA in a sample to be tested relative to a control sample. Using the frequency of the terminal sequences of normal samples as a baseline, a score value (Endmotif score, EMS) representing the deviation degree relative to the terminal distribution of the control sample is calculated. In the formula for calculating EMS, the closer the EMS value is to 1, the smaller the difference between the feature and the normal population; the more the EMS value deviates from 1 (the closer it is to 0), the greater the difference between the feature and the normal population, that is, the greater the probability that the sample to be tested is a cancer sample. And the inventor innovatively groups DNA fragments according to their lengths, and respectively counts the feature differences and EMS values under different DNA fragment lengths. For example, the EMS value of the terminal sequence features of cfDNA fragments within the short fragment size range (50bp - 150bp), the EMS value of the terminal sequence features of cfDNA fragments within the first nucleosome fragment size range (50bp - 250bp), and the EMS value of the terminal sequence features of cfDNA fragments within the second nucleosome fragment size range (259bp - 439bp). And creatively integrates the EMS value of the terminal sequence features of cfDNA fragments within the first nucleosome fragment size range (50bp - 250bp) with the EMS value of the terminal sequence features of cfDNA fragments within the second nucleosome fragment size range (259bp - 439bp), thereby further improving the cancer screening performance. At the same time, the inventor also conducts independent verification on other species (dogs).

[0067] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be understood through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The above and / or additional aspects and advantages of the present invention will become apparent and be readily understood from the following description of embodiments in conjunction with the accompanying drawings, in which:

[0069] Figure 1 A schematic diagram showing the terminal sequence features of plasma cfDNA fragments;

[0070] Figure 2 A schematic diagram showing the definition of the first nucleosome fragment size range and the second nucleosome fragment size range;

[0071] Figure 3 A distribution diagram showing the MDS values of the diversity of the terminal sequence features of cfDNA fragments within the first nucleosome fragment size range of healthy populations using different blood collection tubes (EDTA blood collection tube and Streck blood collection tube) in Example 2;

[0072] Figure 4Shows the distribution of the EMS values of the end - sequence characteristics of DNA fragments within the first nucleosome fragment size range for healthy individuals using different blood collection tubes (EDTA blood collection tube and Streck blood collection tube) in Example 2;

[0073] Figure 5 Shows the distribution of the EMS values of the end - sequence characteristics of all cfDNA fragments in Example 3 between normal individuals and cancer patients;

[0074] Figure 6 Shows the ROC curve graph of the EMS values and MDS values of the end - sequence characteristics of all cfDNA fragments in Example 3;

[0075] Figure 7 Shows the distribution of the EMS values of the end - sequence characteristics of cfDNA fragments within the short - fragment size range (50bp - 150bp) in Example 4 between normal individuals and cancer patients;

[0076] Figure 8 Shows the ROC curve graph of the EMS values and MDS values of the end - sequence characteristics of cfDNA fragments within the short - fragment size range (50bp - 150bp) in Example 4;

[0077] Figure 9 Shows the distribution of the EMS values of the end - sequence characteristics of cfDNA fragments within the first nucleosome fragment size range (50bp - 250bp) in Example 5 between normal individuals and cancer patients;

[0078] Figure 10 Shows the ROC curve graph of the EMS values and MDS values of the end - sequence characteristics of cfDNA fragments within the first nucleosome fragment size range (50bp - 250bp) in Example 5;

[0079] Figure 11 Shows the distribution of the EMS values of the end - sequence characteristics of cfDNA fragments within the second nucleosome fragment size range (259bp - 439bp) in Example 6 between normal individuals and cancer patients;

[0080] Figure 12 Shows the ROC curve graph of the EMS values and MDS values of the end - sequence characteristics of cfDNA fragments within the second nucleosome fragment size range (259bp - 439bp) in Example 6;

[0081] Figure 13 Shows the ROC curve graph of the EMS_Merge value after integrating the EMS values of the end - sequence characteristics of cfDNA fragments within the first and second nucleosome fragment size ranges in Example 7;

[0082] Figure 14Displays a scatter plot of the associated distribution of EMS_Merge values with EMS_mono values and EMS_di values in Example 7;

[0083] Figure 15 Displays the ROC curve graphs of EMS values and MDS values of the terminal sequence characteristics of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome in the independent validation set of plasma samples from 89 dogs in Example 8, and the ROC curve graph of the integrated EMS_Merge value;

[0084] Figure 16 Displays the ROC curve graphs of the training set and the validation set after feature screening and modeling based on machine learning methods in Example 9.

[0085] Figure 17 Displays the ROC curve graphs of the training set and the validation set after feature screening based on machine learning methods and modeling in combination with EMS values in Example 10. Detailed implementation mode

[0086] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as a limitation of the present invention.

[0087] It should be noted that the terms "first" and "second" are only used for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. Further, in the description of the present invention, unless otherwise stated, the meaning of "plurality" is two or more.

[0088] In the ranges disclosed herein, the endpoints and any values are not limited to the exact ranges or values, and these ranges or values should be understood to include values close to these ranges or values. For numerical ranges, between the endpoint values of each range, between the endpoint values of each range and individual point values, and between individual point values, they can be combined with each other to obtain one or more new numerical ranges, and these numerical ranges should be regarded as specifically disclosed herein.

[0089] In this article, the term "comprising" or "including" is an open expression, that is, it includes the content specified by the present invention, but does not exclude other aspects of the content.

[0090] In this article, the terms "optionally", "optional" or "option" generally mean that the subsequent events or conditions may or may not occur, and this description includes the cases where such events or conditions occur, as well as the cases where such events or conditions do not occur.

[0091] As used herein, the term "and / or" encompasses all combinations of the items connected by the term, and each combination should be considered as having been separately listed herein. For example, "A and / or B" encompasses "A", "A and B", and "B". For example, "A, B, and / or C" encompasses "A", "B", "C", "A and B", "A and C", "B and C", and "A and B and C".

