Apparatus for identifying true mutation or sequencing noise in ctdna sequencing data, computer-readable storage medium, and use

Abstract

Provided are an apparatus for identifying true mutation or sequencing noise in ctDNA sequencing data, a computer-readable storage medium, and a use. The noise level of a site is constructed using a data set from healthy individuals, and a true mutation signal is determined via the statistical test difference between the estimated noise level and the mutation frequency of a real observed tumor sample. Compared with the prior art, in which a single positive-determining value is used for all sites, using a positive-determining value for each site allows for more sensitive differentiation between low-frequency noise and mutation signals; further, tumor patients do not need to be paired with data from normal samples, thus reducing testing and sequencing costs; the present invention can be used in the preparation of products for the screening or auxiliary screening of cancer patients, or products for individualized cancer treatment.

Description

Device for identifying true mutations or sequencing noise in ctDNA sequencing data, computer-readable storage medium, and application thereof Technical Field

[0001] The present invention relates to a device, a computer-readable storage medium, and applications for identifying true mutations or sequencing noise in ctDNA sequencing data in the field of bioinformatics. Background Art

[0002] Liquid biopsy technology, a branch of in vitro diagnostics, analyzes and diagnoses diseases like cancer by sampling cerebrospinal fluid, saliva, blood, and urine, to a certain extent, circumventing the impact of tissue heterogeneity on disease. Currently, the detection of circulating tumor DNA (ctDNA) in the blood is a major research focus for liquid biopsy. Circulating tumor DNA (ctDNA) refers to DNA fragments released by tumor cells that carry information about tumor mutations. With the advancement of high-throughput technologies, the purpose of ctDNA NGS data analysis is to analyze specific mutations within tumor cells and, in turn, provide personalized treatment plans for cancer patients. As sequencing throughput increases and data volumes grow, this also presents significant challenges for mutation detection technology. The difficulty in detecting ctDNA data lies in the fact that tumor mutations are often relatively low in frequency. With such high data throughput, distinguishing low-frequency mutations from sequencing noise is particularly important. Existing detection technologies assume that the noise level is uniform across sites and use a uniform threshold for distinguishing between mutations and noise. This can reduce sensitivity and specificity for certain sites, and even lead to false-negative or false-positive sites. For mutation detection using existing technologies, normal control samples paired with tumor patients are also required, which directly increases costs. SUMMARY OF THE INVENTION

[0003] The present invention provides an apparatus and method for detecting the noise level in ctDNA sequencing data, which mainly uses site-based noise levels to detect mutations. The present invention uses a healthy individual data set to construct the background noise level of a site, and can determine the true mutation signal by estimating the noise level and statistically testing the difference in mutation frequency of the actual observed tumor sample. Compared with the prior art that uses a unified positive judgment value for all sites, the present invention uses the positive judgment value of the site, which is more sensitive to distinguishing low-frequency noise / mutation signals. The technology of the present invention detects SNVs and InDels in ctDNA data of single tumor samples, without the need for data from paired normal samples of tumor patients, which can reduce the cost of testing and sequencing. Technical issues

[0004] The technical problem to be solved by the present invention is how to accurately identify or detect whether the mutation detection results in ctDNA sequencing data are true mutations or sequencing noise. Technical Solutions

[0005] In order to solve the above technical problems, the present invention first provides a device for detecting the noise level in ctDNA sequencing data, which may include the following modules:

[0006] A1) ctDNA sequencing data processing module: used to obtain or receive ctDNA sequencing data from healthy subjects, and compare it with the human reference genome after quality control to obtain a comparison file;

[0007] A2) Mutation frequency detection module: used to obtain the sequencing depth of each site in the ctDNA sequencing data, the number of point mutation reads supporting each site, and the number of insertion or deletion mutation reads supporting each site based on the alignment file, calculate the point mutation frequency of each site based on the sequencing depth of each site and the number of point mutation reads supporting each site, and calculate the insertion or deletion mutation frequency of each site based on the sequencing depth of each site and the number of insertion or deletion mutation reads supporting each site;

[0008] A3) a point mutation noise level model construction module: configured to obtain the point mutation noise level of each site based on the point mutation characteristic grouping information of each site and the point mutation frequency of each site; the point mutation characteristic grouping information of each site includes the trinucleotide sequence of each site and the number of alignments of each site;

[0009] A4) an insertion or deletion mutation noise level model construction module: configured to obtain the insertion or deletion mutation noise level of each site based on the insertion or deletion mutation feature grouping information of each site and the insertion or deletion mutation frequency of each site; the insertion or deletion mutation feature grouping information of each site includes the repetitive sequence element information and the chromosome segment duplication information of each site;

[0010] A5) Noise level output module: used to output the noise level of the ctDNA sequencing data; the noise level includes the point mutation (SNV) noise level of each site and / or the insertion or deletion mutation (InDel) noise level of each site.

[0011] The site mentioned above may be a nucleotide, specifically a site included in the sequencing product panel. The trinucleotide sequence of each site includes three nucleotide sequences consisting of the nucleotide of each site on the reference genome and one nucleotide upstream and downstream thereof. The trinucleotide sequence of each site can be obtained from the human reference genome file. The number of alignments for each site can be a unique alignment or multiple alignments. The number of alignments for each site can be obtained from the Mappability file of the human reference genome. The repetitive sequence element information of each site indicates whether each site belongs to a repetitive sequence region and to which specific repetitive region type it belongs. The repetitive sequence element information of each site can be obtained from the repetitive sequence element data file of the human genome. The chromosome segment duplication information of each site indicates whether each site belongs to a large segment duplication region. The chromosome segment duplication information of each site can be obtained from the chromosome segment duplication data file of the human genome.

[0012] In the above device, the point mutation noise level model construction module described in A3) may include the following modules:

[0013] A3-1) Point mutation feature integration and grouping module: configured to group sites with the same trinucleotide sequence and the same number of alignments based on the trinucleotide sequence of each site and the number of alignments of each site to obtain groups with the same point mutation features;

[0014] A3-2) Point mutation prior background noise level acquisition module: used to calculate the average point mutation frequency of the groups with the same point mutation characteristics based on the point mutation frequency of each site in the groups with the same point mutation characteristics and A3-1) the groups with the same point mutation characteristics using formula 1:

[0015] ;

[0016] In formula 1: tm represents the average point mutation frequency of all sites in the group with the same point mutation characteristics, μ ptm represents the point mutation frequency of each site within the group with the same point mutation characteristics as described in A3-1), and n represents the number of sites within the group with the same point mutation characteristics as described in A3-1);

[0017] A3-3) Point mutation noise level acquisition module: used to obtain the point mutation noise level of each site using Formula 2 and Formula 3 based on the average point mutation frequency of the group with the same point mutation characteristics of each site and the point mutation frequency of each site:

[0018] ;

[0019] ;

[0020] In formula 2 and formula 3, p represents the information of a certain site within the range of the sequencing product panel, λ tm represents the average point mutation frequency of all sites with the same point mutation characteristics grouped as site p; μ p represents the point mutation frequency of site p; ω1 represents the weight; ε p represents the point mutation noise level at site p.

[0021] In the above device, the Mappability file can be obtained from the UCSC website (URL: https: / / genome.ucsc.edu / cgi-bin / hgTables / hgsid=1510554375_HANSVqcS3xxfWO2Ui7wYz8FFIlzA, reference genome version: Hg19; 100 bp read length comparison version).

