Detection method and system for NGS sample pollution

Abstract

The invention relates to a method for detecting NGS sample contamination and a system using the method. Specifically, the method provided by the invention judges whether a sample is polluted or not according to the proportion of a new mutation type in a linked SNP gene locus, and the specific pollution degree of the sample is estimated based on the signal intensity of the linked new mutation type, so that the method is used for detecting NGS sample pollution.

Description

Technical Field

[0001] The present invention relates to a method for detecting contamination of NGS samples and a system using the method. The method of the present invention can directly estimate the contamination of a sample without removing CNV regions and LOH regions. Background Art

[0002] Next-Generation Sequencing (NGS) is a high-throughput, fast, and accurate DNA sequencing method that is widely used in genomics, medical research, and biological research. Compared with traditional Sanger sequencing, next-generation sequencing has higher efficiency and lower cost, and is widely used to reveal the association between gene variation and disease, and to study population genetics and evolutionary processes. However, the accuracy and reliability of NGS data may be affected by sample contamination.

[0003] Sample contamination mainly refers to the possibility of contamination from multiple links, including sample collection, processing, transportation, storage, and laboratory operations (such as improper pipetting, cross-use of pipette tips, aerosol contamination, etc.). During the processing, cross-contamination and exogenous genomic contamination will not only reduce data quality, but may also mislead research results and conclusions. Therefore, developing effective methods to detect the degree of contamination in samples is crucial to ensure the reliability of subsequent analysis results.

[0004] SNP (single nucleotide polymorphism) sites are not only important markers of genetic variation, but can also be used to identify the presence and extent of sample contamination. By comparing the expected SNP distribution pattern with the actual detected SNP data, it is possible to infer whether there is exogenous DNA mixed in, thereby quantifying the severity of sample contamination.

[0005] In terms of research methodology, commonly used techniques include statistical analysis of SNP sites based on mutation abundance, using models to estimate contamination levels, and confirming the reliability of results through validation experiments. These methods require not only highly accurate data processing and statistical algorithms, but also quality control measures in laboratory processes to ensure the accuracy and repeatability of the analysis.

[0006] Current methods for assessing sample contamination include VerifyBamID, ContEst, and Conpair. The VerifyBamID2 tool estimates the degree of sample contamination using likelihood probability based on population frequency and Harwin's law. The basic principle of Conpair is the same as that of VerifyBamID2, but it starts with homozygous SNP sites, simplifies the model, and adds control samples for joint judgment, so it can estimate samples with a contamination degree of more than 50%. ContEst uses genotype information and sequencing data to calculate the posterior probability through the Bayesian method to determine the maximum posterior probability estimate of the cross-individual contamination level. However, these software methods are all based on the diploid model. When large-scale copy number changes such as CNV (copy number variation) and LOH (loss of heterozygosity) occur in the sample, the diploid model will no longer be applicable, and the estimation of the degree of sample contamination based on this will be seriously distorted, causing interference and misjudgment in subsequent analysis.

[0007] CN115394357A discloses a site combination for judging sample pairing or contamination and its screening method and application, wherein the Conpair software is used to judge sample pairing or contamination and its degree. However, the disclosure is based on a diploid model and cannot be applied when there are large-scale copy number changes such as CNV (copy number variation) and LOH (loss of heterozygosity) in the sample.

[0008] CN116935966A discloses a method and device for determining contamination of high-throughput sequencing paired data, wherein after SNV (single nucleotide variation), CNV (copy number variation) and LOH (loss of heterozygosity) mutation detection is performed on the sequencing data of the sample and the control sample, the SNV mutation sites in the CNV region and the LOH region are filtered out and then the slice analysis is performed. However, the disclosure cannot directly estimate the contamination degree of the sample without removing the CNV region and the LOH region.

[0009] Therefore, there is an urgent need for a method and system for detecting contamination of NGS samples, which can directly estimate the contamination of samples without removing CNV regions and LOH regions. Summary of the invention

[0010] The inventors performed sample contamination detection based on the newly emerged mutations in the linked SNP loci. Specifically, the contamination of the sample was determined based on the linkage ratio of the newly emerged mutations, and / or the specific contamination degree of the sample was estimated based on the signal intensity of the linked newly emerged mutations. It was found that this method can efficiently and accurately determine whether the NGS sequencing sample is contaminated and give the specific degree of contamination.

[0011] The present invention is based on the above findings. Therefore, one aspect of the present invention relates to a method for detecting contamination of NGS samples, characterized in that the method comprises:

[0012] Extraction of linked loci: Obtain a baseline SNP site set and extract pairwise linked loci;

[0013] Determination of sample contamination characteristics: Count the genotypes in the linked SNP locus list to determine the contamination characteristics;

[0014] Calculation of linkage ratio of new mutations: Count the number of new mutations in the sample and calculate the linkage ratio of new mutations;

[0015] Determine the sample contamination threshold: Determine the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value;

[0016] Sample contamination determination: Determine the contamination status of the sample based on the threshold.