[0092] According to a specific embodiment of the present invention, the present invention provides a method for determining a scoring value of the overall deviation degree of the terminal sequence feature distribution of cfDNA in a test sample relative to a control sample, comprising:

[0093] (1) Obtaining cfDNA of the test sample and the control sample, and respectively performing sequencing processing on the cfDNA of the test sample and the control sample to obtain sequencing reads, wherein the control sample is a sample derived from multiple healthy individuals;

[0094] (2) Respectively obtaining the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp, or 7 bp, or 8 bp at the 5' end of the sequencing reads; and / or the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp upstream and downstream of the 5' end of the sequencing read break point of each cfDNA as the terminal characteristic sequence of the cfDNA, wherein the upstream and downstream of the 5' end of the obtained sequencing read break point are determined according to the position of the break point on the reference genome;

[0095] (3) For the sequencing data of the cfDNA of the test sample and the control sample, obtaining the length of each cfDNA fragment and the corresponding terminal characteristic sequence, and counting the number and proportion of each terminal characteristic sequence at different cfDNA fragment lengths to obtain the terminal sequence feature distribution of the cfDNA of the test sample and the control sample;

[0096] (4) Based on the terminal sequence feature distribution of the cfDNA of the test sample and the control sample, obtaining the frequency P of each terminal characteristic sequence in the cfDNA fragments within a predetermined length range or all cfDNA fragments of the test sample, and the average value of the frequencies of each terminal characteristic sequence in the cfDNA fragments within a predetermined length range or all cfDNA fragments of multiple control samples

[0097] (5) Using the average value As a baseline, the frequency P of each end characteristic sequence in the cfDNA fragments of the sample to be tested is corrected to obtain the deviation degree of each end characteristic sequence in the cfDNA fragments of the sample to be tested from the corresponding end characteristic sequence of the cfDNA fragments of the control sample. The proportion of the deviation degree of each end characteristic sequence in the sum of the deviation degrees of all end characteristic sequences is calculated, and the proportion of the deviation degree is accumulated to obtain a score value of the overall deviation degree of the end sequence characteristics distribution of the cfDNA of the sample to be tested relative to the control sample.

[0098] Studies have shown that there are differences in the expression of nucleic acid shearing enzymes between cancer samples and healthy samples, resulting in different nucleic acid shearing enzymes having a preference for shearing cfDNA fragments of cancer samples and healthy samples, thus resulting in differences in the end sequence characteristics distribution between cancer samples and healthy samples. The method provided by the present invention, by statistically analyzing the deviation of the end sequence characteristics distribution of the cfDNA of the sample to be tested relative to the normal population, calculating the proportion of the deviation degree of each end characteristic sequence in the sum of the deviation degrees of all end characteristic sequences, and accumulating (weighted summing) these deviation proportions to obtain the overall deviation. Using the average value of the frequency of each end characteristic sequence in the cfDNA fragments within the predetermined length range or all cfDNA fragments of the healthy sample as the baseline, the frequency of the corresponding end characteristic sequence in the cfDNA fragments of the sample to be tested is corrected, excluding the interference of different experimental conditions, experimental consumables, different sample collections, different transportation conditions, etc. on the deviation of the end sequence characteristics distribution of the cfDNA of the entire sample to be tested relative to the normal population, thereby further improving the accuracy of determining the sample source and improving the quality of cancer screening.

[0099] It should be noted that the method for obtaining the score value of the overall deviation degree of the end sequence characteristics distribution of the cfDNA of the sample to be tested relative to the control sample provided by the present invention has no particular limitation on the sequencing platform and sequencing depth used for sequencing processing. Any sequencing platform known or commonly used in the art can be applied to the method of the present invention, and the sequencing depth can be the conventional sequencing depth for sequencing cfDNA in the art.

[0100] According to a specific embodiment of the present invention, the sample to be tested or the control sample can be a blood or serum sample derived from a mammal.

[0101] It should be noted that for the sequencing data of the cfDNA of the sample to be tested and the control sample obtained, it further includes a conventional denoising and data filtering process to obtain high-quality aligned reads.

[0102] According to a specific embodiment of the present invention, in step (2) of the above method provided by the present invention, the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp, or 7 bp, or 8 bp at the 5'-end of the sequencing read is obtained, and used as the terminal characteristic sequence of cfDNA. In step (2) of the method of the present invention, when obtaining the base information of 1 bp at the 5'-end of the sequencing read as the terminal characteristic sequence of cfDNA, only the base information of 1 bp at the 5'-end is extracted for each sequencing read. That is to say, in the same method, the base information of a certain length at the 5'-end of the sequencing read extracted must be of a unified length, rather than a mixed length.

[0103] When extracting the base information of 4 bp at the 5'-end of the sequencing read as the terminal characteristic sequence of cfDNA, since the base at each position can be A / T / C / G, there will be 4 4 = 256 kinds of terminal characteristic sequences.

[0104] According to another specific embodiment of the present invention, in step (2) of the above method provided by the present invention, the base information of 1 bp each, or 2 bp each, or 3 bp each, or 4 bp each, or 5 bp each, or 6 bp each at the upstream and downstream of the 5'-end of the breakpoint of the sequencing read of each cfDNA is obtained as the terminal characteristic sequence of cfDNA, where the upstream and downstream of the 5'-end of the breakpoint of the sequencing read obtained are determined according to the position of the breakpoint on the reference genome.

[0105] According to the alignment result of the cfDNA sequencing read and the reference genome, the corresponding position of the 5'-end of a certain sequencing read on the genome is located, and the base information of 1 bp each, or 2 bp each, or 3 bp each, or 4 bp each, or 5 bp each, or 6 bp each at the upstream and downstream of this position is taken and connected as the terminal sequence of this read. When extracting the base information of 2 bp each at the upstream and downstream of the breakpoint of cfDNA as the terminal characteristic sequence of cfDNA, since the base at each position can be A / T / C / G, there will be 4 4 = 256 kinds of terminal characteristic sequences.

[0106] It should be noted that the control samples are blood / serum samples from multiple healthy individuals. These healthy individuals should be at least 50 to ensure that the average value of the frequency of each terminal characteristic sequence in the cfDNA fragments or all cfDNA fragments within the predetermined length range of the obtained control samples is relatively accurate, and the errors between healthy individuals are also excluded.

