Virus co-infection detection method and system based on penalty regression and sequencing physical linkage

Abstract

The invention discloses a virus co-infection detection method and system based on penalty regression and sequencing physical linkage. The method comprises the following steps: collecting virus sequences and corresponding pedigree information thereof; counting and screening high-frequency variation to obtain representative variation of the pedigree; constructing a virus pedigree site variation map as a reference map of a reconstruction model by using representative variation of the pedigree; establishing an index for the reference genome, and comparing a to-be-reconstructed virus sequencing sample with the reference genome by using BWA; converting the compared sequencing sample into a BAM file in a binary format; performing variation detection on the virus sequencing samples after comparison to obtain single nucleotide polymorphism (SNP) and insertion / deletion (INDEL) and frequency information thereof; finally, constructing a regression equation according to the SNP and INDEL of the sample as well as the frequency and the reference map of the sample; and performing regression by using a square error loss function in combination with L0 regularization and L1 regularization, and finally obtaining the spectrum haplotype corresponding to the sample and the frequency thereof.

Description

Technical Field

[0001] The present invention relates to the field of bioinformatics, and in particular to a virus co-infection detection method and system based on penalized regression and sequencing physical linkage. Background Art

[0002] Viruses are a characteristic of many infectious diseases (such as the new coronavirus, acquired immunodeficiency syndrome, and influenza A virus) with large scale. Viral evolution can bring many advantages to the viral population, including stronger drug resistance and stronger transmission ability. In addition, mixed infection of multiple virus strains can also affect the development of the disease. For example, the high ICU hospitalization rate of new coronavirus cases is positively correlated with co-infection with Delta / Omicron virus strains. The strong adaptability of viruses to the environment is due to their genetic diversity. Therefore, accurate reconstruction of viral strain haplotypes, especially calculation of viral sequence haplotypes and their frequencies in co-infection cases, can provide a research basis for clarifying the mechanism of viral recombination evolution, and can also provide a basis for studying the relationship between virus transmission, co-infection and severe cases, and provide a reference for clinical treatment.

[0003] Thanks to the rapid development of biological sequencing technology, the characteristics of viral populations can be characterized by genetic sequencing of viral samples. The quality of sequencing data plays a crucial role in the effect of viral haplotype reconstruction. Although the third-generation sequencing technology can produce 100,000bp sequencing reads, the sequencing error rate is high. The second-generation sequencing technology can produce high-precision sequencing reads with a sequencing accuracy of 99.9%, but the generated sequencing reads are short and difficult to match single nucleotide polymorphism sites that are far away. Therefore, reconstruction of the viral genome through short sequencing reads will lose some linkage information.

[0004] Haplotype reconstruction can be divided into two main methods: reconstruction based on the reference genome and de novo assembly. De novo assembly requires a lot of computing resources and computing time, and the amount of data required is usually large. The assembly process is prone to errors, especially in the splicing of repetitive sequences and complex regions.

[0005] The existing reconstruction methods based on the reference genome have more or less defects. Most of the reconstruction methods only consider single nucleotide polymorphisms (SNPs), ignore insertion / deletion INDELs, and lose a lot of available reconstruction information. In addition, all methods do not consider the linkage information of multiple SNPs or INDELs on the same sequencing read segment; some reconstruction tools require the number of haplotypes to be reconstructed to be specified in advance; and some reconstruction tools are mainly aimed at viruses with high sequence heterogeneity, such as HIV, and have poor universality. Summary of the invention

[0006] The present invention proposes a virus co-infection detection method based on penalized regression and sequencing physical linkage, which is used to solve or at least partially solve the technical problem of low accuracy of reconstructed haplotypes in the methods in the prior art.

[0007] Another technical problem to be solved by the present invention is to provide a virus co-infection detection system based on penalized regression and sequencing physical linkage.

