Gene detection method and device and computer equipment
By constructing a selection signal model and a gamma-Poisson distribution model for whole-genome sequencing data, and combining them with the likelihood ratio test, we achieved accurate detection of coding and non-coding regions. This solved the problem that existing technologies cannot accurately detect driving elements in non-coding regions, and improved the accuracy and comprehensiveness of tumor somatic mutation detection.
Patent Information
- Application Number
- CN202410840619.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-26
- Publication Date
- 2024-11-15
AI Technical Summary
Existing technologies cannot accurately detect driving elements in non-coding regions and coding regions, especially in cancer gene detection. Traditional methods are sensitive to samples with high mutation rates and have limited applicability.
A gene detection method based on whole-genome functional influence scores was adopted. By constructing selection signal models, Poisson regression models, triplet mutation models, and generalized linear models from whole-genome sequencing data, and combining them with a gamma-Poisson distribution model for likelihood ratio testing, the accurate detection of target selection genes was achieved.
It enables the precise detection of selection signals and driver genes for mutations in both coding and non-coding regions, improving the detection capability of tumor somatic mutation heterogeneity and complex mutation maps, and outperforming existing methods in terms of accuracy, enrichment, and Jaccard similarity.
Smart Images

Figure CN118969074A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of gene detection, and in particular to a gene detection method, device and computer equipment. Background Art
[0002] Cancer is a complex disease caused by somatic mutations. Starting from a single ancestral cell, it undergoes multiple cell divisions and accumulation of genetic mutations, and eventually evolves into an infinitely proliferating tumor cell. In this evolutionary process, genetic variation and clonal selection are the two main driving forces. With the development of sequencing technology and somatic mutation detection tools, large-scale tumor somatic mutation data have made it possible to study clonal selection in tumor evolution, including the detection of selection signals and selection genes.
[0003] The coding region refers to the sequence in a gene that encodes proteins, also known as the coding sequence (CDS). The coding region is a part of the DNA of a gene, as opposed to the non-coding region. In eukaryotes, the coding region includes exons. Exons refer to regions in a gene that can encode nucleotide sequences. The non-coding region is a segment that cannot be transcribed into the corresponding messenger RNA and cannot guide protein synthesis (that is, it cannot encode proteins), including introns, promoters, 5' non-transcribed regions, 3' non-transcribed regions, variable splicing sites, and intergenic regions. The non-coding region is located before and after the coding region and controls the strength of gene expression. Early tumor research focused on exon sequencing data, and the detection of driver elements in non-coding regions was relatively small. The analytical tools used to detect non-coding regions were also limited, and mainly relied on background map variability models, that is, predicting the background map variability of somatic mutations in advance and then detecting them, such as DIGDriver, DriverPower, etc., but it is sensitive to samples with high mutation rates and can be used for limited tumors. However, non-coding elements as regulatory elements have been shown to play an important role in a large number of biological processes, and some non-coding elements have also been shown to be significantly associated with the occurrence and development of tumors, such as the promoter element of TERT.
[0004] Some genetic detection methods in related technologies are limited by the detection environment and detection tools and cannot accurately detect the driver elements in the non-coding regions and the coding regions. Summary of the invention
[0005] In view of this, the present invention provides a gene detection method, apparatus and computer equipment to solve the current problem of being unable to accurately detect driver elements in non-coding regions and coding regions.
[0006] According to the first aspect, this embodiment provides a gene detection method, the method comprising:
[0007] Based on the whole genome functional impact score, a selection signal model of genes in whole genome sequencing data is constructed; whole genome sequencing data includes coding and non-coding regions of the genome;
[0008] Based on the gene selection signal model, a Poisson regression model of whole genome sequencing data was constructed;
[0009] A triplet nucleotide mutation model was constructed based on a Poisson regression model of whole genome sequencing data;
[0010] Based on the triple nucleotide mutation model, a generalized linear model was constructed;
[0011] Based on the generalized linear model, a gamma Poisson distribution model of target selection genes was constructed;
[0012] Based on the gamma Poisson distribution model of target selection genes, a likelihood ratio test was performed to detect target selection genes.