[0107] The above-mentioned "cumulative proportion of the deviation degree" refers to the weighted sum of the proportion of the deviation degree of each terminal characteristic sequence to the sum of the deviation degrees of all terminal characteristic sequences.

[0108] According to a specific embodiment of the present invention, the predetermined length range includes at least one of the first nucleosome fragment size range, a partial fragment length range within the first nucleosome fragment size, the second nucleosome fragment size range, and a partial fragment length range within the second nucleosome fragment size. When there are multiple predetermined length ranges, there is no overlap between each predetermined length range. The definitions of the first nucleosome and the second nucleosome are from the published article (Circulating cell-free DNA fragmentation is a stepwise and conserved process linked to apoptosis. BMC Biology, (2023) 21(1), 253), specifically see Figure 2 .

[0109] According to a specific embodiment of the present invention, the first nucleosome fragment size range is 50 - 250 bp in length.

[0110] According to a specific embodiment of the present invention, a partial fragment length range within the first nucleosome fragment size can be 50 - 150 bp in length. It can also be any length range within the first nucleosome fragment size range of 50 - 250 bp, such as 50 - 100 bp, 100 - 200 bp, 150 - 200 bp, 150 - 250 bp, etc. These length ranges are also covered within the protection scope of the partial fragment length range within the first nucleosome fragment size of the present invention.

[0111] According to a specific embodiment of the present invention, the second nucleosome fragment size range is greater than 250 bp and less than or equal to 450 bp in length.

[0112] According to another specific embodiment of the present invention, a partial fragment length range within the second nucleosome fragment size can be 259 bp - 439 bp in length. It can also be any length range within the second nucleosome fragment size range greater than 250 bp and less than or equal to 450 bp, such as 259 - 300 bp, 259 - 350 bp, 300 - 400 bp, 350 - 450 bp, etc. These length ranges are also covered within the protection scope of the partial fragment length range within the second nucleosome fragment size of the present invention.

[0113] It should be noted that the predetermined length range can be the size range of the first nucleosome fragment, or a small range within it, such as a length of 50 - 150 bp; the predetermined length range can also be the size range of the second nucleosome fragment, or a small range within it, such as a length of 259 bp - 439 bp, or the predetermined length range can be a mixture of the size range of the first nucleosome fragment and the size range of the second nucleosome fragment, or a mixture of a small range within the size range of the first nucleosome fragment and a small range within the size range of the second nucleosome fragment. When the predetermined length range is multiple ranges, the end values between the ranges do not overlap. For example, the predetermined length range can be 50 - 150 bp and 259 - 439 bp in length, but it cannot be 50 - 150 bp and 140 - 250 bp (the length ranges overlap), nor can it be 259 - 350 bp and 340 - 450 bp (the length ranges overlap).

[0114] According to a specific embodiment of the present invention, when calculating the degree of deviation of each end characteristic sequence in the cfDNA fragments of the test sample from each end characteristic sequence in the cfDNA fragments of the control sample, the selected length ranges of the cfDNA fragments of the test sample and the control sample are the same. For example, when calculating the degree of deviation of each end characteristic sequence in the cfDNA fragments of the test sample from each end characteristic sequence in the cfDNA fragments of the control sample, the cfDNA fragments of the test sample and the control sample are both cfDNA fragments of the size of the first nucleosome fragment, or both cfDNA fragments of the size of the second nucleosome fragment, or both all cfDNA fragments.

[0115] According to a specific embodiment of the present invention, the calculation formula for the scoring value EMS of the overall degree of deviation of the end sequence characteristics of the cfDNA of the test sample relative to the control sample is as follows:

[0116]

[0117] Wherein, i represents a certain end characteristic sequence;

[0118] Pi is the proportion of the end characteristic sequence i in the cfDNA fragments of the test sample among all end characteristic sequences;

[0119] is the average or median value of the proportion of the end characteristic sequence i in all end characteristic sequences among multiple control samples;

[0120] m is 4 to the power of n, where n is the number of bases of the end characteristic sequence of cfDNA.

[0121] For example, when n is 4 and m is 256, the calculation formula for the scoring value EMS of the overall deviation degree of the terminal sequence feature distribution of cfDNA of the sample to be tested relative to the control sample is as follows:

[0122]

[0123] According to a specific embodiment of the present invention, the method further includes:

[0124] Based on the terminal sequence feature distributions of the cfDNA of the sample to be tested and the control sample, obtain the frequency P of each terminal feature sequence in the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range of the first nucleosome fragment size of the sample to be tested, and the average value of the frequencies of each terminal feature sequence in the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range of the first nucleosome fragment size of multiple control samples

[0125] Use step (5) to obtain the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range of the first nucleosome fragment size of the sample to be tested.

[0126] According to a specific embodiment of the present invention, the method further includes:

[0127] Based on the terminal sequence feature distributions of the cfDNA of the sample to be tested and the control sample, obtain the frequency P of each terminal feature sequence in the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range of the second nucleosome fragment size of the sample to be tested, and the average value of the frequencies of each terminal feature sequence in the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range of the second nucleosome fragment size of multiple control samples

[0128] Use step (5) to obtain the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range of the second nucleosome fragment size of the sample to be tested.

[0129] According to a specific embodiment of the present invention, step (5) further includes:

[0130] ⅰ. Obtain the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range of the first nucleosome fragment size of the sample to be tested;

[0131] ii. Obtain the scoring value EMS of the deviation degree of the end sequence feature distribution of the cfDNA fragments within a part of the fragment length range of the second nucleosome fragment size range or the second nucleosome fragment size of the sample to be tested;

[0132] iii. Based on the EMS values obtained in i and ii, through a linear regression model, in order to obtain the integrated value of the scoring value EMS of the deviation degree of the end sequence feature distribution of the cfDNA fragments within a part of the fragment length range of the first nucleosome fragment size range or the first nucleosome fragment size, and within a part of the fragment length range of the second nucleosome fragment size range or the second nucleosome fragment size.