[0017] Another aspect of the present invention relates to a method for detecting contamination of NGS samples, characterized in that the method comprises:

[0018] Extraction of linked loci: Obtain a baseline SNP site set and extract pairwise linked loci;

[0019] Determination of sample contamination characteristics: Count the genotypes in the linked SNP locus list to determine the contamination characteristics;

[0020] Pollution model fitting: Statistical pollution characteristics, and function fitting of the characteristic statistical mean and pollution theoretical value;

[0021] Sample contamination determination: Determine the degree of sample contamination based on the contamination feature fitting model.

[0022] Another aspect of the present invention relates to a method for detecting contamination of NGS samples, characterized in that the method comprises:

[0023] Extraction of linked loci: Obtain a baseline SNP site set and extract pairwise linked loci;

[0024] Determination of sample contamination characteristics: Count the genotypes in the linked SNP locus list to determine the contamination characteristics;

[0025] Calculation of linkage ratio of new mutations: Count the number of new mutations in the sample and calculate the linkage ratio of new mutations;

[0026] Determine the sample contamination threshold: Determine the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value;

[0027] Pollution model fitting: Statistical pollution characteristics, and function fitting of the characteristic statistical mean and pollution theoretical value;

[0028] Sample contamination determination: The contamination state and degree of contamination of the sample are jointly determined based on the threshold and contamination feature fitting model.

[0029] Another aspect of the present invention relates to a detection system for NGS sample contamination, characterized in that:

[0030] The system comprises: a linkage locus extraction module, a sample contamination feature determination module, a linkage ratio calculation module for new mutation types, a sample contamination threshold determination module, a sample contamination judgment module, and an output module;

[0031] or

[0032] The system comprises: a linked gene locus extraction module, a sample contamination feature determination module, a contamination model fitting module, a sample contamination determination module, and an output module;

[0033] or

[0034] The system comprises: a linkage locus extraction module, a sample contamination feature determination module, a linkage ratio calculation module for new mutation types, a sample contamination threshold determination module, a contamination model fitting module, a sample contamination determination module, and an output module;

[0035] in,

[0036] Linked loci extraction module: obtain the baseline SNP site set and extract the pairwise linked loci;

[0037] Sample contamination feature determination module: count the genotypes in the linked SNP locus list and determine the contamination features;

[0038] Linkage ratio calculation module for new mutations: count the number of new mutations in the sample and calculate the linkage ratio of new mutations

[0039] Sample contamination threshold determination module: Determines the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value;

[0040] Pollution model fitting module: Count pollution characteristics and perform function fitting between characteristic statistical mean and pollution theoretical value;

[0041] Sample contamination determination module: determines the contamination state and / or degree of contamination of the sample based on the threshold and / or contamination feature fitting model

[0042] Output module: outputs sample pollution characteristic statistical information, pollution status and / or pollution degree determination information. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a process analysis diagram of the present invention.

[0044] Figure 2 is the chain ratio of different pollution levels R LD Distribution box plot.

[0045] Figure 3 is the pollution characteristic mean M LD Fitting curve of pollution degree.

[0046] Figure 4 is a box plot of the performance distribution of the asymptotic logarithmic model on the test cases. DETAILED DESCRIPTION

[0047] Hereinafter, various exemplary embodiments of the present invention are described in detail. This detailed description should not be considered as a limitation of the present invention, but should be understood as a more detailed description of certain aspects, characteristics and embodiments of the present invention. It should be understood that the terms described in the present invention are only for describing a particular embodiment and are not intended to limit the present invention. In addition, for the numerical range in the present invention, it should be understood that the upper and lower limits of the range and each intermediate value therebetween are specifically disclosed. Each smaller range between the intermediate value in any stated value or stated range and any other stated value or intermediate value in the range is also included in the present invention. The upper and lower limits of these smaller ranges may be independently included or excluded in the range.

[0048] Unless otherwise indicated, all technical and scientific terms used herein have the same meanings as those generally understood by those skilled in the art. Although the present invention describes only preferred methods and materials, any methods and materials similar or equivalent to those described herein may also be used in the implementation or testing of the present invention. All documents mentioned in this specification are incorporated by reference to disclose and describe the methods and / or materials associated with the documents. In the event of a conflict with any incorporated document, the content of this specification shall prevail.

[0049] The relevant abbreviations and terms involved in the present invention are as follows:

[0050] NGS (Next-Generation Sequencing): The second generation sequencing technology or next generation sequencing technology is a high-throughput, low-cost gene sequencing method. The second generation sequencing is widely used in multiple research directions such as genome sequencing, exome sequencing, transcriptome sequencing (RNA-seq), metagenomics and single-cell sequencing. In addition, it also has important application value in cancer research, genetic disease diagnosis, pathogen detection and personalized medicine.

[0051] Panel: A target region covered by sequencing analysis of multiple sites or genes. Generally speaking, larger panels cover more than 300 genes.