[0013] In an optional embodiment, constructing a selection signal model of genes in whole genome sequencing based on the whole genome functional impact score includes:
[0014] Obtain the functional impact score of each gene mutation in whole genome sequencing data;
[0015] If the functional impact score value of each gene mutation is greater than or equal to the first preset threshold, the gene mutation belongs to the functional mutation group;
[0016] If the functional impact score of each gene mutation is less than the first preset threshold, the gene mutation belongs to the non-functional mutation group;
[0017] Obtain the number of functional mutations and the total number of functional sites in the functional mutation group observed during genome sequencing, as well as the number of non-functional mutations and the total number of non-functional mutation sites in the non-functional mutation group;
[0018] The selection signal ratio value of each gene was calculated based on the number of functional mutations and the total number of functional sites in the functional mutation group, the number of non-functional mutations and the total number of non-functional mutation sites in the non-functional mutation group;
[0019] Based on the selection signal ratio value of each gene, a selection signal model of data genes in whole genome sequencing is constructed.
[0020] In an optional embodiment, the selection signal ratio value of each gene is calculated by the following formula:
[0021]
[0022] ω is the selection signal ratio value of each gene, dN is the first ratio value of the number of functional mutations in the functional mutation group observed during genome sequencing to the total number of functional mutation sites, ds is the second ratio value of the number of non-functional mutations in the non-functional mutation group to the total number of non-functional mutation sites, is the normalized ratio between the first ratio value and the second ratio value, n is the number of functional mutations observed during genome sequencing, N is the number of functional mutations observed during genome sequencing, s is the number of non-functional mutations observed during genome sequencing, and S is the total number of non-functional mutation sites observed during genome sequencing.
[0023] In an optional embodiment, a Poisson regression model of whole genome sequencing data is constructed based on the gene selection signal model, which is performed by the following formula:
[0024] n C>T ~Poisson(λ=d1r C>T *ωN C>T ,N)s C>T ~Poisson(λ=d1r C>T S C>T ,S)
[0025] Where d1 is the density of mutations at each site in the non-functional mutation group, S is the total number of non-functional mutation sites observed during genome sequencing, and S C>T is the total number of C>T sites in the non-functional mutation group on the reference genome, r C>T is the mutation rate of C>T sites in non-functional mutation sites, s C>T is the number of C>T mutations in the non-functional mutation group in the actual observed data during gene sequencing, and λ is the expected value in the Poisson regression process;
[0026] Where N is the total number of functional mutation sites observed during genome sequencing, ω is the selection signal ratio value of each gene, and r C>T *ω is the mutation rate in the functional mutation group, n C>T It is the number of C>T mutations in the non-functional mutation group in the actual observed data during gene sequencing.
[0027] In an optional embodiment, a triplet nucleotide mutation model is constructed based on a Poisson regression model of whole genome sequencing data, and is calculated by the following formula:
[0028] s i ~Poisson(λ=d1 r i S i ,S)
[0029] n i ~Poisson(λ=tri *ωN i ,N)
[0030] Among them, S is the total number of non-functional mutation sites observed during genome sequencing, i is the type of 192 triplet nucleotide mutations, d1 is the density of mutations at each site in the non-functional mutation group, and r i The mutation rate of non-functional mutation sites containing 192 triplet nucleotide mutation types, s i is the number of mutations in the non-functional mutation group containing 192 triplet nucleotide mutation types in the actual observed data during gene sequencing, λ is the expected value in the Poisson regression process, and n i is the number of mutations in the functional mutation group in the actual observed data, which contains 192 triplet nucleotide mutation types, ω is the selection signal ratio value of each gene, t is the classification variable of functional mutation and non-functional mutation, S i is the total number of sites containing 192 triplet nucleotide mutation types in the non-functional mutation group on the reference genome, N i It is the total number of sites containing 192 triplet nucleotide mutation types in the functional mutation group on the reference genome.