[0133] According to a specific embodiment of the present invention, the linear regression model is:

[0134] EMS_Merge = a * EMS_mono + b * EMS_di

[0135] Wherein, EMS_mono is the scoring value of the deviation degree of the end sequence feature distribution of the cfDNA fragments within a part of the fragment length range of the first nucleosome fragment size range or the first nucleosome fragment size, and a is the weight coefficient of EMS_mono;

[0136] EMS_di is the scoring value of the deviation degree of the end sequence feature distribution of the cfDNA fragments within a part of the fragment length range of the second nucleosome fragment size range or the second nucleosome fragment size, and b is the weight coefficient of EMS_di;

[0137] EMS_Merge is the integrated value of the scoring value EMS_mono of the deviation degree of the end sequence feature distribution of the cfDNA fragments within a part of the fragment length range of the first nucleosome fragment size range or the first nucleosome fragment size and the scoring value EMS_di of the deviation degree of the end sequence feature distribution of the cfDNA fragments within a part of the fragment length range of the second nucleosome fragment size range or the second nucleosome fragment size.

[0138] According to a specific embodiment of the present invention, the weight coefficients a and b are the optimal weight coefficients determined by using a machine learning algorithm.

[0139] According to the embodiment of the present invention, the machine learning algorithm includes the exhaustive method and linear fitting.

[0140] According to a specific embodiment of the present invention, in step (2), obtain the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp, or 7 bp, or 8 bp at the 5' end of the sequencing read as the end characteristic sequence of the cfDNA.

[0141] According to a specific embodiment of the present invention, the test sample and the control sample adopt the same sample collection and preservation method and the same sequencing platform. For example, when sampling the test sample and the control sample respectively, the same specification of blood collection tubes and the same preservation conditions are used.

[0142] According to a specific embodiment of the present invention, the test sample is derived from a suspected cancer patient.

[0143] The present invention provides a method for determining the source of a test sample, the method comprising:

[0144] Using the method described above, obtain a scoring value EMS of the overall deviation degree of the terminal sequence characteristics distribution of cfDNA of the test sample relative to the control sample, and determine the source of the test sample based on the scoring value EMS.

[0145] Wherein, the closer the scoring value EMS is to 1, it indicates that the overall deviation of the terminal sequence characteristics distribution of cfDNA of the test sample from that of the control sample is smaller; the closer the scoring value EMS is to 0, it indicates that the overall deviation of the terminal sequence characteristics distribution of cfDNA of the test sample from that of the control sample is larger.

[0146] According to a specific embodiment of the present invention, the threshold value of the scoring value EMS of the test sample being a cancer sample is determined by a predetermined specificity. It should be noted that the threshold value of the scoring value EMS of the test sample being a cancer sample is not fixed, but is determined according to a predetermined specificity.

[0147] The present invention also provides another method for determining the source of a test sample, the method comprising:

[0148] S1. Obtain the terminal sequence characteristics distribution of cfDNA of the test sample and the control sample by using steps (1)-(3) in the method described above;

[0149] S2. Based on the terminal sequence characteristics distribution of cfDNA of the test sample and the control sample, perform a difference test, and select the significantly different terminal sequences as features;

[0150] S3. Construct a prediction model based on the features selected in step S2 to determine the probability that the test sample is a cancer sample.

[0151] The prediction model is obtained by constructing a machine learning model based on the terminal sequence characteristics distribution of cfDNA of known cancer samples and known healthy samples.

[0152] The present invention also provides another method for determining the source of a test sample, the method comprising:

[0153] 1) Obtain the end - sequence feature distributions of cfDNA of the sample to be tested and the control sample using steps (1)-(3) in the method described above.

[0154] 2) Based on the end - sequence feature distributions of cfDNA of the sample to be tested and the control sample, conduct a difference test, and select the end sequences with significant differences as features.

[0155] 3) Based on the features screened in step 2) and the scoring value EMS obtained by the method described above, construct a prediction model to determine the probability that the sample to be tested is a cancer sample.

[0156] The prediction model is obtained by constructing a machine - learning model based on the end - sequence feature distributions of cfDNA of known cancer samples and known healthy samples.

[0157] According to a specific embodiment of the present invention, the "end sequences with significant differences" are determined by t - test and / or rank - sum test, and the q - value is obtained by using FDR to correct the p - value. Among them, the end sequences with q - value < 0.01 can be considered as "end sequences with significant differences".

[0158] According to a specific embodiment of the present invention, in order to ensure the stability of the machine - learning model, the method of repeating a single model multiple times and calculating the mean of repeated modelings, or the method of integrating multiple models can be adopted for modeling.

[0159] According to a specific embodiment of the present invention, during the process of constructing the machine - learning model, the 10 - fold cross - validation method is used and repeated at least 100 times to determine the optimal parameters of the machine - learning model.

[0160] The solutions of the present disclosure will be explained below in conjunction with embodiments. Those skilled in the art will understand that the following embodiments are only used to illustrate the present disclosure and should not be regarded as limiting the scope of the present disclosure. For those not specified in the embodiments regarding specific technologies or conditions, they shall be carried out according to the technologies or conditions described in the literature in the field or according to the product instructions. For the reagents or instruments not specified by the manufacturer, they are all conventional products that can be obtained through commercial purchase.