[0008] In order to solve the above technical problems, the first aspect of the present invention provides a virus co-infection detection method based on penalized regression and sequencing physical linkage, comprising:

[0009] S1: Collect virus sequences and their corresponding lineage information;

[0010] S2: Statistical analysis of single nucleotide polymorphism (SNP) and insertion / deletion INDEL frequencies of lineages worldwide based on viral sequences;

[0011] S3: Screening of high-frequency variants in the pedigree as representative variants of the pedigree;

[0012] S4: Use representative variants of the lineage to construct a site variation map of the viral lineage as a reference map for the reconstruction model;

[0013] S5: Create an index for the reference genome, and use BWA to align the second-generation sequencing virus sample to be reconstructed with the reference genome to obtain the SAM file of the sample;

[0014] S6: Use Samtools to convert the aligned sequencing samples into binary BAM files;

[0015] S7: Use the best analysis route of the GATK genome analysis toolkit to detect mutations in the aligned viral sequencing samples to obtain single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information;

[0016] S8: Construct a regression equation based on the SNP and INDEL of the sample and their frequencies and the reference map; use the square error loss function combined with L0 regularization and L1 regularization for regression, and finally obtain the pedigree haplotype and its frequency corresponding to the sample.

[0017] In one embodiment, in step S2, single nucleotide polymorphism SNPs and insertion / deletion INDEL frequencies of the lineage are counted based on the viral sequences, including: grouping the collected viral sequences by lineage, and comparing them with the reference genome to obtain single nucleotide polymorphism SNPs and insertion / deletion INDELs, and then dividing the number of single nucleotide polymorphism SNPs and insertion / deletion INDELs at each site by the number of viral sequences in the lineage to obtain the single nucleotide polymorphism SNPs and insertion / deletion INDEL frequencies for screening high-frequency variations in the lineage.

[0018] In one embodiment, when screening high-frequency variations in the pedigree as representative variations of the pedigree in step S3, 50% of the highest variation frequency of the pedigree is selected as the threshold for distinguishing whether it is a representative variation of the pedigree, and single nucleotide polymorphisms SNPs and insertion / deletion INDELs with frequencies greater than or equal to the threshold are retained as representative variations of the pedigree for constructing a reference map.

[0019] In one embodiment, when the representative variation of the lineage is used to construct the site variation map of the virus lineage as the reference map for reconstructing the model in step S4, all the lineages of the virus are used as the row index of the reference map; the representative variation of all the lineages of the virus are integrated and sorted according to the position relative to the reference genome as the column index of the reference map. In the reference map, for each lineage row, an ordered vector with the same dimension as the length of the column index is generated, and the value of each dimension in the vector is 0 or 1, 1 represents that the variation corresponding to the column index of the dimension is the representative variation of the lineage, and 0 represents that the variation corresponding to the column index of the dimension is not the representative variation of the lineage.

[0020] In one embodiment, step S7 includes:

[0021] Add sequencing read group information to BAM files;

[0022] Sort the sequencing reads in the BAM file in ascending order according to the position aligned to the reference genome;

[0023] Split the sequencing reads containing the character "N". In this process, all the characters "N" in the sequence reads are identified, and n+1 new sequencing reads are created (n is the number of characters "N" in the original sequencing reads);

[0024] Use Mutect2 to detect single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs and their frequency information to generate preliminary VCF files;

[0025] Use BaseRecalibrator to generate a recalibration table for base quality calibration based on the sequencing read groups, variant quality scores, machine cycles, and nucleotide context in the VCF file;

[0026] Recalibrate the base quality of the sequencing reads according to the recalibration table, and output the recalibrated BAM file;

[0027] Create an index for the BAM file, and use Mutect2 again to detect single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information to obtain the final VCF file;

[0028] In one embodiment, step S8 constructs a regression equation based on the SNPs and INDELs of the sample and their frequencies and the reference map, including:

[0029] Extract the frequency of each representative variation in the VCF file. Y1 represents the vector of length p consisting of the frequencies of the representative variations in the VCF file. Y1 = (freq1, freq2, ..., freq p ), Y2 represents a vector of length q consisting of representative mutations in all reference maps that do not appear in the VCF file, and all element values ​​are 0, Y2 = (0, 0, ..., 0); the sum of p and q is equal to the length of the reference map sequence, which is the number of representative mutations of all lineages of the virus.