[0031] In an optional embodiment, based on the triplet nucleotide mutation model, a generalized linear model is constructed, comprising:
[0032] The generalized linear equation is created by the following formula:
[0033] Log(μ i )=βX i ^Τ+ε
[0034] β a =[β t ,β1,...,β 191 ]
[0035] ω=exp(β t )
[0036] Log(μ i ) is the log link function used to relate the linear predictor to the Poisson parameter used in the statistical analysis, μ i is the expected probability that the mutation observed during gene sequencing falls into the functional mutation group or the non-functional mutation group, X is the number of 192 triplet nucleotide mutations observed during gene sequencing, T is the operation matrix in the generalized linear equation, ε is the residual value of the linear equation, i is the 192 triplet nucleotide mutation types, ω is the selection signal ratio value of each gene, β t is the effect size of the t variable, β a =[β1,...,β 191] is the effect size of 191 trinucleotide covariates;
[0037] Obtaining likelihood estimates of all parameters in the generalized linear equation using Poisson regression;
[0038] The confidence interval of the selection signal ratio value of each gene in the selection signal model of genes in whole genome sequencing was obtained using the maximum likelihood estimation method.
[0039] In an optional embodiment, based on a generalized linear model, a gamma Poisson distribution model of the target selection gene is constructed, comprising:
[0040] The gamma Poisson distribution equation is created by the following formula;
[0041] s i ~Poisson((λ))
[0042] λ~Gamma(α,β)
[0043] s i is the total number of sites containing 192 triplet nucleotide mutation types in the non-functional mutation group on the reference genome, λ is the expected value in the Poisson regression process, α is the shape parameter of the Gamma distribution, and β is the rate parameter;
[0044] The negative binomial regression model was used to estimate the number of non-functional mutations in each gene in whole-genome sequencing; based on the generalized linear model, the maximum likelihood estimate was calculated in Poisson regression to obtain the solution of the selection coefficient.
[0045] In an optional embodiment, based on the gamma Poisson distribution model of the target selection gene, a likelihood ratio test is performed on the target selection gene to detect the target selection gene, comprising:
[0046] The null hypothesis and alternative hypothesis are made by the following formula;
[0047] Null hypothesis (H0):ω j =1
[0048] Alternative hypothesis (H1):ω j ≠1
[0049] Among them, ω j For the jth gene, the test statistic of the likelihood ratio test follows a chi-square distribution with 1 degree of freedom, and the significance value associated with the test statistic is combined to assess whether the gene is selective;
[0050] Multiple weight correction was used to perform multiple test correction on each gene in whole genome sequencing;
[0051] If, after multiple testing correction, the significance value of any gene in whole genome sequencing is less than the second preset threshold, the gene is the target selection gene;
[0052] If, after correction for multiple tests, the significance value of any gene in whole genome sequencing is greater than or equal to the second preset threshold, the gene is not a target selection gene.
[0053] In an optional embodiment, when it is based on the target selection gene, the positive target selection gene or the negative target selection gene is identified by the following formula;
[0054]
[0055] Among them, ω j Genes with values greater than 1 are defined as positive target selection genes, whereas genes with values greater than 1 are defined as positive target selection genes. j Values less than 1 are defined as negative target selection genes, j is the subscript of the gene, i refers to 192 triplet nucleotides, n i,j refers to the number of functional mutations of the i-th triplet of the j-th gene, s i,j Refers to the number of non-functional mutations of the ith triplet of the jth gene.
[0056] According to the second aspect, this embodiment provides a gene detection device, the device comprising:
[0057] The first model building module is used to build a selection signal model of genes in whole genome sequencing data based on the whole genome functional impact score; the whole genome sequencing data includes coding regions and non-coding regions of the genome;
[0058] The second model building module is used to build a Poisson regression model for whole genome sequencing data based on the gene selection signal model;
[0059] The third model building module is used to build a triplet nucleotide mutation model based on a Poisson regression model of whole genome sequencing data;
[0060] The fourth model building module is used to build a generalized linear model based on the triple nucleotide mutation model;
[0061] The fifth model building module is used to build a gamma Poisson distribution model of target selection genes based on a generalized linear model;
[0062] The whole genome detection module is used to perform a likelihood ratio test on the target selection gene based on the gamma Poisson distribution model of the target selection gene to detect the target selection gene.
[0063] According to a third aspect, this embodiment provides a computer device, including:
[0064] The memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the gene detection method in the first aspect or any embodiment of the first aspect by executing the computer instructions.