[0161] Example 1:

[0162] Extract and construct the library for plasma cfDNA according to conventional general methods (e.g., methods of cfDNA extraction, end repair, addition of PolyA, adapter ligation, DNA purification, library amplification and purification), and obtain the cfDNA sequencing data of plasma samples by high-throughput sequencing (Illumina / MGI). Then, use common software (such as cutadapt and / or SOAPnuke) to filter the sequencing data, align it to the genome of the reference species, filter out the repetitive sequences, and retain the reads with alignment quality value greater than or equal to 30 and aligned to unique positions on the reference genome. According to Figure 1 's example, extract several bases (1 - 8 bp) at the 5' end or an equal number of bases upstream and downstream of the breakpoint (1 - 6 bp upstream and downstream of the breakpoint position) as Endmotif. Specifically, as in Figure 1 , obtaining the base information of 2 bp at the 5' end of the sequencing read refers to the base CG; obtaining the base information of 3 bp at the 5' end of the sequencing read refers to the base CGA; obtaining the base information of 4 bp at the 5' end of the sequencing read refers to the base CGAC. Obtaining the base information of 2 bp each upstream and downstream of the breakpoint of the sequencing read of the cfDNA refers to the base CGTC, that is, 2 bp are taken from each of the upstream and downstream, and the combined 4 bp of base information; obtaining the base information of 3 bp each upstream and downstream of the breakpoint of the sequencing read of the cfDNA refers to the base ACGTCG, that is, 3 bp are taken from each of the upstream and downstream, and the combined 6 bp of base information.

[0163] In this example, taking extracting 4 bases as Endmotif (the number of bases can be modified according to requirements) as an example. Obtain the length of each cfDNA fragment and the corresponding End motif. Finally, select different fragment lengths and count the number and proportion of the terminal characteristic sequences of this group of cfDNA.

[0164] Example 2:

[0165] (1) Analyze the samples of two groups of healthy people (Streck: 188 cases; EDTA: 92 cases) using different collection and preservation methods by the method in Example 1. And count the frequency of the terminal characteristic sequences of cfDNA fragments within the size range of the first nucleosome fragment or a part of the size range of the first nucleosome fragment. In this example, take the cfDNA fragments in the range of 50 bp - 250 bp as an example;

[0166] (2) Obtain the diversity value MDS of the terminal sequence characteristics of the sample based on the algorithm of information entropy (see Formula 1);

[0167] Formula 1:

[0168]

[0169] (3) Compare the MDS values of samples from two groups of healthy individuals using different collection and preservation methods, and use box plots to show the distribution of MDS values of the two groups of samples, as shown in Figure 3 . The results show that there are differences in MDS between the two groups of healthy individuals using different collection and preservation methods, indicating that the differences in collection and preservation methods introduce frequency differences in terminal characteristics, resulting in significant differences in information entropy. Therefore, measures need to be taken to correct systematic errors under different experimental conditions such as different experimental consumables, and develop new evaluation methods.

[0170] (4) Adopt the newly developed calculation method of the present invention (see Formula 2), that is, take the average value of the frequency of each terminal sequence of a group of healthy human plasma samples submitted for the same experiment as the baseline, and divide the terminal sequence frequency Pi of the plasma sample to be tested under the same conditions by the average value of the corresponding terminal sequence frequency of the above-mentioned group of healthy human plasma samples Sum up the calculated results after correction for all (256 in this example, m = 256) terminal sequences, and count the proportion of the corrected value of each terminal sequence in the total sum of the corrected values of all terminal sequences to obtain the corrected terminal sequence frequency. At the same time, take the logarithm of the corrected frequency, multiply it by the corrected frequency of the terminal sequence, and then divide it by the logarithm of the sum of the corrected values of the above 256 terminal sequences. Finally, sum up the calculated results of the 256 terminal sequences and take the negative value to calculate the scoring value EMS of the deviation degree relative to the terminal distribution of the control sample, and use box plots to show the distribution of EMS values of the two groups of samples, as shown in Figure 4 . The results show that after correction by the calculation method of the present invention, the EMS values of the two groups of healthy individuals using different collection and preservation methods are at the same level, indicating that the systematic error of terminal characteristics introduced by the differences in collection and preservation methods is corrected and eliminated.

[0171] Formula 2:

[0172]

[0173] Example 3:

[0174] (1) Determine the terminal sequence characteristics of all cfDNA fragments in plasma samples of cancer patients using the method in Example 1;

[0175] (2) Determine the terminal sequence characteristics of all cfDNA fragments in plasma samples of a group of healthy individuals using the method in Example 1;

[0176] (3) Use the EMS calculation method in Example 2 to obtain the scoring value (EMS) of the deviation degree of the end distribution of the plasma sample to be tested. The closer the EMS value is to 1, the smaller the difference between the characteristics and the normal population. Conversely, the greater the difference. See Figure 5 。

[0177] (4) Use the ROC curve evaluation method for the calculated EMS value to evaluate the end sequence characteristics of the plasma sample to be tested, so as to assist in determining the source of the plasma sample to be tested. The samples for evaluation are randomly selected according to a ratio of 3:2 to form the training set data (113 healthy people and 78 liver cancer patients) and the validation set data (75 healthy people and 51 liver cancer patients) respectively. See Table 1. The results show that in the training set data, the AUC value of the EMS method is 0.853, and the AUC value of the MDS method is 0.738. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.003105, Delong test). In the validation set data, the AUC value of the EMS method is 0.812, and the AUC value of the MDS method is 0.708. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.001337, Delong test). See Figure 6 。Based on the above results, it can be known that the scoring value (EMS) of the deviation degree of the end sequence distribution of all cfDNA fragments can be applied to cancer screening, and the screening performance is better than that of the MDS method.

[0178] Dataset Healthy people Liver cancer patients Total Training set 113 78 191 Validation set 75 51 126 Total 188 129 317

[0179] Table 1

[0180] Example 4:

[0181] (1) Use the method in Example 1 to determine the end sequence characteristics of cfDNA fragments within the short fragment size range of plasma samples from cancer patients. In this example, cfDNA fragments in the range of 50bp - 150bp are taken as an example;

[0182] (2) Use the method in Example 1 to determine the end sequence characteristics of cfDNA fragments within the short fragment size (50bp - 150bp) range of a group of healthy human plasma samples;

[0183] (3) Use the EMS calculation method in Example 2 to obtain the scoring value (EMS) of the deviation degree of the end characteristic frequency of the plasma sample. The closer the EMS value is to 1, the smaller the difference between the characteristics and the normal population. Conversely, the greater the difference. See Figure 7 。

[0184] (4) The ROC curve evaluation method is used to evaluate the end - sequence characteristics of the plasma sample to be tested based on the calculated EMS value, so as to assist in determining the source to which the plasma sample to be tested belongs. The cohort used for evaluation is the same as that in Example 3, as shown in Table 1. The results show that in the training set data, the AUC value of the EMS method is 0.805, and the AUC value of the MDS method is 0.719. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.003963, Delong test). In the validation set data, the AUC value of the EMS method is 0.792, and the AUC value of the MDS method is 0.69. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.00206, Delong test), see Figure 8 . Based on the above results, it can be known that the EMS value of the end - sequence of cfDNA fragments within the short - fragment size range, like that of all cfDNA fragments, can also be applied to cancer screening, and its screening performance is also better than that of the MDS method.