[0022] In the above device, the insertion or deletion mutation noise level model construction module described in A4) may include the following modules:

[0023] A4-1) Insertion or deletion mutation feature integration and grouping module: configured to group sites having the same repetitive sequence element information and the same chromosome segment duplication information based on the repetitive sequence element information of each site and the chromosome segment duplication information of each site to obtain groups with the same insertion or deletion mutation feature;

[0024] A4-2) Insertion or deletion mutation prior background noise level acquisition module: used to calculate the average insertion or deletion mutation frequency of the insertion or deletion mutation group with the same insertion or deletion mutation characteristics based on the insertion or deletion mutation frequency of each site and the insertion or deletion mutation characteristics of the group with the same insertion or deletion mutation characteristics in A4-1 using Formula 4:

[0025] ;

[0026] In formula 4: rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics, μ prsl represents the insertion or deletion mutation frequency of each site within the group with the same insertion or deletion mutation characteristics as described in A4-1), and m represents the number of sites within the group with the same insertion or deletion mutation characteristics as described in A4-1);

[0027] A4-3) Insertion or deletion mutation noise level acquisition module: used to obtain the insertion or deletion mutation noise level of each site using Formula 5 and Formula 6 based on the average insertion or deletion mutation frequency of the same insertion or deletion mutation characteristic group of each site and the insertion or deletion mutation frequency of each site:

[0028] ;

[0029] In Equations 5 and 6, p represents the information of a certain site within the panel; ω2 represents the weight, λ rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics of site p, μ pl represents the insertion or deletion mutation frequency of site p, ε pl represents the insertion or deletion mutation noise level at site p.

[0030] In the above-mentioned device, the repetitive sequence element information includes eight types of repetitive elements: DNA repeats, long insert repeats, short insert repeats, low complexity region repeats, simple repeat region repeats, long repeat region repeats, non-repetitive regions, and other repeats. The chromosome segment repeat information includes two types: those located in non-large repeat regions and those located in large repeat regions. The human genome repetitive sequence element data file (Repeats file) can be obtained from the UCSC website (URL: https: / / genome.ucsc.edu / cgi-bin / hgTables, Reference Genome Version: Hg19, Select RepeatMasker). The human genome chromosome segment repeat data file (Segmental Duplication file) can be obtained from the UCSC website (URL: https: / / genome.ucsc.edu / cgi-bin / hgTables, Reference Genome Version: Hg 19, Select Segmental Dups).

[0031] The sequencing data mentioned above may be the sequencing data of the panel corresponding to the sequencing product.

[0032] In order to solve the above technical problems, the present invention also provides a device for identifying whether a mutation detection result in ctDNA sequencing data is a true mutation or sequencing noise. The device may include the following modules:

[0033] C1) ctDNA sequencing data quality control and alignment module: used to obtain or receive ctDNA sequencing data of the sample to be tested, and compare it with the human reference genome after quality control to obtain the alignment file;

[0034] C2) ctDNA sequencing data mutation analysis module: used to obtain the sequencing depth, mutation read support number and mutation frequency of the mutation site in the ctDNA sequencing data of the test sample based on the comparison file, and obtain the mutation detection result in the ctDNA sequencing data of the test sample;

[0035] C3) Statistical testing module: used to determine whether the mutation detection result is a true mutation or sequencing noise using a statistical hypothesis testing method based on the mutation detection result and the noise level. The noise level can be obtained using the device described above.

[0036] In the above device, the mutation may be SNV and / or InDel.

[0037] In the above device, the noise level may be the SNV noise level and / or the InDel noise level.

[0038] In the above device, the statistical hypothesis test can use Formula 7 and Formula 8 to calculate the statistical significance level of the hypothesis test:

[0039] ;

[0040] In Equations 7 and 8, p represents the information of a certain site within the range of the sequencing product panel, δ p is the noise level of site p (point mutation noise level, insertion or deletion mutation noise level); d p is the sequencing depth of site p; p k is the expected number of noise mutation reads supported at site p; p is the number of reads supporting the mutation (point mutation, insertion or deletion mutation) of site p in the sample to be tested; j represents the value of the mutation read support, j∈(0,k p ); P(k p ,λ p ) represents k p Greater than or equal to λ p The statistical P value size.

[0041] In order to solve the above technical problems, the present invention also provides a method for detecting the noise level in ctDNA sequencing data, which may include the following steps:

[0042] B1) ctDNA sequencing data processing: ctDNA sequencing data from healthy subjects are obtained and compared with the human reference genome after quality control to obtain an alignment file;

[0043] B2) Mutation frequency detection: Based on the alignment file, the sequencing depth of each site, the number of point mutation reads supporting each site, and the number of insertion or deletion mutation reads supporting each site in the ctDNA sequencing data are obtained; the point mutation frequency of each site is calculated based on the sequencing depth of each site and the number of point mutation reads supporting each site; and the insertion or deletion mutation frequency of each site is calculated based on the sequencing depth of each site and the number of insertion or deletion mutation reads supporting each site;

[0044] B3) Point mutation noise level model construction: The point mutation noise level of each site is obtained based on the point mutation characteristic grouping information of each site and the point mutation frequency of each site; the point mutation characteristic grouping information of each site includes the trinucleotide sequence of each site and the number of alignments of each site;

[0045] B4) Construction of an insertion or deletion mutation noise level model: Obtaining the insertion or deletion mutation noise level for each site based on the insertion or deletion mutation characteristic grouping information for each site and the insertion or deletion mutation frequency for each site; the insertion or deletion mutation characteristic grouping information for each site includes repetitive sequence element information and chromosome segment duplication information for each site;

[0046] B5) Noise level output: Output the noise level of the ctDNA sequencing data; the noise level includes the point mutation (SNV) noise level of each site and / or the insertion or deletion mutation (Indel) noise level of each site.

[0047] The site mentioned above may be a nucleotide, specifically a site included in the sequencing product panel. The trinucleotide sequence of each site includes three nucleotide sequences consisting of the nucleotide of each site on the reference genome and one nucleotide upstream and downstream thereof. The trinucleotide sequence of each site can be obtained from the human reference genome file. The number of alignments for each site can be a unique alignment or multiple alignments. The number of alignments for each site can be obtained from the Mappability file of the human reference genome. The repetitive sequence element information of each site indicates whether each site belongs to a repetitive sequence region and to which specific repetitive region type it belongs. The repetitive sequence element information of each site can be obtained from the repetitive sequence element data file of the human genome. The chromosome segment duplication information of each site indicates whether each site belongs to a large segment duplication region. The chromosome segment duplication information of each site can be obtained from the chromosome segment duplication data file of the human genome.

[0048] In the above method, the construction of the point mutation noise level model in B3) may include the following steps:

[0049] B3-1) Point mutation feature integration grouping: Based on the trinucleotide sequence of each site and the number of alignments of each site, sites with the same trinucleotide sequence and the same number of alignments are grouped together to obtain groups with the same point mutation features;

[0050] B3-2) Obtaining the prior background noise level of point mutations: Based on the point mutation frequency of each site in the group with the same point mutation characteristics and B3-1) the average point mutation frequency of the group with the same point mutation characteristics is calculated using Formula 1:

[0051] ;

[0052] In formula 1: tm represents the average point mutation frequency of all sites in the group with the same point mutation characteristics, μ ptm represents the point mutation frequency of each site within the group with the same point mutation characteristics as described in B3-1), and n represents the number of sites within the group with the same point mutation characteristics as described in B3-1);

[0053] B3-3) Obtaining the point mutation noise level: Based on the average point mutation frequency of the group with the same point mutation characteristics of each site and the point mutation frequency of each site, the point mutation noise level of each site is obtained using Formula 2 and Formula 3:

[0054] ;

[0055] In Equations 2 and 3: p represents the information of a certain site within the panel; λ tm represents the average point mutation frequency of all sites with the same point mutation characteristics grouped as site p; μ p represents the point mutation frequency of site p; ω1 represents the weight; ε p represents the point mutation noise level at site p.