[0065] According to the fourth aspect, this embodiment provides a computer-readable storage medium, on which computer instructions are stored, and the computer instructions are used to enable a computer to execute the gene detection method in the first aspect or any embodiment of the first aspect.
[0066] According to the fifth aspect, this embodiment provides a computer program product, including computer instructions, and the computer instructions are used to enable a computer to execute the gene detection method in the first aspect or any embodiment of the first aspect.
[0067] The technical solution of the present invention has the following advantages:
[0068] The present invention relates to the field of gene detection technology, and discloses a gene detection method, device and computer equipment, wherein the gene detection method can detect the heterogeneity of tumor somatic mutations and complex mutation maps, etc., expand the simple dN / dS algorithm to a generalized linear model (GLM-Poisson) based on Poisson distribution, and use mutation characteristics as covariates to correct the influence of non-selection factors, thereby realizing the detection of selection signals in the whole genome. After realizing the detection of selection signals, the joint analysis of the Gamma-Poisson distribution model and the likelihood ratio test is applied to develop the accurate detection of selection genes in the whole genome, and finally realize the detection of selection signals and accurate detection of driver genes for mutations in coding and non-coding regions. BRIEF DESCRIPTION OF THE DRAWINGS
[0069] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0070] Figure 1 is a schematic diagram of a process of a gene detection method according to an embodiment of the present invention;
[0071] Figure 2 is a specific schematic diagram of another gene detection method according to an embodiment of the present invention;
[0072] Figure 3 is a comparative schematic diagram of various gene detection methods according to an embodiment of the present invention;
[0073] Figure 4 is a structural block diagram of a gene detection device according to an embodiment of the present invention;
[0074] Figure 5 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION
[0075] 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 those skilled in the art without creative work are within the scope of protection of the present invention.
[0076] According to an embodiment of the present invention, an embodiment of a genetic detection method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.
[0077] In this embodiment, a gene detection method is provided, which can be used in mobile terminals, such as mobile phones, tablet computers, etc. Figure 1 is a flow chart of a gene detection method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:
[0078] Step S101, constructing a selection signal model of genes in whole genome sequencing data based on the whole genome functional impact score; the whole genome sequencing data includes coding regions and non-coding regions of the genome.
[0079] Since the existing technology has limited analytical tools for detecting non-coding regions, and the traditional gene detection method cannot accurately detect the driver elements and coding regions of non-coding regions. Therefore, the gene detection method in the embodiment of the present disclosure can realize the selection signal detection and driver gene detection of mutations in coding regions and non-coding regions in the following way.
[0080] In an optional embodiment, the above step S101 constructs a selection signal model of genes in whole genome sequencing based on the whole genome functional impact score; the whole genome sequencing includes the coding region and non-coding region of the genome, including:
[0081] Step a1, obtaining the functional impact score value of each gene mutation in the whole genome sequencing data.
[0082] Specifically, the genome-wide functional impact score can be used to quantitatively assess the selection pressure on genomic variants in non-coding regions. The functional impact score is a set of continuous scores, with each mutation having a unique score.
[0083] Step a2: If the functional impact score of each gene mutation is greater than or equal to a first preset threshold, the gene mutation belongs to the functional mutation group.
[0084] Step a3: If the functional impact score of each gene mutation is less than the first preset threshold, the gene mutation belongs to the non-functional mutation group.
[0085] Specifically, in order to divide the continuous variable into two groups of data with different selection pressures similar to synonymous and non-synonymous variations, the disclosed embodiment divides the continuous variable into two groups, becoming a binary variable, and the mutation with a lower score constitutes a non-functional mutation group. For example, the functional impact score of a mutation is greater than or equal to the first preset threshold (classification threshold), and the mutation with a higher score constitutes a functional mutation group, while the functional impact score of the mutation is less than the first preset threshold (classification threshold), and the mutation with a lower score constitutes a functional mutation group. The specific classification threshold will be compared and tested during the test process. Figure 2 , illustrating a dichotomous functional impact score.
[0086] Step a4, obtaining the number of functional mutations and the total number of functional sites in the functional mutation group observed during genome sequencing, as well as the number of non-functional mutations and the total number of non-functional mutation sites in the non-functional mutation group.