[0185] Example 5:

[0186] (1) The method in Example 1 is used to determine the end - sequence characteristics of cfDNA fragments within the first nucleosome fragment size range of plasma samples from cancer patients. In this example, cfDNA fragments in the range of 50bp - 250bp are taken as an example;

[0187] (2) The method in Example 1 is used to determine the end - sequence characteristics of cfDNA fragments within the first nucleosome fragment size (50bp - 250bp) range of a group of healthy human plasma samples;

[0188] (3) The EMS calculation method in Example 2 is used to obtain the scoring value (EMS) of the deviation degree of the end - distribution of the plasma sample to be tested. The closer the EMS value is to 1, the smaller the difference in characteristics from the normal population; conversely, the greater the difference, see Figure 9 .

[0189] (4) The ROC curve evaluation method is used to evaluate the end - sequence characteristics of the plasma sample to be tested based on the calculated EMS value, so as to assist in determining the source to which the plasma sample to be tested belongs. The cohort used for evaluation is the same as that in Example 3, as shown in Table 1. The results show that in the training set data, the AUC value of the EMS method is 0.851, and the AUC value of the MDS method is 0.748. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.0006459, Delong test). In the validation set data, the AUC value of the EMS method is 0.808, and the AUC value of the MDS method is 0.719. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.001224, Delong test), as shown in Figure 10 . From the above - mentioned results, it can be seen that the cancer screening performance of the end - sequence characteristics of cfDNA fragments within the first nucleosome fragment size range based on the EMS method is better than that of the MDS method.

[0190] Example 6:

[0191] (1) The method in Example 2 is used to determine the end - sequence characteristics of cfDNA fragments in the second nucleosome fragment size range of the plasma samples of cancer patients. In this example, the cfDNA fragments in the range of 259bp - 439bp are taken as an example;

[0192] (2) The method in Example 2 is used to determine the end - sequence characteristics of cfDNA fragments within the first nucleosome fragment size (259bp - 439bp) of a group of healthy human plasma samples;

[0193] (3) The EMS calculation method in Example 2 is used to obtain the scoring value (EMS) of the deviation degree of the end - distribution of the plasma sample to be tested. The closer the EMS value is to 1, the smaller the difference in characteristics from the normal population, and vice versa, as shown in Figure 11 .

[0194] (4) The ROC curve evaluation method is used to evaluate the end - sequence characteristics of the plasma sample to be tested based on the calculated EMS value, so as to assist in determining the source of the plasma sample to be tested. The cohort used for evaluation is the same as that in Example 3, as shown in Table 1. The results show that in the training set data, the AUC value of the EMS method is 0.841, and the AUC value of the MDS method is 0.696. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 1.189e - 05, Delong test). In the validation set data, the AUC value of the EMS method is 0.803, and the AUC value of the MDS method is 0.694. The Delong test results show that the AUC value of the EMS method is significantly higher than that of the MDS method (P - value = 0.000701, Delong test), as shown in Figure 12 . Based on the above results, it can be seen that in the range of the second nucleosome fragment size, the cancer screening performance of the end - sequence characteristics of cfDNA fragments based on the EMS method is better than that of the MDS method.

[0195] Example 7:

[0196] (1) Based on Examples 5 and 6, the algorithms based on the end - sequence characteristics of cfDNA fragments within the first and second nucleosome fragment sizes using EMS can be used for cancer screening, and their performance is better than that of the MDS algorithm. Therefore, through the linear regression model method, that is, EMS_Merge = a * EMS_mono + b * EMS_di (EMS_mono represents the end - sequence EMS value of cfDNA fragments within the first nucleosome fragment size, a represents its weight coefficient; EMS_di represents the end - sequence EMS value of cfDNA fragments within the second nucleosome fragment size, b represents its weight coefficient; EMS_Merge represents the integrated value of the end - sequence EMS values of cfDNA fragments within the first and second nucleosome fragment sizes), the end - sequence characteristics of cfDNA fragments within the first and second nucleosome fragment sizes are integrated, aiming to improve cancer screening performance.

[0197] (2) Using the linear regression model method, the optimal weight coefficients of the end - sequence EMS values of cfDNA fragments within the first and second nucleosome fragment sizes are obtained respectively, and the integrated value EMS_Merge can be obtained by substituting them into the formula.

[0198] (3) The integrated value (EMS_Merge) of the end sequence EMS values of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome was evaluated using the receiver operating characteristic (ROC) curve method to assess the end sequence characteristics of the plasma sample to be tested, so as to assist in determining the source of the plasma sample to be tested. The cohort used for evaluation was the same as in Example 3, as shown in Table 1. The results showed that in the training set data, the AUC value of EMS_Merge was 0.857, which was higher than the AUC values of the end sequence EMS values of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome (0.851, 0.841). In the validation set, the AUC value of EMS_Merge was 0.813, which was higher than the AUC values of the end sequence EMS values of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome (0.808, 0.803), as shown in Figure 13 . Among them, in the training set, at 100% specificity, EMS_Merge identified 2 and 4 more positive samples than EMS_mono and EMS_di respectively, as shown in Figure 14 . Based on the above results, it can be seen that integrating the end sequence EMS values of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome can improve the performance of cancer screening.

[0199] Example 8:

[0200] (1) Plasma samples from 89 dogs (11 cancerous dogs and 78 healthy dogs) were enrolled as an independent validation set. Based on the methods of Example 3, Example 4, and Example 5, the EMS values of the end sequences of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome of the enrolled dogs and the integrated EMS_Merge values were obtained respectively.