[0052] SNP (Single Nucleotide Polymorphism): Single nucleotide polymorphism refers to the variation of base types at a single nucleotide site with a higher frequency in the genomic DNA sequence. The frequency can be set as needed. Generally, base variations at a single nucleotide site on an autosome with a gene frequency higher than 1% can be considered SNPs. In some cases, the frequency can be set to 5%.

[0053] CNV (Copy Number Variation): Copy number variation refers to the change in the number of DNA copies of a larger segment in the genome compared to a normal individual. Unlike SNP, CNV involves the duplication or deletion of a larger DNA sequence in the genome (usually ranging from a few thousand base pairs to millions of base pairs).

[0054] LOH (Loss of Heterozygosity): Loss of heterozygosity refers to the loss or inactivation of one of the two alleles at a locus in certain diseases such as tumors, causing the region to change from a heterozygous state to a homozygous state. This phenomenon may occur in tumor cells and is an important mechanism for the occurrence and development of cancer.

[0055] Linked inheritance: The phenomenon that two or more genes located on the same chromosome tend to be passed on to offspring together during the inheritance process due to their physical proximity. This concept is used in the present invention, specifically referring to the phenomenon that two SNPs within a specified distance range tend to be passed on to offspring together during the inheritance process.

[0056] Linked SNP loci: Two or more SNPs in a genome that are genetically linked.

[0057] De novo mutation: A genetic variation that appears in an individual's genome for the first time and is not inherited from a parent.

[0058] Quartile method: A statistical descriptive analysis method used to describe the distribution of data, such as the central tendency and dispersion of data. The interquartile range (IQR) can be used to identify outliers. When a data set is divided into four equal parts, namely Q1 (1st quartile), Q2 (2nd quartile) and Q3 (3rd quartile), the IQR is defined as Q3–Q1, which measures the dispersion of the data set. And, data outside Q3+1.5*IQR or Q1-1.5*IQR are considered outliers.

[0059] Standard deviation method: a method to describe the degree of dispersion of data. The standard deviation is the square root of the arithmetic mean of the squared deviations from the mean, and is most commonly used as the degree of statistical distribution in probability statistics. Outliers can be identified using the mean and standard deviation of the data. First, calculate the mean and standard deviation of the data; then define a threshold, usually a multiple of the standard deviation (usually 2 or 3 times the standard deviation) as the threshold, which determines what kind of data points are considered outliers; then, calculate the difference between each data point and the mean, and then compare this difference with the threshold. If the difference exceeds the threshold, the data point is considered an outlier.

[0060] The present invention is based on the principle of SNP site linkage inheritance to judge and estimate the contamination of the sample. A linked SNP locus (SNP1: ref1>alt1, SNP2: ref2>alt2) has four genotypes according to the combination of ref and alt, namely: ref1 / ref2, ref1 / alt2, alt1 / ref2, alt1 / alt2. In a normal sample, a completely linked SNP locus is composed of a paternal genome and a maternal genome, and its own genotype may only appear at most two (if the paternal and maternal genomes are the same, only one will appear, and if the paternal and maternal genomes are different, two will appear). Chromosome crossover, background sequencing noise, false positives caused by sample damage, and exogenous genome contamination will all lead to new genotypes in actual sample detection, which are also called new mutations in the present invention. The present invention first limits the genotypes caused by chromosome crossover by screening SNP pairs within a specified distance range, and counts the reads support numbers of the four genotypes of each linked SNP locus, records with no reads support number as 0, and sorts the four genotypes from small to large according to the reads support number. The first two digits in a normal sample must be 0, because a genotype can only be inherited in two ways at most, and the inherited genotype occupies the last two digits after sorting. If the first two digits are not 0, it indicates that there is a new mutation in the sample, that is, there may be contamination in the sample. In a preferred embodiment, the interference of background noise caused by sequencing and false positives caused by sample damage can be further eliminated by setting the minimum detection limit of the reads support number to improve accuracy. This method uses new mutations as contamination features, does not rely on the diploid model, and can eliminate the interference of CNV and LOH in the sample to be tested, and has superior and stable detection performance.

[0061] The present invention determines whether a sample is contaminated by identifying the proportion of newly emerged mutations in linked SNP loci in the sequencing sample, and estimates the specific contamination degree of the sample based on the signal strength of the linked newly emerged mutations.

[0062] This method can efficiently and accurately determine whether the NGS sequencing sample is contaminated and give a specific degree of contamination. The method is also compatible with CNV and LOH that may exist in NGS sequencing, that is, in the case of CNV and LOH in the sample, the method of the present invention can also be undisturbed and has strong robustness and stability.