[0087] Step a5, calculating the selection signal ratio value of each gene based on the number of functional mutations and the total number of functional sites in the functional mutation group, the number of non-functional mutations and the total number of non-functional mutation sites in the non-functional mutation group.
[0088] Step a6, constructing a selection signal model of genes in the whole genome sequencing data based on the selection signal ratio value of each gene.
[0089] In the disclosed embodiment, the above step S101 is equivalent to constructing a dN / dS model in whole genome sequencing. The dN / dS value in the whole genome represents the ratio of the number of functional mutations to the total number of functional sites (dN) and the normalized ratio between the number of non-functional mutations and the total number of non-functional mutation sites (dS), and ω is used to represent the selection signal ratio value (dN / dS).
[0090] The traditional dN / dS ratio, also known as the Ka / Ks ratio, is a natural selection metric that compares the number of non-synonymous variants at non-synonymous loci to the ratio of the synonymous variant rates at synonymous loci. In order to extend this framework to the non-coding genome, the genome-wide functional impact score is combined to quantitatively assess the selection pressure on genomic variants in non-coding regions. Of course, the disclosed embodiments are also applicable to coding genomes.
[0091] In a specific embodiment, in the above step a5, the selection signal ratio value of each gene is calculated by the following formula (1):
[0092]
[0093] ω is the selection signal ratio value of each gene, dN is the first ratio value of the number of functional mutations in the functional mutation group observed during genome sequencing to the total number of functional mutation sites, ds is the second ratio value of the number of non-functional mutations in the non-functional mutation group to the total number of non-functional mutation sites, is the normalized ratio between the first ratio value and the second ratio value, n is the number of functional mutations observed during genome sequencing, N is the number of functional mutations observed during genome sequencing, s is the number of non-functional mutations observed during genome sequencing, and S is the total number of non-functional mutation sites observed during genome sequencing.
[0094] Step S102, constructing a Poisson regression model for whole genome sequencing based on the gene selection signal model.
[0095] Specifically, according to the research suggestions on low-density mutations in cancer genome data, the present disclosure embodiment adopts the Poisson regression model to analyze somatic mutations. In constructing the Poisson regression model of whole genome sequencing, taking the number of C>T mutations in the non-functional mutation group as an example, its Poisson distribution process is modeled by the following formulas (2)-(3).
[0096] In a specific embodiment, based on the gene selection signal model, a Poisson regression model of whole genome sequencing data is constructed, which is performed by the following formulas (2)-(3):
[0097] s C>T ~Poisson(λ=d1r C>T S C>T ,S) (2)
[0098] n C>T ~Poisson(λ=d1r C>T *ωN C>T ,N) (3)
[0099] Where d1 is the density of mutations at each site in the non-functional mutation group, S is the total number of non-functional mutation sites observed during genome sequencing, and S C>T is the total number of C>T sites in the non-functional mutation group on the reference genome, r C>T is the mutation rate of C>T sites in non-functional mutation sites, s C>T is the number of C>T mutations in the non-functional mutation group in the actual observed data during gene sequencing, and λ is the expected value in the Poisson regression process;
[0100] Where N is the total number of functional mutation sites observed during genome sequencing, ω is the selection signal ratio value of each gene, and r C>T *ω is the mutation rate in the functional mutation group, n C>T It is the number of C>T mutations in the non-functional mutation group in the actual observed data during gene sequencing.
[0101] Step S103, constructing a triplet nucleotide mutation model based on the Poisson regression model of the whole genome sequencing data.
[0102] It has been confirmed in the related COSMIC mutational signatures study that the trinucleotide mutation model can effectively capture the contextual dependence of somatic mutations. Relevant researchers have also shown that the trinucleotide model has sufficient ability to capture the impact of background mutation rate and maintains strong consistency with the pentanucleotide substitution model in various cancer types. In order to consider the contextual impact of mutations, the disclosed embodiment uses 192 trinucleotide mutation types in the Poisson regression model. These 192 trinucleotide types include the base of the mutation site plus one base each upstream and downstream (6*4*4), while also considering the impact of transcription and non-transcription chains (*2), forming 192 (6*4*4*2) trinucleotide combinations. Therefore, its mutation is expressed by the following formula. As Figure 2 Shown are schematic diagrams of genomic triplets.