[0030] Traverse all sequencing reads in the SAM file, and take the frequencies of two representative variants in the same sequencing read (or paired-end sequencing read) as the linkage variant frequency. Y3 represents the vector of linkage variant frequencies, Y3 = (Freq1, Freq2, ..., Freq t ).at last,

[0031] Based on the reference map, the above linkage variation is also used as the column of the map. For each pedigree row, if the two representative variations corresponding to the linkage variation in the column are both representative variations of the pedigree, the value of the column is 1, otherwise it is 0. The matrix based on the reference map X = (X1, X2, ..., X m ), m is equal to the length of the reference map row sequence, that is, the number of all lineages of the virus. m The dimension of X is equal to the dimension of Y. m In , 1 means that the variation corresponding to the column index of this dimension is the representative variation of this pedigree, and 0 means that the variation corresponding to the column index of this dimension is not the representative variation of this pedigree. Different weights can be set for single point variation and linkage variation.

[0032] Finally, the regression equation is:

[0033] Y~βX,

[0034] Use the square error loss function combined with L0 regularization and L1 regularization for regression, the loss function is:

[0035]

[0036] Among them, λ is the L0 regularization coefficient, and γ is the L1 regularization coefficient, both of which are adjustable. Finally, the β vector obtained by regression is the frequency of the corresponding spectrum.

[0037] The second aspect of the present invention provides a virus co-infection detection system based on penalized regression and sequencing physical linkage, which is used to perform the following operations:

[0038] Collect virus sequences and their corresponding lineage information;

[0039] The frequency of single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs of lineages worldwide was calculated based on viral sequences;

[0040] Screening high-frequency variants of the pedigree as representative variants of the pedigree;

[0041] Use representative variants of the lineage to construct a site variation map of the viral lineage as a reference map for the reconstruction model;

[0042] Create an index for the reference genome, and use BWA to align the second-generation sequencing virus sample to be reconstructed with the reference genome to obtain the SAM file of the sample;

[0043] Use Samtools to convert the aligned sequencing samples into binary format BAM files;

[0044] The best analysis route of the GATK genome analysis toolkit was used to detect mutations in the aligned viral sequencing samples to obtain single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information;

[0045] The regression equation was constructed by taking the sample's SNPs and INDELs and their frequencies as dependent variables and the reference map as independent variables;

[0046] Referring to steps S5-S7, the SNP and INDEL of the sample to be tested and their frequency information are collected, and the regression equation constructed in S8 is input to perform regularized regression on the equation to obtain the pedigree and frequency information of the sample to be tested.

[0047] Beneficial effects of the present invention

[0048] The present invention provides a virus co-infection detection method and system based on penalized regression and sequencing physical linkage. On the one hand, by using single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs and frequencies, combined with the linkage variation frequencies of the same sequencing read (or double-end sequencing read), the accuracy of haplotype sequence reconstruction of the virus population can be effectively improved; on the other hand, by adding L0 regularization and L1 regularization penalty terms to the loss function, overfitting can be avoided, and the method for reconstructing the haplotype sequence of the virus population can control the number of reconstructed haplotypes, thereby effectively reducing the false positive rate of the reconstruction result. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the prior art of the embodiments of the present invention, the drawings involved in the embodiments or the prior art are briefly introduced below. Obviously, the drawings in the following description show some embodiments of the present invention; for ordinary technicians in this field, other possible drawings can still be derived based on these drawings without creative work.

[0050] Figure 1 A schematic diagram of constructing a reference map using representative variations of a pedigree provided in an embodiment of the present invention.

[0051] Figure 2 Schematic diagram of the reference map structure.

[0052] Figure 3 Schematic diagram of the physical linkage structure of sequencing.

[0053] Figure 4 A schematic diagram of the variation detection process based on the optimal analysis route of the GATK genome analysis toolkit provided in an embodiment of the present invention.

[0054] Figure 5 A schematic flow chart of a method for detecting viral co-infection based on penalized regression and sequencing physical linkage provided in an embodiment of the present invention.

[0055] Figure 6 The figure is a comparison of the effects of the embodiment of the present invention and other existing reconstruction methods in a simulated data test.