[0056] In the above method, the Mappability file can be obtained from the UCSC website (URL: https: / / genome.ucsc.edu / cgi-bin / hgTables / hgsid=1510554375_HANSVqcS3xxfWO2Ui7wYz8FFIlzA, reference genome version: Hg19; 100 bp read length comparison version).

[0057] In the above method, the construction of the insertion or deletion mutation noise level model in B4) may include the following steps:

[0058] B4-1) Integration and grouping of insertion or deletion mutation characteristics: Based on the repetitive sequence element information and the chromosome segment duplication information of each site, sites with the same repetitive sequence element information and the same chromosome segment duplication information are grouped together to obtain groups with the same insertion or deletion mutation characteristics;

[0059] B4-2) Obtaining the insertion or deletion a priori background noise level: Based on the insertion or deletion mutation frequency of each site and B4-1) the insertion or deletion mutation characteristics of the same group, the average insertion or deletion mutation frequency of the group with the same insertion or deletion mutation characteristics is calculated using Formula 4:

[0060] ;

[0061] In formula 4: rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics, μ prsl represents the insertion or deletion mutation frequency of each site within the group with the same insertion or deletion mutation characteristics as described in B4-1), and m represents the number of sites within the group with the same insertion or deletion mutation characteristics as described in B4-1);

[0062] B4-3) Obtaining the insertion or deletion mutation noise level: Based on the average insertion or deletion mutation frequency of the group with the same insertion or deletion mutation characteristics for each site and the insertion or deletion mutation frequency of each site, the insertion or deletion mutation noise level of each site is obtained using Formula 5 and Formula 6:

[0063] ;

[0064] ;

[0065] In Equations 5 and 6, p represents the information of a certain site within the panel; ω2 represents the weight, λ rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics of site p, μ pl represents the insertion or deletion mutation frequency of site p, ε pl represents the insertion or deletion mutation noise level at site p.

[0066] In the above method, the repetitive sequence element information includes eight types of repetitive elements: DNA repeats, long insert repeats, short insert repeats, low complexity region repeats, simple repeat region repeats, long repeat region repeats, non-repetitive regions, and other repeats. The chromosome segment duplication information includes two types: those located in non-large repeat regions and those located in large repeat regions. The human genome repetitive sequence element data file (Repeats file) can be obtained from the UCSC website (URL: https: / / genome.ucsc.edu / cgi-bin / hgTables, Reference Genome Version: Hg19, Select RepeatMasker). The human genome chromosome segment duplication data file (Segmental Duplication file) can be obtained from the UCSC website (URL: https: / / genome.ucsc.edu / cgi-bin / hgTables, Reference Genome Version: Hg 19, Select Segmental Dups).

[0067] The sequencing data mentioned above may be the sequencing data of the panel corresponding to the sequencing product.

[0068] In order to solve the above technical problems, the present invention also provides a method for identifying whether a mutation detection result in ctDNA sequencing data is a true mutation or sequencing noise, and the method may include the following steps:

[0069] D1) ctDNA sequencing data quality control and alignment: Obtain ctDNA sequencing data from the sample to be tested, and then align it with the human reference genome after quality control to generate an alignment file;

[0070] D2) Mutation analysis of ctDNA sequencing data: obtaining the sequencing depth, mutation read support number, and mutation frequency of the mutation sites in the ctDNA sequencing data of the sample to be tested based on the alignment file, and obtaining mutation detection results in the ctDNA sequencing data of the sample to be tested;

[0071] D3) Statistical test analysis: used to determine whether the mutation detection result is a true mutation or sequencing noise based on the mutation detection result and the noise level using a statistical hypothesis test method; the noise level can be obtained using the method described above.

[0072] In the above method, the mutation may be SNV and / or InDel.

[0073] In the above method, the noise level may be the SNV noise level and / or the InDel noise level.

[0074] In the above method, the statistical hypothesis test can use Formula 7 and Formula 8 to calculate the statistical significance level of the hypothesis test:

[0075] ;

[0076] In Equations 7 and 8, p represents the information of a certain site within the range of the sequencing product panel, δ p is the noise level of site p (point mutation noise level, insertion or deletion mutation noise level); d p is the sequencing depth of site p; p k is the expected number of noise mutation reads supported at site p; p is the number of reads supporting the mutation (point mutation, insertion or deletion mutation) of site p in the sample to be tested; j represents the value of the mutation read support, j∈(0,k p ); P(k p ,λ p ) represents k p Greater than or equal to λ p The statistical P value size.

[0077] In order to solve the above technical problems, the present invention also provides a computer-readable storage medium storing a computer program, which can be a computer-readable storage medium for detecting noise levels in ctDNA sequencing data or a computer-readable storage medium for identifying true mutations or sequencing noise in ctDNA sequencing data. The computer program for the computer-readable storage medium for detecting noise levels in ctDNA sequencing data or the computer-readable storage medium for identifying true mutations or sequencing noise in ctDNA sequencing data can cause a computer to execute the steps of the method described above or run the modules of the device described above.

[0078] In order to solve the above technical problems, the present invention also provides a computer device, including a memory, a processor and a computer program stored in the memory. The processor executes the computer program to implement steps B1)-B5) or D1)-D3) of the above method.

[0079] In order to solve the above technical problem, the present invention further provides a computer program product, including a computer program, which can implement the steps of the above method when executed by a processor.

[0080] Any of the following applications of the device described above also fall within the scope of protection of the present invention:

[0081] E1) Use in the preparation of products for screening or assisting in screening cancer patients;

[0082] E2) Application in the development or preparation of products for personalized cancer treatment.

[0083] Any of the following applications of the method described above also fall within the scope of protection of the present invention:

[0084] E1) Use in the preparation of products for screening or assisting in screening cancer patients;

[0085] E2) Application in the development or preparation of products for personalized cancer treatment.

[0086] Any of the following applications of the computer-readable storage medium described above also falls within the scope of protection of the present invention:

[0087] E1) Use in the preparation of products for screening or assisting in screening cancer patients;

[0088] E2) Application in the development or preparation of products for personalized cancer treatment.

[0089] Any of the following applications of the computer device described above also falls within the scope of protection of the present invention:

[0090] E1) Use in the preparation of products for screening or assisting in screening cancer patients;

[0091] E2) Application in the development or preparation of products for personalized cancer treatment.

[0092] Any of the following applications of the computer program product described above also falls within the scope of protection of the present invention:

[0093] E1) Use in the preparation of products for screening or assisting in screening cancer patients;

[0094] E2) Application in the development or preparation of products for personalized cancer treatment.

[0095] The computer program product mentioned above may be a software product that implements its solution mainly through a computer program.

[0096] The computer-readable storage medium mentioned above refers to a carrier for storing data, which may be a magnetic tape, magnetic disk, floppy disk, optical disk, magneto-optical disk, ROM, PROM, VCD, DVD, hard disk, flash memory, USB flash disk, CF card, SD card, MMC card, SM card, memory stick (Memory Stick) or xD card, etc.