[0103] In an optional embodiment, a triplet nucleotide mutation model is constructed based on a Poisson regression model of whole genome sequencing data, and is calculated by the following formulas (4)-(5):
[0104] s i ~Poisson(λ=d1 r i S i ,S) (4)
[0105] n i ~Poisson(λ=tr i *ωN i ,N) (5)
[0106] Among them, S is the total number of non-functional mutation sites observed during genome sequencing, i is the 192 types of triplet nucleotide mutations, such as AAA>ATA, d1 is the density of mutations at each site in the non-functional mutation group, and r i The mutation rate of non-functional mutation sites containing 192 triplet nucleotide mutation types, s i is the number of mutations in the non-functional mutation group containing 192 triplet nucleotide mutation types in the actual observed data during gene sequencing, λ is the expected value in the Poisson regression process, and n i is the number of mutations in the functional mutation group in the actual observed data, which contains 192 triplet nucleotide mutation types, ω is the selection signal ratio value of each gene, t is the classification variable of functional mutation and non-functional mutation, S i is the total number of sites containing 192 triplet nucleotide mutation types in the non-functional mutation group on the reference genome, N i It is the total number of sites containing 192 triplet nucleotide mutation types in the functional mutation group on the reference genome.
[0107] Step S104, constructing a generalized linear model based on the triple nucleotide mutation model.
[0108] Specifically, in order to obtain the specific numerical value of the selection signal ratio value of each gene, the embodiment of the present disclosure adopts a general linear model (GLM), and the specific equation is expressed by the following formulas (6)-(8).
[0109] In a specific embodiment, based on the triple nucleotide mutation model, a generalized linear model is constructed, comprising:
[0110] Step b1, create a generalized linear equation using the following formula:
[0111] Log(μ i )=βX i ^Τ+ε (6)
[0112] β a =[β t ,β1,…,β 191 ] (7)
[0113] ω=exp(β t ) (8)
[0114] Log(μ i ) is the log link function used to relate the linear predictor to the Poisson parameter used in the statistical analysis, μ iis the expected probability that the mutation observed during gene sequencing falls into the functional mutation group or the non-functional mutation group, X is the number of 192 triplet nucleotide mutations observed during gene sequencing, T is the operation matrix in the generalized linear equation, ε is the residual value of the linear equation (the difference between the actual observed value and the fitted value), which is used to measure the degree of fit and prediction accuracy of the equation, i is the type of 192 triplet nucleotide mutations, ω is the selection signal ratio value of each gene, β t is the effect size of the t variable, β a =[β1,…,β 191 ] are the effect sizes of the 191 trinucleotide covariates.
[0115] Specifically, in the above generalized linear equation, the logarithmic link function is used to link the linear predictor with the Poisson parameter (μ i ) is associated with the variable μ. i represents the expected probability that a mutation observed in gene sequencing falls into the functional or non-functional mutation group. The variable X represents the number of 192 triplet nucleotide mutations observed. The categorical variable t is used to distinguish functional mutations from non-functional mutations (represented as 1 or 0 in the generalized linear model). The parameter i represents one of the 192 trinucleotide types, such as AAA>ATA. In this GML model, β t The exponent exp(β t ) corresponds to the selectivity ω in the dN / dS framework.
[0116] Step b2, using Poisson regression to obtain likelihood estimates of all parameters in the generalized linear equation;
[0117] Step b3, using the maximum likelihood estimation method, obtain the confidence interval of the selection signal ratio value of each gene in the selection signal ratio value model of whole genome sequencing.
[0118] Specifically, in order to consider the effects of different trinucleotides in the generalized linear model, a unit matrix is introduced into the covariate of the embodiment of the present disclosure to correct the differences in mutation rates between different trinucleotides. a =[β1,…,β 191 ] represents the effects of 191 trinucleotide covariates. Then, Poisson regression can be used to efficiently obtain likelihood estimates of all parameters in the generalized linear equation. By using the maximum likelihood estimation method, confidence intervals (CIs) for the parameter ω can be obtained.