[0056] Figure 7 The figure shows a comparison of the effects of the embodiments of the present invention in real data testing of SARS-CoV-2 virus with other existing reconstruction methods. DETAILED DESCRIPTION

[0057] The inventors of the present application have discovered through extensive research and practice that the current haplotype reconstruction methods often have low reconstruction accuracy when faced with highly similar virus strains within a virus population, and result in a high false positive rate. In addition, the calculation process often consumes a large amount of computing resources and time. Therefore, an accurate, efficient and universally applicable haplotype reconstruction algorithm is urgently needed.

[0058] Therefore, in order to solve the problems of low reconstruction accuracy, high computational and time costs, and lack of general applicability of existing methods, the present invention proposes a virus co-infection detection method and system based on penalized regression and sequencing physical linkage, which improves reconstruction accuracy and is applicable to situations where the similarity of virus strains within a population is high by incorporating single-point mutations and linkage mutations of the same sequencing read (or double-end sequencing read) into the model at the same time. Penalized regression combined with L0 regularization and L1 regularization is used to control the false positive rate of the reconstruction results. Coordinate descent and local combinatorial search are used to reduce computational and time costs.

[0059] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0060] Embodiment 1

[0061] The embodiment of the present invention provides a method for detecting viral co-infection based on penalized regression and sequencing physical linkage, comprising:

[0062] S1: Collect virus sequences and their corresponding lineage information;

[0063] S2: Statistical analysis of single nucleotide polymorphism (SNP) and insertion / deletion INDEL frequencies of lineages worldwide based on viral sequences;

[0064] S3: Screening of high-frequency variants in the pedigree as representative variants of the pedigree;

[0065] S4: Use representative variants of the lineage to construct a site variation map of the viral lineage. Use representative variants of the lineage to construct a site variation map of the viral lineage as a reference map for the reconstruction model.

[0066] See also Figure 1 , which is a schematic diagram of constructing a reference map using representative variations of a pedigree provided in an embodiment of the present invention.

[0067] S5: Create an index for the reference genome, and use BWA to align the second-generation sequencing virus sample to be reconstructed with the reference genome to obtain the SAM file of the sample;

[0068] S6: Use Samtools to convert the aligned sequencing samples into binary BAM files;

[0069] S7: Use the best analysis route of the GATK genome analysis toolkit to detect mutations in the aligned viral sequencing samples to obtain single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information;

[0070] See also Figure 4 , is a schematic diagram of the variation detection process based on the optimal analysis route of the GATK genome analysis toolkit provided in an embodiment of the present invention.

[0071] S8: The SNPs and INDELs of the sample and their frequencies were used as dependent variables and the reference map was used as the independent variable to construct a regression equation;

[0072] S9: Collect the SNP and INDEL of the sample to be tested and their frequency information according to steps S5-S7, input the regression equation constructed in S8, perform regularized regression on the equation, and obtain the pedigree and frequency information of the sample to be tested.

[0073] See also Figure 5 , is a schematic diagram of a process flow of a virus co-infection detection method based on penalized regression and sequencing physical linkage provided in an embodiment of the present invention.

[0074] In one embodiment, in step S2, single nucleotide polymorphism SNPs and insertion / deletion INDEL frequencies of the lineage are counted based on the viral sequences, including: grouping the collected viral sequences by lineage, and comparing them with the reference genome to obtain single nucleotide polymorphism SNPs and insertion / deletion INDELs, and then dividing the number of single nucleotide polymorphism SNPs and insertion / deletion INDELs at each site by the number of viral sequences in the lineage to obtain the single nucleotide polymorphism SNPs and insertion / deletion INDEL frequencies for screening high-frequency variations in the lineage.

[0075] In one embodiment, when screening high-frequency variations in the pedigree as representative variations of the pedigree in step S3, 50% of the highest variation frequency of the pedigree is selected as the threshold for distinguishing whether it is a representative variation of the pedigree, and single nucleotide polymorphisms SNPs and insertion / deletion INDELs with frequencies greater than or equal to the threshold are retained as representative variations of the pedigree for constructing a reference map.