[0097] In view of the above problems, the present invention aims to provide a method for detecting mutations based on the noise level of sites.

[0098] The technology of the present invention detects SNVs and InDels in ctDNA data from a single tumor sample without the need for data from paired normal samples from tumor patients, which reduces the cost of testing and sequencing. Beneficial effects

[0099] 1. The present invention detects point mutations (SNVs) and / or insertion / deletion mutations (InDels) in ctDNA sample sequencing data without the need to provide normal control samples, thus reducing cost investment.

[0100] 2. This invention utilizes a dataset of healthy individuals to construct the background noise level at each site. This estimation of the noise level can be used to statistically compare the difference between the estimated noise level and the observed mutation frequency in tumor samples to determine the true mutation signal. Compared to the prior art, which uses a uniform positive judgment value for all sites, this invention uses a site-specific positive judgment value, which is more sensitive in distinguishing low-frequency noise from mutation signals.

[0101] 3. The present invention simulates specific noise levels for different sites in sequencing data, which has obvious advantages in evaluating the minimum detection limit (LoD) of the site.

[0102] 4. This method is simple to operate, consumes minimal resources, and is fast. Because it uses a healthy individual dataset to construct a noise level model, when analyzing real tumor samples, if the panel sequencing product and experimental conditions remain unchanged, there is no need to rebuild the model; statistical tests can be performed directly based on the noise level and the actual observed mutation frequency. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] Figure 1 is a schematic diagram of the process of constructing the noise level model for ctDNA sequencing data.

[0104] Figure 2 shows a comparison of the detection results of mutation sites using different detection methods, where AF represents the mutation frequency. Modes for Carrying Out the Invention

[0105] The present invention will be further described in detail below in conjunction with specific embodiments. The examples provided are only for illustrating the present invention and are not intended to limit the scope of the present invention. The examples provided below can serve as a guide for further improvements by those skilled in the art and are not intended to limit the present invention in any way.

[0106] Unless otherwise specified, the experimental methods in the following examples are conventional methods and were performed according to the techniques or conditions described in the literature in the field or according to the product instructions. The materials and reagents used in the following examples, unless otherwise specified, were all commercially available.

[0107] Example 1: Constructing a site noise level model based on ctDNA sequencing data of healthy individuals.

[0108] This study is based on 40 healthy individuals (with normal annual blood test results) collected by Beijing Genetron Health Technology Co., Ltd. Using the Onco Sonar product (CT_DNA_170 Solid Tumor Gene Mutation Detection Kit (Reversible Termination Sequencing), Beijing Genetron Health Technology Co., Ltd., product number RSN113), according to standard experimental procedures, ctDNA raw sequencing data from 40 healthy individuals without known cancer was obtained through DNA extraction, library construction, probe capture, and sequencing. This example includes steps such as preprocessing the raw data, obtaining mutation frequencies, determining mutation grouping characteristics, and constructing a noise level model.

[0109] 1. Raw data preprocessing.

[0110] Raw sequencing data in FASTQ format were analyzed using standard data analysis procedures, including quality control (including data volume, pre- and post-deduplication depth, Q30, and capture efficiency) and alignment. The quality control and alignment software used were Trimmomatic (v0.33, http: / / www.usadellab.org / cms / index.php / page=trimmomatic) and Biomarkers for Biotechnology Information Systems (BWA) (v0.7.10-r789, https: / / bio-bwa.sourceforge.net / ), respectively.

[0111] The quality control software Trimmomatic requires parameter settings, including the adapter sequences used for sequencing, a sliding window clipping window size of 4 bp, and an average base quality threshold of at least 15 within the sliding window. Bases with a quality score below 3 were removed from the initial ends of sequencing reads in the raw data, and bases with a quality score below 3 were removed from the end of sequencing reads. The software defaults were used unless otherwise specified. ctDNA samples were qualified for quality control with a Q30 ≥ 80%, a sequencing depth ≥ 15,000x before deduplication (referred to as "deduplication"), and a deduplication ratio ≥ 2.5. Sequencing data that passed quality control were aligned to the human reference genome (Hg19) using BWA software. BAM files (containing valid alignment data) were generated for healthy individual samples for subsequent analysis.

[0112] 2. Obtain the mutation frequency of the detection sites within the sequencing product panel.

[0113] Based on the BAM file, the sequencing depth and read counts for each possible A / C / T / G point mutation were calculated for each locus within the panel (Onco Sonar product, panel size approximately 0.3 MB). The point mutation frequencies for each of the four nucleotides were calculated (the frequency of no detected point mutations was recorded as 0). Specifically, the BAM file was generated using the mpileup command in Samtools (v1.7, http: / / www.htslib.org / doc / samtools.html) to generate a pileup format file. The sequencing depth (i.e., total number of sequencing reads per locus) and the number of mutation reads were calculated using the ctDNA-SNV-InDel-pipeline program (Beijing PanGenomics Gene Technology Co., Ltd.). The frequency of the point mutation at each locus was calculated by dividing the number of reads supporting the point mutation by the sequencing depth at that locus. Similarly, the sequencing depth (total number of sequencing reads for the site) and the number of reads supporting insertion or deletion mutations for each site within the panel are counted based on the BAM file. The mutation frequency of the insertion or deletion mutation at each site is calculated by dividing the number of reads supporting insertion or deletion mutations at each site by the sequencing depth of the site.

[0114] 3. Screen feature groups and construct noise level models for point mutations at the sites.

[0115] 3.1 Screen the point mutation feature groups and determine the prior background noise level of point mutations in each group.

[0116] The Mappability file for the human reference genome (Hg19 version, 100-bp read length alignment) was downloaded from the database developed and maintained by the University of California, Santa Cruz (UCSC database) (https: / / genome.ucsc.edu / cgi-bin / hgTables / hgsid=1510554375_HANSVqcS3xxfWO2Ui7wYz8FFIlzA). This file indicates whether a particular genomic sequence position has unique or multiple alignments across the entire genome. Based on the Hg19 reference genome file, the tri-nucleotide context of each position (i.e., the nucleotide sequence of a particular position in the reference genome, along with a sequence consisting of one nucleotide upstream and one nucleotide downstream) was extracted. The tri-nucleotide context and the number of alignments (i.e., unique or multiple) were used as features for subsequent model construction.

[0117] The trinucleotide sequence and alignment number of each site are combined to perform feature grouping. Specifically, sites with the same trinucleotide sequence and the same alignment number are grouped together. The sites included in the group are sites with the same point mutation characteristics. Based on different trinucleotide sequences and alignment numbers, up to 128 groups can be divided (each site has 4 possible forms of nucleotides: A / T / C / G, and there are up to 4 possible arrangements of three nucleotides). 3 = 64; the number of alignments is divided into two cases: unique alignment and multiple alignments; the trinucleotide sequence features and the number of alignments can be divided into a maximum of 64 × 2 = 128 groups, each of which is called a group with identical point mutation features. Based on the above, the sites within the panel sequencing range are grouped (the trinucleotide sequence and alignment number corresponding to each site are unique, so each site will belong to a unique group). Based on the point mutation frequency of the site obtained in step 2, the point mutation frequency of all sites within the group with identical point mutation features (sites with identical features) is averaged to calculate the average point mutation frequency of the group (Equation 1). This average point mutation frequency serves as the prior background noise level for the group:

[0118] .