[0201] (2) The receiver operating characteristic (ROC) curve method was used to evaluate the end sequence characteristics of the plasma samples of the above dogs to assist in determining the source of the plasma sample to be tested. The results showed that the AUC values of the end sequence characteristics of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome of the dogs based on the EMS algorithm were 0.911 and 0.879 respectively, indicating that both could be used for cancer screening and their performance was better than that of the MDS algorithm, as shown in Figure 15 . The AUC value of EMS_Merge was 0.913, which was higher than the AUC values of the end sequence characteristics of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome, as shown in Figure 15 . This indicates that integrating the end sequence characteristics of cfDNA fragments within the size ranges of the first nucleosome and the second nucleosome can also improve the performance of cancer screening in the validation dataset of dogs.

[0202] Example 9:

[0203] (1) End sequence features of cfDNA fragments within the first nucleosome fragment size range obtained based on Example 5. First, screen out the end sequence features with significant differences between healthy and cancer populations, and then use machine learning methods such as Logistic Regression to model these significantly different end sequence features. Repeat the modeling 100 times, and extract the contribution degree of each feature in each modeling. Calculate the average value of the contribution degrees of each feature in 100 modelings, and screen out the top 20 features with the average model contribution degree as the features for the final modeling.

[0204] (2) Based on the 20 selected end sequence frequency features, use machine learning methods such as Logistic Regression for modeling. During the modeling process, use the 10-fold cross-validation method and repeat it 100 times to improve the detection accuracy and determine the optimal model parameters.

[0205] (3) The training set and validation set data for model training and validation are the same as those in Example 3, as shown in Table 1. Use the Receiver Operating Characteristic (ROC) evaluation method to evaluate the model performance, as shown in Figure 16 . The AUC value of the training set is 0.999, and the AUC value of the validation set is 0.91. The results show that the strategy of feature screening and modeling based on machine learning methods can be used for the prediction of cancer-derived samples.

[0206] Example 10

[0207] (1) Based on Example 5, obtain the End Distribution Deviation Score (EMS) of the test plasma sample within the first nucleosome fragment size range;

[0208] (2) Based on Example 9, obtain the top 20 end sequence frequency features with the average model contribution degree within the first nucleosome fragment size range;

[0209] (3) Based on the End Distribution Deviation Score (EMS) of the test plasma sample within the first nucleosome fragment size range and the 20 selected end sequence frequency features, use machine learning methods such as Logistic Regression for modeling. During the modeling process, use the 10-fold cross-validation method and repeat it 100 times to improve the detection accuracy and determine the optimal model parameters; (4) The training set and validation set data for model training and validation are the same as those in Example 3, as shown in Table 1. Use the Receiver Operating Characteristic (ROC) evaluation method to evaluate the model performance, as shown in Figure 17 . The AUC value of the training set is 0.983, and the AUC value of the validation set is 0.955. The results show that the strategy of modeling by combining the features screened by machine learning methods with the EMS value can be used for the prediction of cancer-derived samples, and the performance of the validation set has been improved.

[0210] In the description of this specification, the descriptions with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", "some implementation manners" or "some examples", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0211] Although the embodiments of the present invention have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for determining a score value of the overall deviation degree of the terminal sequence feature distribution of cfDNA in a sample to be tested relative to a control sample, characterized in that, Comprising: (1) Obtaining cfDNA of a sample to be tested and a control sample, respectively performing sequencing processing on the cfDNA of the sample to be tested and the control sample to obtain sequencing reads, wherein the control sample is a sample derived from multiple healthy individuals; (2) Respectively obtaining the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp, or 7 bp, or 8 bp at the 5'-end of the sequencing reads; and / or the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp upstream and downstream of the 5'-end of the sequencing read break point of each cfDNA as the terminal characteristic sequence of the cfDNA, wherein the upstream and downstream of the 5'-end of the sequencing read break point are determined according to the position of the 5'-end of the sequencing read break point on the reference genome; (3) For the sequencing data of the cfDNA of the sample to be tested and the control sample, obtaining the length of each cfDNA fragment and the corresponding terminal characteristic sequence, and counting the number and proportion of each terminal characteristic sequence under different cfDNA fragment lengths to obtain the terminal sequence characteristic distribution of the cfDNA of the sample to be tested and the control sample; (4) Based on the end sequence feature distributions of the cfDNA of the sample to be tested and the control samples, obtain the frequency P of each end feature sequence in the cfDNA fragments within a predetermined length range or all cfDNA fragments of the sample to be tested, and the average value of the frequencies of each end feature sequence in the cfDNA fragments within a predetermined length range or all cfDNA fragments of multiple control samples ; (5)Using the average value as a baseline, correct the frequency P of each end characteristic sequence in the cfDNA fragments of the sample to be tested, obtain the deviation degree of each end characteristic sequence in the cfDNA fragments of the sample to be tested from the corresponding end characteristic sequence of the cfDNA fragments of the control sample, calculate the proportion of the deviation degree of each end characteristic sequence in the sum of the deviation degrees of all end characteristic sequences, and accumulate the proportion of the deviation degree, so as to obtain the scoring value EMS of the overall deviation degree of the end sequence characteristics of the cfDNA of the sample to be tested relative to the cfDNA of the control sample.

2. The method according to claim 1, characterized in that, The predetermined length range includes at least one of the first nucleosome fragment size range, a part of the fragment length range in the first nucleosome fragment size, the second nucleosome fragment size range, and a part of the fragment length range in the second nucleosome fragment size, and when there are multiple predetermined length ranges, there is no overlap between the respective predetermined length ranges; Optionally, the first nucleosome fragment size range is 50 - 250 bp in length; Optionally, a part of the fragment length range in the first nucleosome fragment size is 50 - 150 bp in length; Optionally, the second nucleosome fragment size range is greater than 250 bp and less than or equal to 450 bp in length; Optionally, a part of the fragment length range in the second nucleosome fragment size is 259 bp - 439 bp in length.