[0076] In one embodiment, when the representative variation of the lineage is used to construct the site variation map of the virus lineage as the reference map for reconstructing the model in step S4, all the lineages of the virus are used as the row index of the reference map; the representative variation of all the lineages of the virus are integrated and sorted according to the position relative to the reference genome as the column index of the reference map. In the reference map, for each lineage row, an ordered vector with the same dimension as the length of the column index is generated, and the value of each dimension in the vector is 0 or 1, 1 represents that the variation corresponding to the column index of the dimension is the representative variation of the lineage, and 0 represents that the variation corresponding to the column index of the dimension is not the representative variation of the lineage.

[0077] See also Figure 2 , which is a schematic diagram of the reference atlas structure.

[0078] In one embodiment, step S7 includes:

[0079] Add sequencing read group information to BAM files;

[0080] Sort the sequencing reads in the BAM file in ascending order according to the position aligned to the reference genome;

[0081] Split the sequencing reads containing the character "N". In this process, all the characters "N" in the sequence reads are identified, and n+1 new sequencing reads are created (n is the number of characters "N" in the original sequencing reads);

[0082] Use Mutect2 to detect single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs and their frequency information to generate preliminary VCF files;

[0083] Use BaseRecalibrator to generate a recalibration table for base quality calibration based on the sequencing read groups, variant quality scores, machine cycles, and nucleotide context in the VCF file;

[0084] Recalibrate the base quality of the sequencing reads according to the recalibration table, and output the recalibrated BAM file;

[0085] Create an index for the BAM file, and use Mutect2 again to detect single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information to obtain the final VCF file.

[0086] In one embodiment, step S8 constructs a regression equation based on the SNPs and INDELs of the sample and their frequencies and the reference map, including:

[0087] Extract the frequency of each representative variation in the VCF file. Y1 represents the vector of length p consisting of the frequencies of the representative variations in the VCF file. Y1 = (freq1, freq2, ..., freq p ) T , Y2 represents a vector of length q consisting of representative variants in all reference maps that do not appear in the VCF file, and all element values ​​are 0, Y2 = (0, 0, ..., 0) T ; The sum of p and q is equal to the length of the reference map sequence, which is the number of representative mutations of all lineages of the virus.

[0088] Traverse all sequencing reads in the SAM file, and take the frequencies of two representative variants in the same sequencing read (or paired-end sequencing read) as the linkage variant frequency. Y3 represents the vector of linkage variant frequencies, Y3 = (Freq1, Freq2, ..., Freq t ) T .at last,

[0089] Please refer to Figure 3 , which is a schematic diagram of the physical linkage structure of sequencing.

[0090] Based on the reference map, the above linkage variation is also used as the column of the map. For each pedigree row, if the two representative variations corresponding to the linkage variation in the column are both representative variations of the pedigree, the value of the column is 1, otherwise it is 0. The matrix based on the reference map X = (X1, X2, ..., X m ), m is equal to the length of the reference map row sequence, that is, the number of all lineages of the virus. m The dimension of X is equal to the dimension of Y. m In , 1 means that the variation corresponding to the column index of this dimension is the representative variation of this pedigree, and 0 means that the variation corresponding to the column index of this dimension is not the representative variation of this pedigree. Different weights can be set for single point variation and linkage variation.

[0091] Finally, the regression equation is:

[0092] Y~βX,

[0093] Use the square error loss function combined with L0 regularization and L1 regularization for regression, the loss function is:

[0094]

[0095] Among them, λ is the L0 regularization coefficient, and γ is the L1 regularization coefficient, both of which are adjustable. Finally, the β vector obtained by regression is the frequency of the corresponding spectrum.

[0096] Please refer to Figure 6 and Figure 7 , Figure 6 The effect of the embodiment of the present invention in the simulation data test is compared with other existing reconstruction methods. Figure 7 The results of the embodiments of the present invention in the real data test of SARS-CoV-2 virus are compared with other existing reconstruction methods. Figure 6 and Figure 7 All indicate that the system of the present invention is significantly superior to the existing reconstruction system in terms of the similarity between the reconstructed haplotype and the gold standard haplotype, the false positive rate of the reconstructed haplotype, the difference in abundance between the reconstructed haplotype and the gold standard haplotype in the sample, and the number of reconstructed haplotypes.