[0119] Step S105, constructing a gamma Poisson distribution model of the target selection gene based on the generalized linear model.
[0120] Specifically, in the disclosed embodiment, the gamma-Poisson distribution model is a Gamma-Poisson model. In the disclosed embodiment, the selection signal model (dNdS-Fun framework) of each gene in whole genome sequencing is extended to the detection of selection elements or selection genes.
[0121] Further, in order to better explain the gamma Poisson distribution model, the embodiment of the present disclosure takes genes as an example to illustrate the gene detection method in the embodiment of the present disclosure. In actual application, genes can be divided into coding sequences, proximal regulatory sequences (such as promoters, splicing regions and untranslated regions), intron sequences, or replaced with intergenic region sequences. In the global selection signal detection, the embodiment of the present disclosure represents the mutation density of each site as the parameter d1. From the gene level, the parameter d1 is actually the average density of all genes, because the mutation rates between individual genes are not the same, but they all vary around the average density of d1. The embodiment of the present disclosure models the normalized mutation rate on the gene as a Gamma distribution, so the number of somatic mutations observed in gene sequencing is converted into a Gamma-Poisson distribution. The Poisson distribution characterizes the average level of observed mutations, while the Gamma distribution captures the randomness of the number of mutations occurring between genes. The number of functional mutations actually observed is affected by different gene densities d1 and selection ω. Therefore, in a given data set, the number of non-functional mutations observed for an individual gene j (s i ) can be modeled as a Poisson-Gamma process, which is specifically expressed by the following formulas (8)-(9).
[0122] In an optional embodiment, based on the triple nucleotide mutation model, a generalized linear model is constructed, comprising:
[0123] Step c1, create a gamma Poisson distribution equation using the following formula.
[0124] s i ~Poisson((λ)) (8)
[0125] λ~Gamma(α,β) (9)
[0126] s i is the total number of sites containing 192 triplet nucleotide mutation types in the non-functional mutation group on the reference genome, λ is the expected value in the Poisson regression process, α is the shape parameter of the Gamma distribution, and β is the rate parameter.
[0127] Step c2, using a negative binomial regression model to estimate the number of non-functional mutations of each gene in whole genome sequencing;
[0128] Step c3, based on the generalized linear model, in Poisson regression, obtain the solution of the selection coefficient by calculating the maximum likelihood estimate.
[0129] Wherein, α is the shape parameter of the Gamma distribution, and β is the rate parameter (the inverse of the scale). Under this Gamma Poisson distribution mode, the disclosed embodiment can regard the size of the gene, the gene sequence, and the influence of the variant model as offset terms to reflect the exposure degree of the gene. In the disclosed embodiment, a negative binomial regression model (negative binomial regression framework) is used to estimate the number of non-functional mutations of each gene, which is a composite distribution of Gamma-Poisson. During operation, the α and β parameters can be obtained by the following formulas (10)-(11):
[0130] α j =θ j (10)
[0131] β j =θ j / μ j (11)
[0132] Among them, μ j is the number of non-functional mutations of gene j predicted by the negative binomial regression model, θ j The following R-based code is used to find α and β:
[0133] model=glm.nb(obs_s~offset(log(exp_s))+covariate_matrix)
[0134] Where obs_s is the number of non-functional mutations observed during gene sequencing, exp_s is the expected number of non-functional variants in gene j, covariate_matrix is a unit matrix containing the covariate effects of 192 triplet nucleotides, and glm.nb is a negative binomial distribution function in R. obs_s~offset(log(exp_s)) represents the observed number of mutations that follow the negative binomial distribution function of the mutation site in the reference genome. Then, in Poisson regression, by solving d j The maximum likelihood estimate of is used to obtain the solution for the selection coefficient. Therefore, d j The likelihood function of the maximum likelihood estimate is now limited by the global mutation rate in the type of mutations between genes and is also affected by the number of non-functional mutations in the gene, as shown in the following formula (12):
[0135] L(d j )=L Poisson (d j |s j )L Gamma(d j |α,β) (12)
[0136] Under the constraints of Poisson and Gamma, d j The analytical solution of the maximum likelihood estimation can be obtained by solving the likelihood function L(d j ) About d j The derivative of (dL / dd j ) is equal to 0. Poisson represents the likelihood function value under Poisson regression. Gamma represents the likelihood function value under the gamma distribution. In the disclosed embodiment, the effect of the Gamma-Poisson model is to detect the selection signal of the gene. Among them, the Gamma distribution is used to correct the differences in length and mutation rate between different genes.