3. The method according to claim 2, characterized in that, When calculating the deviation degree between each terminal characteristic sequence in the cfDNA fragment of the sample to be tested and the corresponding terminal characteristic sequence in the cfDNA fragment of the control sample, the selected length ranges of the cfDNA fragments of the sample to be tested and the control sample are the same.

4. The method according to claim 3, characterized in that The calculation formula for the scoring value EMS of the overall deviation degree of the terminal sequence characteristic distribution of the cfDNA of the sample to be tested relative to the control sample is: Wherein, i represents a certain terminal characteristic sequence; Pi is the proportion of the terminal characteristic sequence i in the cfDNA fragment of the sample to be tested among all terminal characteristic sequences; is the average or median value of the proportion of the terminal feature sequence i in all terminal feature sequences among multiple control samples; m is 4 to the power of n, where n is the number of bases of the terminal characteristic sequence of the cfDNA.

5. The method according to claim 4, characterized in that The method further includes: Based on the end - sequence characteristic distribution of cfDNA of the sample to be tested and the control sample, obtain the frequency P of each end - characteristic sequence in the cfDNA fragments within the first nucleosome fragment size range of the sample to be tested or a part of the fragment length range within the first nucleosome fragment size, and the average value of the frequencies of each end - characteristic sequence in the cfDNA fragments within the first nucleosome fragment size range of multiple control samples or a part of the fragment length range within the first nucleosome fragment size Using step (5) to obtain the scoring value EMS of the deviation degree of the terminal sequence characteristic distribution of the cfDNA fragment in the first nucleosome fragment size range or a part of the fragment length range in the first nucleosome fragment size of the sample to be tested.

6. The method according to claim 4, wherein The method further includes: Based on the terminal sequence feature distribution of cfDNA of the sample to be tested and the control sample, obtain the frequency P of each terminal feature sequence in the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of the sample to be tested, and the average value H of the frequencies of each terminal feature sequence in the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of multiple control samples. Use step (5) to obtain the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of the sample to be tested.

7. The method according to claim 4, characterized in that, Step (5) further includes: ⅰ. Obtain the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size of the sample to be tested. ⅱ. Obtain the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size of the sample to be tested. ⅲ. Based on the EMS values obtained in ⅰ and ⅱ, through a linear regression model, obtain the integrated value of the scoring value EMS of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size and the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size.

8. The method according to claim 7, characterized in that, The linear regression model is: EMS_Merge = a * EMS_mono + b * EMS_di Wherein, EMS_mono is the scoring value of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size, and a is the weight coefficient of EMS_mono. EMS_di is the scoring value of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size, and b is the weight coefficient of EMS_di. EMS_Merge is the integrated value of the scoring value EMS_mono of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the first nucleosome fragment size range or a part of the fragment length range within the first nucleosome fragment size and the scoring value EMS_di of the deviation degree of the terminal sequence feature distribution of the cfDNA fragments within the second nucleosome fragment size range or a part of the fragment length range within the second nucleosome fragment size.

9. The method according to claim 8, wherein The weight coefficients a and b are the optimal weight coefficients determined by using a machine learning algorithm. Optionally, the machine learning algorithm includes the exhaustive method and linear fitting.

10. The method according to claim 1, characterized in that, In step (2), obtain the base information of 1 bp, or 2 bp, or 3 bp, or 4 bp, or 5 bp, or 6 bp, or 7 bp, or 8 bp at the 5'-end of the sequencing reads as the terminal characteristic sequence of cfDNA.

11. The method according to claim 1, wherein The test sample is derived from a suspected cancer patient.

12. The method according to claim 1, characterized in that, The number of healthy people from whom the control sample is derived is not less than 50.

13. A method for determining the source of a sample to be tested, characterized in that, Including: Using the method described in any one of claims 1-12, obtain the scoring value EMS of the overall deviation degree of the terminal sequence characteristics of cfDNA in the test sample relative to the control sample, and based on the scoring value EMS, determine the source of the test sample. Among them, the closer the scoring value EMS is to 1, it indicates that the overall deviation of the terminal sequence characteristics of cfDNA in the test sample from the terminal sequence characteristics of cfDNA in the control sample is smaller; the closer the scoring value EMS is to 0, it indicates that the overall deviation of the terminal sequence characteristics of cfDNA in the test sample from the terminal sequence characteristics of cfDNA in the control sample is larger.

14. A method for determining the source of a sample to be tested, characterized in that, Including: S1. Use steps (1)-(3) in the method described in any one of claims 1-12 to obtain the terminal sequence characteristics distributions of cfDNA in the test sample and the control sample. S2. Based on the terminal sequence characteristics distributions of cfDNA in the test sample and the control sample, perform a difference test, and select the significantly different terminal sequences as features. S3. Based on the features selected in step S2, construct a prediction model to determine the probability that the test sample is a cancer sample. The prediction model is obtained by constructing a machine learning model based on the terminal sequence characteristics distributions of cfDNA in known cancer samples and known healthy samples.

15. A method for determining the source of a sample to be tested, characterized in that, Including: 1) Use steps (1)-(3) in the method described in any one of claims 1-12 to obtain the terminal sequence characteristics distributions of cfDNA in the test sample and the control sample. 2) Based on the terminal sequence characteristics distributions of cfDNA in the test sample and the control sample, perform a difference test, and select the significantly different terminal sequences as features. 3) Based on the features selected in step 2) and the scoring value EMS obtained by the method described in any one of claims 1-12, construct a prediction model to determine the probability that the test sample is a cancer sample. The prediction model is obtained by constructing a machine learning model based on the terminal sequence characteristics distributions of cfDNA in known cancer samples and known healthy samples.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method for determining the scoring value of the overall deviation degree of the terminal sequence characteristics of cfDNA in the test sample relative to the control sample described in any one of claims 1-12 or the steps in the method for determining the source of the test sample described in any one of claims 13-15.

17. An electronic device, characterized in that, Including: The computer-readable storage medium described in claim 16; And One or more processors for executing the program in the computer-readable storage medium.

Images