[0097] The embodiment of the present invention provides a virus co-infection detection system based on penalized regression and sequencing physical linkage, which is used to perform the following operations:

[0098] Collect virus sequences and their corresponding lineage information;

[0099] The frequency of single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs of lineages worldwide was calculated based on viral sequences;

[0100] Screening high-frequency variants of the pedigree as representative variants of the pedigree;

[0101] Use representative variants of the lineage to construct a site variation map of the viral lineage as a reference map for the reconstruction model;

[0102] Create an index for the reference genome, and use BWA to align the second-generation sequencing virus sample to be reconstructed with the reference genome to obtain the SAM file of the sample.

[0103] Samtools was used to convert the aligned sequencing samples into binary BAM files.

[0104] The optimal analysis route of the GATK genome analysis toolkit was used to perform variation detection on the aligned viral sequencing samples to obtain single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information.

[0105] The regression equation was constructed by taking the sample's SNPs and INDELs and their frequencies as dependent variables and the reference map as independent variables;

[0106] Referring to steps S5-S7, the SNP and INDEL of the sample to be tested and their frequency information are collected, and the regression equation constructed in S8 is input to perform regularized regression on the equation to obtain the pedigree and frequency information of the sample to be tested.

[0107] Since the system introduced in the second embodiment of the present invention is a system used to implement a virus co-infection detection method based on penalized regression and sequencing physical linkage in the embodiment of the present invention, a person skilled in the art can understand the specific architecture and possible variations of the system in combination with the method in the first embodiment. Therefore, any software used in the first embodiment of the present invention belongs to a part of the protection scope of the present invention, and the specific content will not be repeated.

[0108] The above embodiments are only for illustrating the technical solutions of the present invention and are not intended to limit the present invention; although the present invention has been described in detail through the above embodiments, it should be understood by those skilled in the art that the technical solutions in these embodiments can still be modified or some of the technical features can be replaced by equivalents. These modifications or replacements will not cause the corresponding technical solutions to deviate from the core concept and protection scope of the embodiments of the present invention.

Claims

1. A method for detecting viral co-infection based on penalized regression and sequencing physical linkage, characterized in that: include: S1: Collect virus sequences and their corresponding lineage information; S2: Statistical analysis of single nucleotide polymorphism (SNP) and insertion / deletion INDEL frequencies of lineages worldwide based on viral sequences; S3: Screening of high-frequency variants in the pedigree as representative variants of the pedigree; S4: Use representative variants of the lineage to construct a site variation map of the viral lineage as a reference map for the reconstruction model; S5: Create an index for the reference genome, and compare the second-generation sequencing virus sample to be reconstructed with the reference genome to obtain the SAM file of the sample; S6: converting the aligned SAM file into a BAM file in binary format; S7: Perform mutation detection on the BAM file to obtain single nucleotide polymorphism (SNP), insertion / deletion INDEL and their frequency information; S8: The SNPs and INDELs of the sample and their frequencies were used as dependent variables and the reference map was used as the independent variable to construct a regression equation; S9: Collect the SNP and INDEL of the sample to be tested and their frequency information according to steps S5-S7, input the regression equation constructed in S8, perform regularized regression on the equation, and obtain the viral lineage and frequency information of the sample to be tested.

2. The method for detecting viral co-infection according to claim 1, characterized in that: In step S2, the single nucleotide polymorphism SNP and insertion / deletion INDEL frequencies of the lineage are counted according to the viral sequences, including: grouping the collected viral sequences by lineage, and comparing them with the reference genome to obtain single nucleotide polymorphism SNP and insertion / deletion INDEL, and then dividing the number of single nucleotide polymorphism SNP and insertion / deletion INDEL at each site by the number of viral sequences of the lineage to obtain the single nucleotide polymorphism SNP and insertion / deletion INDEL frequencies for screening high-frequency variations in the lineage.