[0063] The method of the present invention comprises the following steps:

[0064] 1Select linked SNP loci:

[0065] 1.1 Baseline SNPs can be obtained from public population databases, such as ExAC, GnomAD, 1KG, iJGVD, etc. Preferably, the East Asian Thousand People Database (1KG_EAS) is recommended. If a large number of large panel clinical examination samples have been accumulated, a baseline database can be constructed based on the sample data. The principle of SNP selection is to select autosomal single base sites with stable coverage depth and high population heterozygosity:

[0066] (1) extracting single-base SNP sites on autosomes with a gene frequency higher than 1% in the population database within the current detection range, and the preferred population gene frequency threshold can be set to 5%;

[0067] (2) Eliminate SNPs located in the genomic repetitive interval and high-frequency somatic mutation regions (such as TP53, KRAS, EGFR, etc.);

[0068] (3) Eliminate SNPs with unstable coverage depth (the threshold set here can be selected in the range of less than or equal to 10×, preferably, less than or equal to 80×) to obtain a baseline SNP site set;

[0069] 1.2 Extract a pair of SNPs within the specified distance range of the SNP candidate list, called linked SNP loci (the total number of current linked SNP loci is recorded as N LD ), preferably, the distance between a pair of SNPs is set to be less than 80bp, preferably less than 35bp.

[0070] 2 Filter sample sequencing reads:

[0071] 2.1 Use software such as bwa to align sample sequencing information to the genome; use software such as sambamba to mark duplicate reads in bam and extract reads covering two SNP sites in any linked SNP locus.

[0072] 2.2 Perform quality control filtering on the extracted reads: remove reads containing DUP (flag: 1024) in flag and softclipped type in Cigar, and remove reads with low alignment quality.

[0073] 3. Determine the pollution characteristics:

[0074] 3.1 There are four genotypes (ref1 / ref2, ref1 / alt2, alt1 / ref2, alt1 / alt2) in a linked SNP locus. The reads support number of each genotype is counted, and the record with no reads support number is 0. Each genotype is sorted from small to large according to the reads support number, and the genotype ranked first or second after sorting from small to large is selected as the first genotype, the second genotype (i.e., the new mutation type) or a combination of both or one of them as the contamination feature. In other words, when the four genotypes of the linked SNP locus are sorted from small to large according to the reads support number of the four genotypes, the contamination feature is the first genotype, or the contamination feature is the second genotype, or the contamination feature is a combination of the first genotype and one or more selected from the second genotype, the third genotype or the fourth genotype, or the contamination feature is a combination of the second genotype and one or more selected from the first genotype, the third genotype or the fourth genotype. Preferably, the contamination characteristic is the second genotype, or the contamination characteristic is the second genotype and one or more combinations selected from the first genotype, the third genotype or the fourth genotype. Most preferably, the contamination characteristic is the second genotype.

[0075] 3.2 Screening for available new mutations: New mutations are the genotypes identified as contamination features as described above. Considering the background noise caused by sequencing errors and sequencing technology limitations, and SNP false positives caused by DNA sequence damage during sample storage and processing, in order to further improve the accuracy, the reads support number V for new mutations can be adjusted based on the SNP positive judgment value of the current process. LD Set the filter threshold, V LD Only those above the threshold are considered to be usable new mutations. Preferably, the threshold can be set to 5. LD All refer to newly emerged mutations available after filtering.

[0076] 4. Determine the sample contamination status:

[0077] Count the number of all new mutations available in the sample n LD ; and calculate the linkage ratio of the new mutation type: R LD =n LD / N LD .

[0078] There is basically no signal of linked new mutation in normal samples. If the sample is contaminated, the linkage ratio R LD will increase significantly.

[0079] 4.1 Obtaining R under the specified pollution level of the training set LD Distribution;

[0080] 4.2 Using a data processing method to propose discrete values; as the data processing method, the quartile method (or standard deviation method) etc. may be used;

[0081] 4.3 Determine R at the current pollution level according to the fluctuation range LD Threshold, if the R LD If the value is higher than the threshold, it means that the sample to be tested has reached the current contamination level. LD The threshold can be determined by combining the specific analysis process to simulate the degree of contamination. For example, if the sample with a contamination degree exceeding 5% is required to be detected, the sample with 5% contamination can be statistically simulated. LD The fluctuation range of the value is determined, and the judgment threshold is determined according to the maximum value of the fluctuation range.

[0082] 5 Fitting the asymptotic logarithmic model:

[0083] The first genotype and the second genotype can be used as pollution features alone and modeled based on their reads support numbers to estimate the degree of pollution, or the pollution feature can be a combination of genotypes, that is, the combined expectation of the reads support numbers of the four genotypes is used for modeling. When the pollution feature is counted, the characteristic statistical value is based on the reads support number of the pollution feature or the value obtained based on the combined expectation of the reads support numbers of the four genotypes. In view of the fact that after sorting, the first genotype statistical value will only be non-0 when the sample and the pollution source are both heterozygous, and the second genotype statistical value can be non-0 as long as one of them is heterozygous. Therefore, preferably, the characteristic statistical value is based on the reads support number of the second genotype. The following uses the second genotype as a pollution feature for modeling to illustrate.