[0119] In formula 1: tm represents the average point mutation frequency of all sites in the group with the same point mutation characteristics, μ ptm represents the point mutation frequency of each site in the group with the same point mutation characteristics, and n represents the number of sites in the group with the same point mutation characteristics.

[0120] 3.2 Construct a point mutation noise level model for the site.

[0121] A model is constructed by combining the groups with the same point mutation characteristics at each site and their corresponding point mutation prior background noise level (the average point mutation frequency obtained in step 3.1) and the point mutation frequency of each site detected in step 2 to calculate the noise level of the point mutation at the current site:

[0122] .

[0123] In formula 2 and formula 3, p represents the information of a certain site within the range of the sequencing product panel, such as chr7: 55249071; λ tm represents the average point mutation frequency of all sites with the same point mutation characteristics grouped as site p; μ p represents the point mutation frequency of site p; ω1 represents the weight; ε p represents the point mutation noise level at site p.

[0124] 4. Screen the insertion or deletion mutation feature groups and construct the noise level model of the insertion or deletion mutation at the site.

[0125] 4.1 Screen the grouping characteristics of insertion or deletion mutations and determine the a priori background noise level of insertion or deletion mutations in each group.

[0126] The human genome repetitive sequence element data file (Repeats file, URL: https: / / genome.ucsc.edu / cgi-bin / hgTables, genome version Hg19, select RepeatMasker) and human genome chromosome segment duplication data file (Segmental Duplication file, URL: https: / / genome.ucsc.edu / cgi-bin / hgTables, genome version Hg 19, select Segmental Dups) were downloaded from the UCSC database. The Repeats file indicates whether a genomic locus belongs to a repetitive sequence region and the specific type of repetitive region. It is categorized into eight different types: DNA, LINE, SINE, Low_complexity, Simple_repeats, LTR, Non_Repeats, and Other. (DNA type represents DNA repeats; LINE type represents long insert repeats; SINE type represents short insert repeats; Low complexity type represents repeats in low-complexity regions, such as G-rich and AT-rich regions; Simple repeats type represents repeats in simple repeat regions, such as (TTAGGG)n and (GATATA)n; LTR type represents long repeats; Non_Repeats type represents repeats in non-repeats; and Other type represents other repeats. For detailed information on specific types, please refer to the UCSC database.) The Segmental Duplication file indicates whether a genomic locus belongs to a large repeat region. (Chromosomal repeat regions can be categorized as those located in non-large repeat regions and those located in large repeat regions.) The repetitive sequence element information (Repeats information) of the site and its chromosome segment duplication information (Segmental Duplication information) are used as features for the subsequent construction of the insertion or deletion mutation noise level model.

[0127] Sites with the same Repeats information (i.e., belonging to the same of the eight aforementioned types, such as DNA repeat elements) and Segmental Duplication information are grouped together, referred to as groups with identical insertion or deletion mutation characteristics. The sites within this group are considered to have identical insertion or deletion mutation characteristics. Repeats information can be divided into eight groups; Segmental Duplication information can be divided into two groups. The combination of Repeats and Segmental Duplication information can be divided into a maximum of 8 × 2 = 16 groups, each of which is referred to as a group with identical insertion or deletion mutation characteristics. Sites within the panel are grouped (each site's Repeats information and Segmental Duplication information are unique, so each site belongs to a unique group). Based on the insertion or deletion mutation frequency of the site obtained in Step 2, the average insertion or deletion mutation frequency of all sites within the group with identical insertion or deletion mutation characteristics is calculated (Equation 4). This information serves as the a priori background noise level for insertion or deletion mutations in that group:

[0128] .

[0129] In formula 4: rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics, μ prsl represents the insertion or deletion mutation frequency of each site in the group with the same insertion or deletion mutation characteristics, and m represents the number of sites in the group with the same insertion or deletion mutation characteristics.

[0130] 4.2 Construct a noise level model for insertion or deletion mutations at the site.

[0131] Combining the a priori background noise level of each site group and the corresponding insertion or deletion mutation (the average mutation frequency of the groups with the same characteristics obtained in step 4.1) and the insertion or deletion mutation frequency of each site obtained in step 2, use Equations 5 and 6 to construct a model to calculate the noise level of the insertion or deletion mutation at the current site.

[0132]

[0133] .

[0134] In Equations 5 and 6, p represents the information of a site within the range of the sequencing product panel, ω2 represents the weight, and λ rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics of site p, μ pl represents the insertion or deletion mutation frequency of site p; ε pl Indicates the insertion or deletion mutation noise level at locus p.

[0135] The point mutation noise level model of the locus and the noise level model of the insertion or deletion mutation of the locus together constitute the ctDNA NGS sequencing data noise level model.

[0136] Example 2. Process for identifying whether the mutation detection result in ctDNA sequencing data is a true mutation or sequencing noise.

[0137] For the test sample, based on the same standard test process as in Example 1, ctDNA sequencing data is obtained through processes such as DNA extraction, library construction, probe capture, and on-machine sequencing; after data preprocessing, alignment and other steps, for samples that pass quality control, the sequencing depth, the number of mutant reads supported, and the mutation frequency of the locus are counted to obtain the mutation detection result of the ctDNA sequencing data of the test sample. According to the ctDNA sequencing data noise level model constructed in Example 1, for the locus where mutation occurs in the test sample, a statistical test method is used to identify whether the mutation (point mutation, insertion or deletion mutation) detection result in the sample ctDNA sequencing data is a true mutation or sequencing noise.

[0138] The steps of quality control and alignment of the sample ctDNA NGS sequencing data to the reference genome (reference genome version: Hg19) using the method in Example 1 are performed to obtain a BAM file; based on the BAM file, the sequencing depth of the mutant locus and the number of mutant supports (the number of reads supporting the mutation) are counted, and the mutation frequency of the locus is calculated; based on the information of the locus (i.e., the chromosome number and genomic position of the locus where mutation occurs) and the type of mutation (whether the mutation is a point mutation or an insertion or deletion mutation type), the noise level of the locus obtained by matching the model constructed in Example 1 is obtained (point mutation type matches the point mutation noise level, insertion or deletion mutation matches the insertion or deletion mutation noise level). According to the idea of statistical hypothesis testing, H0: The mutation at this locus is sequencing noise; H1: The mutation at this locus is a true mutation; the statistical significance level (P-value) of the locus is calculated according to the sequencing depth, the number of mutant supports, and the mutation noise level of this locus (Equations 7 and 8); if P-value < alpha (positive threshold), then H0 should be rejected and H1 should be accepted, that is, the detected mutation is a true mutation; otherwise, H0 is accepted, that is, the detected mutation is sequencing noise.

[0139] 。<00​​​​​is the sequencing depth of site p, λ p is the number of noise mutation reads supporting site p; k p is the number of reads supporting the mutation (point mutation, insertion or deletion mutation) of site p in the sample to be tested; j represents the value of the mutation read support, j∈(0,k p ); P (k p ,λ p ) represents k p Greater than or equal to λ p The statistical P value size.

[0141] In this example, the mutation positive threshold is 0.0001.

[0142] Example 3: The present invention verifies the mutation detection results based on tumor sample ctDNA sequencing data.

[0143] The mutation detection method established in Examples 1 and 2 of the present invention was compared with existing technologies based on ctDNA sequencing data from tumor samples. The evaluation metric was the number of known positive variants (i.e., point mutations, insertions, or deletions) detected. Ten previously clinically tested tumor samples (including five lung cancer and five colorectal cancer samples) were collected from Beijing Genetron Health Gene Technology Co., Ltd. Following the Onco Sonar product testing process, NGS sequencing data was generated through DNA extraction, equal sample mixing, library construction, probe capture, and sequencing.