3. The method for detecting viral co-infection according to claim 1, characterized in that: When screening high-frequency variants of the pedigree as representative variants of the pedigree in step S3, 50% of the highest variant frequency of the pedigree is selected as the threshold for distinguishing whether it is a representative variant of the pedigree, and single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs with frequencies greater than or equal to the threshold are retained as representative variants of the pedigree for constructing a reference map.

4. The method for detecting viral co-infection according to claim 1, characterized in that: In step S4, when the representative variation of the lineage is used to construct the virus lineage site variation map as the reference map for reconstructing the model, all the virus lineages are used as the row index of the reference map; the representative variation of all the virus lineages are integrated and sorted according to the position relative to the reference genome as the column index of the reference map; in the reference map, for each lineage row, an ordered vector with the same dimension as the column index length is generated, and the value of each dimension in the vector is 0 or 1, 1 represents that the variation corresponding to the column index of the dimension is the representative variation of the lineage, and 0 represents that the variation corresponding to the column index of the dimension is not the representative variation of the lineage.

5. The method for detecting viral co-infection according to claim 1, characterized in that: Step S7 includes: Add sequencing read group information to BAM files; Sort the sequencing reads in the BAM file in ascending order according to the position aligned to the reference genome; Split the sequencing reads containing the character "N"; wherein, all the characters "N" in the sequence reads are identified, and n+1 new sequencing reads are created, where n is the number of characters "N" in the original sequencing reads; Use Mutect2 to detect single nucleotide polymorphisms (SNPs) and insertion / deletion INDELs and their frequency information to generate preliminary VCF files; Use BaseRecalibrator to generate a recalibration table for base quality calibration based on the sequencing read groups, variant quality scores, machine cycles, and nucleotide context in the VCF file; Recalibrate the base quality of the sequencing reads according to the recalibration table, and output the recalibrated BAM file; The BAM file is indexed, and Mutect2 is used again to detect single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information to obtain a final VCF file, which contains single nucleotide polymorphisms (SNPs), insertion / deletion INDELs, and their frequency information.

6. The method for detecting viral co-infection according to claim 5, characterized in that: Step S8 constructs a regression equation based on the SNPs and INDELs of the sample and their frequencies and the reference map: Y~βX Use the square error loss function combined with L0 regularization and L1 regularization for regression, the loss function is: Among them, l is the square error loss function, X is the matrix composed of representative variations of each pedigree, Y is the single nucleotide polymorphism SNP and insertion / deletion INDEL and its frequency vector, λ is the L0 regularization coefficient, γ is the L1 regularization coefficient, both of which are adjustable. Finally, the β vector obtained by regression is the frequency of the corresponding pedigree; Specifically: Y1=(freq1,freq2,…,freq p ), represents a vector of length p consisting of the frequencies of representative variants in the VCF file, and freq represents the frequency of representative variants corresponding to the dimension; obtained by extracting the frequency of each representative variant in the VCF file; Y2 = (0, 0, ..., 0), which means a vector of length q consisting of representative variants in all reference maps that do not appear in the VCF file, with all element values ​​being 0; The sum of p and q is equal to the column sequence length of the reference map, which is the number of representative mutations of all lineages of the virus; Y3=(Freq1,Freq2,…,Freq t ), represents the vector of linked mutation frequencies, Y3 = (Freq1, Freq2,…, Freq t ) represents the linkage variation frequency corresponding to the dimension; by traversing all sequencing reads in the SAM file, the frequencies of two representative variations in the same sequencing read or paired-end sequencing read are obtained as the linkage variation frequency; The matrix X based on the reference map is (X1, X2, ..., X m ), m is equal to the length of the reference map row sequence, that is, the number of all lineages of the virus; X m The dimension of X is equal to the dimension of Y. m In the equation, 1 represents that the variation corresponding to the column index of this dimension is the representative variation of the pedigree, and 0 represents that the variation corresponding to the column index of this dimension is not the representative variation of the pedigree; different weights are set for single-point variation and linkage variation.

7. A virus co-infection detection system based on penalized regression and sequencing physical linkage, characterized in that The system comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs or instructions; The processor is used to obtain and execute a computer program or instruction in a memory so that the system executes the method according to any one of claims 1 to 6.

Images