[0084] 5.1 Obtain the available SNP second genotype reads support number V according to step 3 LD , for the same sample, all the V values ​​of the linked SNP loci after screening LD Sum, the corresponding depth DP LD Sum and divide the two to obtain the sample's pollution characteristic statistical value M i (M i =ΣV LD / ΣDP LD );

[0085] 5.2 Statistics of pollution characteristics M of all training samples with the same pollution level i ;

[0086] 5.3 After removing discrete values ​​using the quartile method (or standard deviation method), calculate the mean value of the contamination characteristic M LD ;

[0087] 5.4 Since the newly emerged mutant type is the second genotype after sorting, this value has a theoretical upper limit. LD The theoretical value of the current pollution level is fitted into the asymptotic logarithmic function α LD =-a(1-b ln M LD ), solve the parameters a and b, and obtain the pollution fitting model.

[0088] 6. Predict pollution level:

[0089] Based on the characteristic statistical method in step 5.1, obtain the contamination characteristic statistical value M of the sample to be tested i , substitute into the model to estimate the contamination degree of the sample to be tested.

[0090] A variety of models can be used to fit the sample contamination level, including logarithmic models, polynomial models, exponential models, linear models, nonlinear models, Bayesian models, generalized linear models, etc. Any regression model, probability distribution model or machine learning model based on new mutations can be used in this step to flexibly adapt to different application requirements and data characteristics.

[0091] The key point of the present invention is to detect sample contamination based on the newly emerged mutations in the linked genotype, determine whether the sample is contaminated by the proportion of newly emerged mutations, or directly estimate the degree of sample contamination by modeling the reads support number of newly emerged mutations. Compared with the existing contamination detection methods, the present invention has the following advantages:

[0092] 1. Contamination detection based on baseline SNP sites can improve accuracy and reliability. Baseline SNP sites are ubiquitous in the entire sample set, have high polymorphism and stability, and can therefore more accurately reflect the mutation status of the sample.

[0093] 2 The detection method of the present invention is universal. The baseline SNP site can be selected according to specific needs, and does not depend on a specific sample type or panel design. It can be applied to samples of different sources and types, and has strong applicability and versatility, especially compatible with NGS tumor samples. Therefore, no matter what panel or sample is used, the detection method can be directly applied, reducing the workload and time cost of algorithm development and verification.

[0094] 3 The detection method of the present invention is statistically simple and efficient. By detecting the genotype of the linked SNP loci, the contamination status of the sample can be quickly obtained. The algorithm has low computational complexity and is suitable for large-scale NGS data analysis.

[0095] 4 The detection method of the present invention is stable and robust. In the fitting of the contamination degree, compared with the traditional diploid model, the present invention starts from a completely new perspective and directly uses the newly emerged mutation type of linked inheritance to fit the contamination degree. It is not limited by the diploid genotype and has obvious advantages in tumor samples with ploidy changes such as CNV and LOH. Therefore, the method of the present invention can be appropriately used for the detection of contamination in NGS samples from tumors.

[0096] In summary, the present invention provides a method and system for determining whether an NGS sample is contaminated and estimating the specific degree of contamination based on baseline SNP loci and linkage genetic estimation. The detection method has high accuracy, versatility and computational efficiency, and the invention has important application prospects and commercial value in the field of oncology research and clinical diagnosis.

[0097] Those skilled in the art will appreciate that all or part of the functions of the above method can be implemented by hardware or by computer program. When all or part of the functions in the above method are implemented by computer program, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, disk, CD, hard disk, etc., and the program is executed by a computer to implement the above functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above functions can be implemented. In addition, when all or part of the functions in the above embodiment are implemented by computer program, the program can also be stored in a storage medium such as a server, another computer, disk, CD, flash disk or mobile hard disk, and saved in the memory of the local device by downloading or copying, or the system of the local device is updated, and when the program in the memory is executed by the processor, all or part of the functions in the above method can be implemented.

[0098] Therefore, the present invention also discloses a system for detecting contamination of NGS samples according to the present invention, wherein

[0099] The system comprises: a linkage locus extraction module, a sample contamination feature determination module, a linkage ratio calculation module for new mutation types, a sample contamination threshold determination module, a sample contamination judgment module, and an output module;

[0100] or

[0101] The system comprises: a linked gene locus extraction module, a sample contamination feature determination module, a contamination model fitting module, a sample contamination determination module, and an output module;

[0102] or

[0103] The system comprises: a linkage locus extraction module, a sample contamination feature determination module, a linkage ratio calculation module for new mutation types, a sample contamination threshold determination module, a contamination model fitting module, a sample contamination determination module, and an output module;

[0104] a) Linked loci extraction module: obtain a baseline SNP site set and extract pairwise linked loci;

[0105] b) Sample contamination feature determination module: Count the genotypes in the linked SNP locus list and determine the contamination features;

[0106] c) Module for calculating linkage ratio of new mutations: Count the number of new mutations in the sample and calculate the linkage ratio of new mutations

[0107] d) Sample contamination threshold determination module: Determines the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value;

[0108] e) Pollution model fitting module: statistical pollution characteristics, function fitting of characteristic statistical mean and pollution theoretical value;

[0109] f) Sample contamination determination module: determines the contamination state and / or contamination degree of the sample based on a threshold value and / or contamination feature fitting;

[0110] g) Output module: outputs sample pollution characteristic statistical information, pollution status and / or pollution degree determination information.