[0144] Sequencing data from 10 samples were analyzed separately (without mixing DNA samples) using existing techniques (i.e., using the samtools software mpileup command to calculate the total depth, mutation support, and mutation frequency of variant sites within ctDNA sequencing data. True mutations were determined as a mutation frequency ≥ 0.1% and a mutation support ≥ 4; see http: / / www.htslib.org / doc / 1.7 / samtools.html). The results (see Figure 2) showed a total of 648 positive mutation sites, including 486 sites with low-frequency mutations (mutation frequency < 0.5%). Among these low-frequency mutations, 192 sites had an allele frequency (AF) ≤ 0.1%, 96 sites had an AF < 0.2%, 96 sites had an AF < 0.2%, and 102 sites had an AF < 0.5%.

[0145] The mixed sequencing data were analyzed using both existing techniques and the method established in Example 2, and the mutation results were compared. Comparing the mutation detection ratios at known positive mutation sites revealed that the mutation detection method of the present invention is superior in detecting low-frequency mutations (less than 0.5%), improving performance by approximately 26%.

[0146] Calculation method: As shown in Figure 2, for a certain frequency group, the difference between the mutation detection ratio of the method of the present invention and the mutation detection ratio of the existing method is divided by the mutation detection ratio of the existing method to obtain the performance improvement ratio of the current group; specific calculation steps: AF≤0.1% group performance improvement ratio: (0.197-0.0989) / 0.0989=99%, 0.1<AF≤0.2% group performance improvement ratio: (0.5-0.3437) / 0.3437=45%, 0.2<AF≤0.3% group performance improvement ratio: (0.8333-0.7395) / 0.7395=13%, 0.3<AF≤0.5% group performance improvement ratio: (0.8725-0.8828) / 0.8823=-1%, 0.5<AF≤2.5% group performance improvement ratio: (1-1) / 1=0%, AF>2.5% Group performance improvement ratio: (1-1) / 1 = 0%. Based on the performance improvement ratio of each frequency group mentioned above, the average performance improvement of the six frequency groups is calculated to obtain the overall performance improvement ratio: (99% + 45% + 13% - 1% + 0% + 0%) / 6 = 26%.

[0147] In summary, the present invention calculates the statistical significance level P-value of the site based on the sequencing depth of the site in the ctDNA sequencing data, the number of mutation reads support and the noise level of the site. It no longer relies on the unified judgment criteria for positive mutations. It is particularly suitable for the detection of low-frequency mutations and will not lead to missed detection due to failure to reach the threshold. The present invention does not require the provision of normal control samples, reducing cost investment and shortening the analysis cycle.

[0148] The present invention further provides a computer device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the following steps:

[0149] B1) ctDNA sequencing data processing: ctDNA sequencing data from healthy subjects are obtained and compared with the human reference genome after quality control to obtain an alignment file;

[0150] B2) Mutation frequency detection: Based on the alignment file in B1), obtain the sequencing depth, point mutation reads support number, and insertion or deletion mutation reads support number of each site in the ctDNA sequencing data. Calculate the point mutation frequency of each site based on the sequencing depth and point mutation reads support number of each site. Calculate the insertion or deletion mutation frequency of each site based on the sequencing depth and insertion or deletion mutation reads support number of each site.

[0151] B3) Point mutation noise level model construction: The point mutation noise level of each site is obtained based on the point mutation characteristic grouping information and the point mutation frequency of each site. The point mutation characteristic grouping information of each site includes the trinucleotide sequence of each site and the number of alignments at each site.

[0152] B4) Construction of an insertion or deletion mutation noise level model: Obtaining the insertion or deletion mutation noise level for each site based on the insertion or deletion mutation characteristic grouping information and the insertion or deletion mutation frequency of each site; the insertion or deletion mutation characteristic grouping information for each site includes the repetitive sequence element information and the chromosome segment duplication information for each site;

[0153] B5) Noise level output: Outputs the noise level of ctDNA sequencing data; this noise level includes the SNV noise level of each site and / or the InDel noise level of each site.

[0154] B3) Construction of the point mutation noise level model includes the following steps:

[0155] B3-1) Point mutation feature integration and grouping: Based on the trinucleotide sequence of each site and the number of alignments for each site, sites with the same trinucleotide sequence and the same number of alignments were grouped together to obtain groups with the same point mutation features;

[0156] B3-2) Obtaining the prior background noise level of point mutations: Based on the point mutation frequency of each site in the group with the same point mutation characteristics and B3-1) the average point mutation frequency of the group with the same point mutation characteristics is calculated using formula 1:

[0157] ;

[0158] In formula 1: tm represents the average point mutation frequency of all sites in the group with the same point mutation characteristics, μ ptm represents the point mutation frequency of each site within the group with the same point mutation characteristics in B3-1, and n represents the number of sites within the group with the same point mutation characteristics in B3-1;

[0159] B3-3) Obtaining the point mutation noise level: Based on the average point mutation frequency of the group with the same point mutation characteristics for each site and the point mutation frequency of each site, the point mutation noise level of each site is obtained using Equations 2 and 3:

[0160] ;

[0161] In formula 2 and formula 3: p represents the information of a certain site within the panel range, λ tmrepresents the average point mutation frequency of all sites with the same point mutation characteristics grouped as site p; μ p represents the point mutation frequency of site p; ω1 represents the weight; ε p represents the point mutation noise level at site p.

[0162] B4) Construction of the insertion or deletion noise level model includes the following steps:

[0163] B4-1) Integration and grouping of insertion or deletion mutation characteristics: Based on the repetitive sequence element information and the chromosome segment duplication information of each site, sites with the same repetitive sequence element information and the same chromosome segment duplication information are grouped together to obtain the same insertion or deletion mutation characteristics group;

[0164] B4-2) Obtaining the a priori background noise level of insertion or deletion mutations: Based on the insertion or deletion mutation frequency of each site and B4-1) the average insertion or deletion mutation frequency of groups with the same insertion or deletion mutation characteristics, the following formula (4) is used to calculate:

[0165] ;

[0166] In formula 4: rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics, μ prsl represents the insertion or deletion mutation frequency of each site within the group with the same insertion or deletion mutation characteristics in B4-1, and m represents the number of sites within the group with the same insertion or deletion mutation characteristics in B4-1;

[0167] B4-3) Obtaining the insertion or deletion mutation noise level: Based on the average insertion or deletion mutation frequency within the same group of insertion or deletion mutation characteristics of each site and the insertion or deletion mutation frequency of each site, the insertion or deletion mutation noise level of each site is obtained using Equations 5 and 6:

[0168] ;

[0169] In Equations 5 and 6, p represents the information of a certain site within the panel; ω2 represents the weight, λ rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics of site p, μ pl represents the insertion or deletion mutation frequency of site p, ε pl represents the insertion or deletion mutation noise level at site p.

[0170] The present invention further provides a computer device comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the following steps:

[0171] D1) ctDNA sequencing data quality control and alignment: Obtain ctDNA sequencing data from the sample to be tested, and then align it with the human reference genome after quality control to generate an alignment file;

[0172] D2) Mutation analysis of ctDNA sequencing data: Based on the alignment file in D1), the sequencing depth, mutation read support number, and mutation frequency of the mutation sites in the ctDNA sequencing data of the test sample are obtained to obtain the mutation detection results in the ctDNA sequencing data of the test sample;

[0173] D3) Statistical analysis: Based on the mutation detection results and the noise level, a statistical hypothesis test is used to determine whether the mutation detection result is a true mutation or sequencing noise; the mutation is a SNV and / or InDel; the noise level is the SNV noise level and / or the InDel noise level, and the noise level is obtained using the method described above.