[0111] The present invention also discloses a device for detecting contamination of NGS samples, which includes a memory and a processor; the memory includes a device for storing programs; the processor includes a device for implementing the method for detecting contamination of NGS samples of the present invention by executing the programs stored in the memory.

[0112] The present invention also discloses a computer-readable storage medium, in which a program is stored. The program can be executed by a processor to implement the NGS sample contamination detection method of the present invention.

[0113] The present invention is further described below with reference to the embodiments.

[0114] Example

[0115] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content of the present invention, those skilled in the art may make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the present invention equally.

[0116] In all embodiments of the present invention, the following training samples and test samples are used.

[0117] 17 sequencing samples were mixed and simulated to construct a specified contamination degree, of which 12 (training case numbers S1-S9, test case numbers S10-S12) were used as contaminated samples, and 5 (training case numbers C1-C3, test case numbers C4-C5) were used as contamination sources. 13 contamination sources with different contamination degrees were constructed (the gradients of the 13 contamination sources were 0.99:0.01, 0.97:0.03, 0.95:0.05, 0.92:0.08, 0.9:0.1, 0.88:0.12, 0.85:0.15, 0.8:0.2, 0.75:0.25, 0.7:0.3, 0.65:0.35, 0.6:0.4, and 0.55:0.45, respectively). A total of 429 (9×3×13+3×2×13) fastq files were constructed.

[0118] Example 1: Sample contamination detection based on linkage ratio of new mutations

[0119] 1.1 Selection of baseline SNP loci and acquisition of linked SNPs

[0120] The autosomal single-base SNP sites with a population frequency higher than 5% within the specified detection interval in the population database 1KG_EAS were extracted, and those with coordinates located in the repeated segments of the genome and common hotspot intervals such as high-frequency mutation regions (TP53, KRAS, EGFR, etc.) were eliminated. The SNP sites detected in the training samples with stable coverage depth (DP average higher than 80×) were retained as the baseline SNPs site set (total number of SNPs N, a total of 25,529 SNP sites that can be used for the baseline were obtained in this example).

[0121] Extract a pair of SNPs within the specified distance range of the SNP candidate list (total number of linked SNP loci N LD In this embodiment, a total of 2273 linked loci were obtained). In order to achieve complete linkage as much as possible, a maximum linkage distance of 80 bp is recommended. In this embodiment, a distance of 35 bp is used to extract linked SNP loci.

[0122] 1.2 Filtering sample sequencing reads

[0123] The fastq was quality-controlled and filtered, the adapters were removed, and low-quality unqualified reads were eliminated. The software bwa (v0.7.15) was used to align it to the reference genome (hg19). The software sambamba (v1.0.0) subcommand sort was used for sorting, and the subcommand markdup was used to mark the reads for duplication, and finally the result file in bam format was obtained.

[0124] After obtaining the candidate linked SNP list, it was converted into a bed file according to the coordinates, and the python package pysam was used to obtain all reads in the linked bed interval range, and the reads were quality-controlled and filtered: flags containing DUP (flag: 1024) were removed, softclipped types in Cigar were removed, and reads with low alignment quality averages were removed (the alignment quality threshold in this example was 20). Then, the mutant types of the linked SNPs pairs were prepared for statistics.

[0125] 1.3 Determination of pollution characteristics

[0126] There are four combined genotypes (ref1 / ref2, ref1 / alt2, alt1 / ref2, alt1 / alt2) for a linked SNP locus. The reads support number corresponding to these four combined genotypes and the total coverage depth of the locus are counted, and the four combined genotypes are sorted from small to large according to the reads support number. The second genotype (i.e., the new mutation type) is selected as the contamination feature for subsequent analysis. Considering the false positives caused by background sequencing noise and sample damage, the reads support number V for the new mutation type is LD Set the filtering threshold (in this embodiment, the threshold is set to 5 according to the SNP positive judgment value of the current process), V LD A mutation above the threshold is considered to be an available new mutation.

[0127] 1.4 Chain ratio calculation

[0128] By setting the minimum coverage depth of 80× on the linked loci to ensure the stability of the mutation signal, the total number of linked SNP loci N was obtained. LD , count the number of new mutations currently available n LD , and calculate the proportion of new mutations, recorded as linkage proportion R LD , R LD =n LD / N LD .

[0129] 1.5 Determination of sample contamination status

[0130] Statistical simulation of pollution in 351 training samples R LD The distribution of the linkage ratio R LD The distribution of Figure 2 As shown, R is defined as needed according to the fluctuation range. LD The threshold value for determining whether a sample is contaminated is used. In this embodiment, the contamination level should not exceed 0.05, so R LD>0.085 indicates that the sample is contaminated.

[0131] According to the chain ratio R LD The 78 test samples were individually tested for contamination, and the test results are shown in Table 1. The results show that all contamination tests with a theoretical contamination level lower than 0.05 were judged as contaminated, and all contamination tests with a theoretical contamination level higher than 0.05 were judged as normal. The results were in line with expectations.

[0132] Table 1 Pollution detection statistics

[0133]

[0134]

[0135]

[0136] Note:

[0137] Test: Test contaminated samples

[0138] Ctrl: Test pollution source sample

[0139] Expect: Theoretical contamination level

[0140] R LD : Linkage ratio, the ratio of the number of new mutations to the total number of linked SNP loci

[0141] Status: Sample contamination status, no contamination is recorded as PASS, contamination is recorded as Fail

[0142] Example 2: Sample contamination detection based on signal intensity of linked de novo mutations

[0143] 2.1 Constructing a fitted asymptotic logarithmic model

[0144] According to steps 1.1-1.3 described in Example 1, the mutation reads support numbers of the four genotypes of each linked SNP are sorted from small to large, and the reads support number of the second genotype after screening (V LD ), V for all linked SNPs in the same sample LD Sum, the corresponding depth DP LD Sum and divide the two to obtain the pollution characteristic statistical value M i (M i =ΣV LD / ΣDP LD ).

[0145] Count the pollution feature statistics of all training samples M i , group them according to the same pollution level, use the quartile method to eliminate discrete values ​​within the group, and calculate the pollution mean M for each groupLD , as shown in Table 2, the pollution degree is fitted to the asymptotic logarithmic function α LD =-a(1-b ln M LD ), the contamination mean M detected LD The simulated pollution theoretical value is brought into the curve_fit in the scipy module for curve fitting, and the function parameters a and b are solved to obtain: a = -0.36, b = 3.428, and the complete function relationship is α LD =-0.36(1-3.428lnM LD ), where R2 = 0.9963, the model curve is as follows Figure 3 shown.

[0146] Table 2 M under different pollution levels LD Statistics

[0147] Contamination <![CDATA[M LD ]]> 0.01 0.003 0.03 0.03 0.05 0.043 0.08 0.059 0.1 0.069 0.12 0.079 0.15 0.095 0.2 0.121 0.25 0.144 0.3 0.165 0.35 0.184 0.4 0.196 0.45 0.204

[0148] 2.2 Sample contamination degree detection

[0149] According to steps 1.1-1.3 described in Example 1, the available new mutation type (i.e., the second genotype after screening) reads support number V of the sample to be tested is obtained. LD , V of all linked SNPs in the sample to be tested LD Sum, the corresponding depth DP LD Sum and divide the two to obtain the pollution characteristic statistical value M i , substitute into the fitted function relationship to obtain the predicted value α of the contamination degree of the sample to be tested LD The contamination degree of 78 samples to be tested was estimated. The results are shown in Table 3. The theoretical value of simulated contamination is consistent with the predicted value α LD The distribution box plot of Figure 4 The coefficient of determination R of the model is shown in 2 =0.9227, the determination coefficient for the range of medium and low pollution levels 0.01-0.15 is R 2 =0.9548.

[0150] The results show that the model has good prediction ability, especially under medium and low pollution levels, indicating that the model has strong sensitivity for pollution detection.

[0151] Table 3 Statistics of pollution detection results

[0152]

[0153]

[0154] Note:

[0155] Test: Test contaminated samples

[0156] Ctrl: Test pollution source sample Expect: Theoretical pollution degree M i : Pollution characteristic statistical value α LD : Model predicts the contamination degree of samples Example 3: Sample contamination detection based on combination method

[0157] The linkage ratio R is obtained according to 1.1-1.4 described in Example 1. LD , according to Example 2, the model prediction value α is obtained LD According to the chain ratio R LD and the model predicted value α LD The 78 test samples were tested jointly, and the pollution prediction results were corrected twice. Specifically, when the pollution status of the 1.5 samples described in Example 1 was determined to be non-polluted but the model prediction value was higher than the pollution threshold, the prediction value was corrected to the specified value (0.0499 in this example), and the rest were output as they were. The results are shown in Table 4.

[0158] The results show that all samples with theoretical contamination levels not greater than 0.05 are judged as PASS, and all test samples with theoretical contamination levels greater than 0.05 are judged as Fail. The test results of all samples are in line with expectations.

[0159] Table 4 Model performance statistics under different pollution levels

[0160]

[0161]

[0162] Note:

[0163] Test: Test the contaminated sample Ctrl: Test the contamination source sample

[0164] Expect: Theoretical contamination level

[0165] R LD : Linkage ratio, the ratio of the number of new mutations to the total number of linked SNP loci

[0166] M i: Pollution characteristic statistics

[0167] α LD :The model predicts the degree of contamination of the sample

[0168] Status: Sample contamination status, no contamination is recorded as PASS, contamination is recorded as Fail.