[0174] Statistical hypothesis testing uses Equations 7 and 8 to calculate the statistical significance level of the hypothesis test:

[0175] ;

[0176] In Equations 7 and 8, p represents the information of a certain site within the Panel, δ p is the noise level at site p; d p is the sequencing depth of site p, λ p k is the expected number of noise mutation reads supported at site p; p is the number of mutation reads supporting the site p in the sample to be tested; j represents the value of the mutation reads supporting the site p, j∈(0,k p ); P (k p ,λ p ) represents k p Greater than or equal to λ p The statistical P value size.

[0177] The present invention has been described in detail above. For those skilled in the art, without departing from the purpose and scope of the present invention, and without the need to carry out unnecessary experimental conditions, the present invention can be implemented in a wide range under equivalent parameters, concentrations and conditions. Although the present invention provides specific embodiments, it should be understood that further improvements can be made to the present invention. In short, according to the principles of the present invention, this application is intended to include any changes, uses or improvements to the present invention, including changes that depart from the disclosed scope in this application and are made using conventional techniques known in the art.

[0178] CROSS-REFERENCE TO RELATED APPLICATIONS

[0179] This application claims priority to the Chinese patent application (application number 202310929989.3) filed on July 27, 2023, the entire contents of which are incorporated herein by reference. Industrial Applicability

[0180] The technical solution of this invention detects SNVs and InDels in ctDNA data from a single tumor sample, eliminating the need for paired normal sample data from the tumor patient, thereby reducing experimental and sequencing costs. This invention can be applied to detect low-frequency mutations in tumor samples or to develop products for detecting low-frequency mutations in tumor samples, thereby providing personalized medication guidance for cancer patients.

Claims

1. A device for detecting noise level in ctDNA sequencing data, characterized in that: The device comprises the following modules: A1) ctDNA sequencing data processing module: used to obtain ctDNA sequencing data of healthy subjects, and compare it with the human reference genome after quality control to obtain a comparison file; A2) Mutation frequency detection module: used to obtain the sequencing depth of each site in the ctDNA sequencing data, the number of point mutation reads supporting each site, and the number of insertion or deletion mutation reads supporting each site based on the comparison file, calculate the point mutation frequency of each site based on the sequencing depth of each site and the number of point mutation reads supporting each site, and calculate the insertion or deletion mutation frequency of each site based on the sequencing depth of each site and the number of insertion or deletion mutation reads supporting each site; A3) a point mutation noise level model building module: used to obtain the point mutation noise level of each site based on the point mutation feature grouping information of each site and the point mutation frequency of each site; the point mutation feature grouping information of each site includes the trinucleotide sequence of each site and the number of alignments of each site; A4) an insertion or deletion mutation noise level model construction module: used to obtain the insertion or deletion mutation noise level of each site based on the insertion or deletion mutation feature grouping information of each site and the insertion or deletion mutation frequency of each site; the insertion or deletion mutation feature grouping information of each site includes the repetitive sequence element information of each site and the chromosome segment duplication information of each site; A5) Noise level output module: used to output the noise level of the ctDNA sequencing data; the noise level includes the SNV noise level of each site and / or the InDel noise level of each site.

2. The device according to claim 1, characterized in that: A3) The point mutation noise level model construction module includes the following modules: A3-1) Point mutation feature integration grouping module: used for grouping sites with the same trinucleotide sequence and the same number of alignments into a group based on the trinucleotide sequence of each site and the number of alignments of each site to obtain groups with the same point mutation feature; A3-2) Point mutation prior background noise level acquisition module: used to calculate the average point mutation frequency of the group with the same point mutation feature based on the point mutation frequency of each site in the group with the same point mutation feature and A3-1) the group with the same point mutation feature using formula 1: ； In formula 1: tm represents the average point mutation frequency of all sites in the same group with the same point mutation characteristics, μ ptm represents the point mutation frequency of each site in the group with the same point mutation characteristics as described in A3-1), and n represents the number of sites in the group with the same point mutation characteristics as described in A3-1); A3-3) A point mutation noise level acquisition module: used to obtain the point mutation noise level of each site using Formula 2 and Formula 3 based on the average point mutation frequency of the same point mutation feature group of each site and the point mutation frequency of each site: ； In formula 2 and formula 3: p represents the information of a certain site within the panel; λ tm represents the average point mutation frequency of all sites with the same point mutation characteristics grouped as site p; μ p represents the point mutation frequency of site p; ω1 represents the weight; ε p Represents the point mutation noise level at site p.

3. The device according to claim 1, characterized in that: A4) The insertion or deletion mutation noise level model construction module includes the following modules: A4-1) Insertion or deletion mutation feature integration grouping module: used for grouping sites with the same repetitive sequence element information and the same chromosome segment duplication information into a group based on the repetitive sequence element information of each site and the chromosome segment duplication information of each site to obtain the same insertion or deletion mutation feature grouping; A4-2) Insertion or deletion mutation prior background noise level acquisition module: used to calculate the average insertion or deletion mutation frequency of the insertion or deletion mutation characteristics of the same group based on the insertion or deletion mutation frequency of each site and A4-1) the insertion or deletion mutation characteristics of the same group using formula 4: ； In formula 4: rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics, μ prsl represents the insertion or deletion mutation frequency of each site in the group with the same insertion or deletion mutation characteristics as described in A4-1), and m represents the number of sites in the group with the same insertion or deletion mutation characteristics as described in A4-1); A4-3) Insertion or deletion mutation noise level acquisition module: used to obtain the insertion or deletion mutation noise level of each site using Formula 5 and Formula 6 based on the average insertion or deletion mutation frequency of the same group of insertion or deletion characteristics of each site and the insertion or deletion mutation frequency of each site: ； In Equation 5 and Equation 6, p represents the information of a certain site within the Panel; ω2 represents the weight, λ rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics as site p, μ pl represents the insertion or deletion mutation frequency of site p, ε pl Represents the insertion or deletion mutation noise level at site p.

4. A device for identifying whether a mutation detection result in ctDNA sequencing data is a true mutation or sequencing noise, characterized in that: The device comprises the following modules: C1) ctDNA sequencing data quality control and comparison module: used to obtain ctDNA sequencing data of the sample to be tested, and compare it with the human reference genome after quality control to obtain a comparison file; C2) ctDNA sequencing data mutation analysis module: used to obtain the sequencing depth, mutation reads support number and mutation frequency of the mutation site in the ctDNA sequencing data of the sample to be tested based on the comparison file, and obtain the mutation detection result in the ctDNA sequencing data of the sample to be tested; C3) Statistical test module: used to determine whether the mutation detection result is a true mutation or sequencing noise based on the mutation detection result and the noise level using a statistical hypothesis test method; the mutation is SNV and / or InDel; the noise level is the SNV noise level and / or the InDel noise level, and the noise level is obtained using the device described in any one of claims 1 to 3.