Claims

1. A method for detecting NGS sample contamination, characterized in that: The method comprises: Extraction of linked loci: Obtain a baseline SNP site set and extract pairwise linked loci; Determination of sample contamination characteristics: Count the genotypes in the linked SNP locus list to determine the contamination characteristics; Calculation of linkage ratio of new mutations: Count the number of new mutations in the sample and calculate the linkage ratio of new mutations; Determine the sample contamination threshold: Determine the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value; Sample contamination determination: Determine the contamination status of the sample based on the threshold.

2. A method for detecting NGS sample contamination, characterized in that: The method comprises: Extraction of linked loci: Obtain a baseline SNP site set and extract pairwise linked loci; Determination of sample contamination characteristics: Count the genotypes in the linked SNP locus list to determine the contamination characteristics; Pollution model fitting: Statistical pollution characteristics, and function fitting of the characteristic statistical mean and pollution theoretical value; Sample contamination determination: Determine the degree of sample contamination based on the contamination feature fitting model.

3. A method for detecting NGS sample contamination, characterized in that: The method comprises: Extraction of linked loci: Obtain a baseline SNP site set and extract pairwise linked loci; Determination of sample contamination characteristics: Count the genotypes in the linked SNP locus list to determine the contamination characteristics; Calculation of linkage ratio of new mutations: Count the number of new mutations in the sample and calculate the linkage ratio of new mutations; Determine the sample contamination threshold: Determine the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value; Pollution model fitting: Statistical pollution characteristics, and function fitting of the characteristic statistical mean and pollution theoretical value; Sample contamination determination: The contamination state and degree of contamination of the sample are jointly determined based on the threshold and contamination feature fitting model.

4. The method according to any one of claims 1 to 3, characterized in that The baseline SNPs are from a public population database or a self-constructed baseline SNP database.

5. The method according to any one of claims 1 to 3, characterized in that The contamination feature is a new mutation type. Preferably, when the four genotypes are sorted from small to large according to the reads support number of the four genotypes of the linked SNP locus, The contaminating characteristic is the first genotype, or The contaminating characteristic is a second genotype, or The contamination characteristic is a combination of the first genotype and one or more selected from the second genotype, the third genotype or the fourth genotype, or The contamination characteristic is a combination of the second genotype and one or more selected from the first genotype, the third genotype or the fourth genotype; More preferably, the contamination characteristic is the second genotype, or the contamination characteristic is a combination of the second genotype and one or more selected from the first genotype, the third genotype or the fourth genotype; Most preferably, the contamination characteristic is a second locus genotype.

6. The method according to claim 2 or 3, characterized in that: The feature statistical mean is the average value of the feature statistical value, and the feature statistical value is based on the reads support number of the contamination feature or based on the combined expectation of the reads support number of four genotypes. Preferably, the feature statistical value is based on the reads support number of the second genotype.

7. The method according to any one of claims 1 to 3, characterized in that It also includes a data analysis process for filtering sample sequencing reads.

8. The method according to claim 2 or 3, characterized in that: The fitting model is a regression model, a machine learning model or a probability distribution model constructed based on pollution characteristics, preferably a logarithmic fitting model or a polynomial model fitting model.

9. The method according to any one of claims 1 to 3, characterized in that: In determining the sample contamination characteristics, it also includes screening for emerging mutants; preferably, emerging mutants with a reads support number higher than a filtering threshold are selected; preferably, the filtering threshold is 5.

10. A detection system for NGS sample contamination, characterized in that: The system comprises: a linkage locus extraction module, a sample contamination feature determination module, a linkage ratio calculation module for new mutation types, a sample contamination threshold determination module, a sample contamination judgment module, and an output module; or The system comprises: a linked gene locus extraction module, a sample contamination feature determination module, a contamination model fitting module, a sample contamination determination module, and an output module; or The system comprises: a linkage locus extraction module, a sample contamination feature determination module, a linkage ratio calculation module for new mutation types, a sample contamination threshold determination module, a contamination model fitting module, a sample contamination determination module, and an output module; in, Linked loci extraction module: obtain the baseline SNP site set and extract the pairwise linked loci; Sample contamination feature determination module: count the genotypes in the linked SNP locus list and determine the contamination features; Linkage ratio calculation module for new mutations: count the number of new mutations in the sample and calculate the linkage ratio of new mutations; Sample contamination threshold determination module: Determines the threshold of whether the sample is contaminated based on the simulated contamination theoretical value and the actual detection value; Pollution model fitting module: Count pollution characteristics and perform function fitting between characteristic statistical mean and pollution theoretical value; Sample contamination determination module: determines the contamination state and / or contamination degree of the sample based on a threshold value and / or a contamination feature fitting model; Output module: outputs sample pollution characteristic statistics, pollution status and / or pollution degree determination information.

11. A detection device, characterized in that: The device comprises a memory and a processor; the memory comprises a device for storing a program; the processor comprises a device for implementing the method according to any one of claims 1 to 3 by executing the program stored in the memory.

12. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program can be executed by a processor to implement the method according to any one of claims 1 to 3.

13. The method according to any one of claims 1 to 3, the system according to claim 10, the detection device according to claim 11, and the computer-readable medium according to claim 12, characterized in that: The NGS samples were from tumors.

Images