5. The device according to claim 4, characterized in that: The statistical hypothesis test uses Formula 7 and Formula 8 to calculate the statistical significance level of the hypothesis test: ； In formula 7 and formula 8: p represents the information of a certain site within the panel range, δ p is the noise level at site p; d p is the sequencing depth of site p, λ p is the number of noise mutation reads support expected to be observed at site p; k p is the number of mutation reads supporting site p in the sample to be tested; j represents the value of mutation reads supporting site p, j∈(0,k p ); P(k p ,λ p ) represents k p Greater than or equal to λ p The statistical P value size.

6. A method for detecting noise levels in ctDNA sequencing data, characterized in that: The method comprises the following steps: B1) ctDNA sequencing data processing: ctDNA sequencing data of healthy subjects are obtained, and the comparison files are obtained by comparing with the human reference genome after quality control; B2) Mutation frequency detection: based on the alignment file, the sequencing depth of each site in the ctDNA sequencing data, the number of point mutation reads supporting each site, and the number of insertion or deletion mutation reads supporting each site are obtained; the point mutation frequency of each site is calculated based on the sequencing depth of each site and the number of point mutation reads supporting each site; the insertion or deletion mutation frequency of each site is calculated based on the sequencing depth of each site and the number of insertion or deletion mutation reads supporting each site; B3) Point mutation noise level model construction: based on the point mutation feature grouping information of each site and the point mutation frequency of each site, the point mutation noise level of each site is obtained; the point mutation feature grouping information of each site includes the trinucleotide sequence of each site and the number of alignments of each site; B4) Construction of insertion or deletion mutation noise level model: based on the insertion or deletion mutation feature grouping information of each site and the insertion or deletion mutation frequency of each site, the insertion or deletion mutation feature grouping information of each site includes the repetitive sequence element information of each site and the chromosome segment duplication information of each site; B5) Noise level output: output the noise level of the ctDNA sequencing data; the noise level includes the SNV noise level of each site and / or the InDel noise level of each site.

7. The method according to claim 6, characterized in that: B3) The point mutation noise level model construction comprises the following steps: B3-1) Point mutation feature integration grouping: Based on the trinucleotide sequence of each site and the number of alignments of each site, sites with the same trinucleotide sequence and the same number of alignments are grouped together to obtain groups with the same point mutation features; B3-2) Obtaining the prior background noise level of point mutations: Based on the point mutation frequency of each site in the group with the same point mutation characteristics and B3-1) the group with the same point mutation characteristics is calculated using formula 1 to obtain the average point mutation frequency of the group with the same point mutation characteristics: ； In formula 1: tm represents the average point mutation frequency of all sites in the same group with the same point mutation characteristics, μ ptm represents the point mutation frequency of each site in the group with the same point mutation characteristics as described in B3-1), and n represents the number of sites in the group with the same point mutation characteristics as described in B3-1); B3-3) Obtaining the point mutation noise level: Based on the average point mutation frequency of the group with the same point mutation characteristics of each site and the point mutation frequency of each site, the point mutation noise level of each site is obtained using Formula 2 and Formula 3: ； In formula 2 and formula 3, p represents the information of a certain site within the panel range, λ tm represents the average point mutation frequency of all sites with the same point mutation characteristics grouped as site p; μ p represents the point mutation frequency of site p; ω1 represents the weight; ε p Represents the point mutation noise level at site p.

8. The method according to claim 6, characterized in that: B4) The construction of the insertion or deletion mutation noise level model comprises the following steps: B4-1) Integration and grouping of insertion or deletion mutation characteristics: Based on the repetitive sequence element information of each site and the chromosome segment duplication information of each site, sites with the same repetitive sequence element information and the same chromosome segment duplication information are grouped together to obtain groups with the same insertion or deletion mutation characteristics; B4-2) Obtaining the a priori background noise level of insertion or deletion mutations: Based on the insertion or deletion mutation frequency of each site and B4-1) the insertion or deletion mutation characteristics of the same group, the average insertion or deletion mutation frequency of the same group of insertion or deletion mutation characteristics is calculated using Formula 4: ； In formula 4: rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics, μ prsl represents the insertion or deletion mutation frequency of each site in the group with the same insertion or deletion mutation characteristics as described in B4-1), and m represents the number of sites in the group with the same insertion or deletion mutation characteristics as described in B4-1); B4-3) Obtaining the insertion or deletion mutation noise level: Based on the average insertion or deletion mutation frequency in the same group of insertion or deletion mutation characteristics of each site and the insertion or deletion mutation frequency of each site, the insertion or deletion mutation noise level of each site is obtained using Formula 5 and Formula 6: ； In Equations 5 and 6, p represents the information of a certain site within the Panel; ω2 represents the weight, λ rsl represents the average insertion or deletion mutation frequency of all sites in the same group with the same insertion or deletion mutation characteristics as site p, μ pl represents the insertion or deletion mutation frequency of site p, ε pl Represents the insertion or deletion mutation noise level at site p.

9. A method for identifying whether a mutation detection result in ctDNA sequencing data is a true mutation or sequencing noise, characterized in that: The method comprises the following steps: D1) ctDNA sequencing data quality control and comparison: Obtain ctDNA sequencing data of the sample to be tested, and compare it with the human reference genome after quality control to obtain a comparison file; D2) mutation analysis of ctDNA sequencing data: obtaining the sequencing depth, mutation reads support number and mutation frequency of the mutation site in the ctDNA sequencing data of the sample to be tested based on the comparison file, and obtaining the mutation detection result in the ctDNA sequencing data of the sample to be tested; D3) Statistical test analysis: Based on the mutation detection result and the noise level, a statistical hypothesis test method is used to determine whether the mutation detection result is a true mutation or sequencing noise; the mutation is SNV and / or InDel; the noise level is SNV noise level and / or InDe noise level, and the noise level is obtained by the method described in any one of claims 6 to 8.

10. The method according to claim 9, characterized in that: The statistical hypothesis test uses Formula 7 and Formula 8 to calculate the statistical significance level of the hypothesis test: ； In formula 7 and formula 8: p represents the information of a certain site within the panel range, δ p is the noise level at site p; d p is the sequencing depth of site p, λ p is the number of noise mutation reads support expected to be observed at site p; k p is the number of mutation reads supporting site p in the sample to be tested; j represents the value of mutation reads supporting site p, j∈(0,k p ); P(k p ,λ p ) represents k p Greater than or equal to λ p The statistical P value size.

11. A computer-readable storage medium storing a computer program, characterized in that: The computer-readable storage medium is a computer-readable storage medium for detecting noise levels in ctDNA sequencing data; the computer program enables the computer to execute the steps of the method described in any one of claims 6 to 8.

12. A computer-readable storage medium storing a computer program, characterized in that: The computer-readable storage medium is a computer-readable storage medium for identifying whether the mutation detection results in ctDNA sequencing data are true mutations or sequencing noise; the computer program enables the computer to run the steps of the method described in claim 9 or 10.

13. Any of the following uses of the device according to any of claims 1 to 5: E1) Use in the preparation of products for screening or assisting in screening of cancer patients; E2) Application in the development or preparation of products for personalized cancer treatment.

14. Any of the following uses of the device according to claim 4 or 5: E1) Use in the preparation of products for screening or assisting in screening of cancer patients; E2) Application in the development or preparation of products for personalized cancer treatment.

15. Any of the following uses of the method according to any of claims 6 to 8: E1) Use in the preparation of products for screening or assisting in screening of cancer patients; E2) Application in the development or preparation of products for personalized cancer treatment.

16. Any of the following uses of the method according to claim 9 or 10: E1) Use in the preparation of products for screening or assisting in screening of cancer patients; E2) Application in the development or preparation of products for personalized cancer treatment.

Images