[0137] Step S106, based on the gamma Poisson distribution model of the target selection gene, a likelihood ratio test is performed on the target selection gene to detect the target selection gene.
[0138] Specifically, in order to infer whether there is a selection signal in gene j, the embodiment of the present disclosure uses a likelihood ratio test to compare the goodness of fit of the selection model and the no-selection model. In the embodiment of the present disclosure, by inferring whether there is a selection signal in gene j, it is possible to know whether gene j is selected in the tumorigenesis process, that is, whether it is related to the tumorigenesis process, and the purpose is to detect genes related to tumorigenesis.
[0139] In an optional embodiment, the above step S106, based on the gamma Poisson distribution model of the target selection gene, performs a likelihood ratio test on the target selection gene to detect the target selection gene, including:
[0140] Step c1, make null hypothesis and alternative hypothesis using the following formula.
[0141] Null hypothesis (H0):ω j =1 (13)
[0142] Alternative hypothesis (H1):ω j ≠1 (14)
[0143] Among them, ω j For the jth gene, the test statistic of the likelihood ratio test follows a chi-square distribution with 1 degree of freedom, and the significance value associated with the test statistic is used to assess whether the gene is selective.
[0144] Specifically, after performing these tests, the present disclosure embodiment uses the Benjamini-Hochberg method to perform multiple test correction on all evaluated genes, and the corrected significance value is less than 0.05, which is considered to be a selected gene.j represents the ratio of the number of functional mutations to the number of non-functional mutations observed in gene j.
[0145] Step c2, using a multiple weight correction method to perform multiple test correction on each gene in the whole genome sequencing.
[0146] Step c3: if the significance value of any gene in the whole genome sequencing is less than the second preset threshold after multiple test correction, the gene is the target selection gene.
[0147] Specifically, the second preset threshold is 0.05 mentioned above.
[0148] Step c4: if, after multiple testing correction, the significance value of any gene in the whole genome sequencing is greater than or equal to the second preset threshold, the gene is not a target selection gene.
[0149] Specifically, after performing these tests, all evaluated genes were corrected for multiple tests using the Benjamini-Hochberg method in the disclosed embodiments, and genes were considered to be target selection genes when the corrected significance value was less than 0.05. j It represents the ratio of the number of functional mutations to the number of non-functional mutations observed in gene j, which can be expressed by the following formula (15):
[0150]
[0151] Among them, ω j Genes with values greater than 1 are defined as positive target selection genes, whereas genes with values greater than 1 are defined as positive target selection genes. j Values less than 1 are defined as negative target selection genes, j is the subscript of the gene, i refers to 192 triplet nucleotides, n i,j refers to the number of functional mutations of the i-th triplet of the j-th gene, s i,j Refers to the number of non-functional mutations of the ith triplet of the jth gene.
[0152] Specifically, in Figure 2 , is a schematic diagram of the gene detection method in the embodiment of the present disclosure, wherein a represents the construction of a generalized linear model (construction of a GLM-Poisson model) based on a triplet nucleotide mutation model, and the construction of a generalized linear model includes: 1) genome-wide triplet nucleotide extraction; 2) using mutation function impact scores to divide the mutations in the whole genome into functional mutation groups and non-functional mutation groups; 3) calculating the selection signal (dN / dS) value between the functional mutation group and the non-functional mutation group. b represents the biological process behind the selection signal calculated based on the selection signal model (dNdS-Fun) of genes in whole genome sequencing.
[0153] The disclosed embodiment provides a gene detection method that can detect problems such as the heterogeneity of tumor somatic mutations and complex mutation maps, expands the simple dN / dS algorithm to a generalized linear model (GLM-Poisson) based on Poisson distribution, and uses mutation features as covariates to correct the influence of non-selective factors, thereby realizing genome-wide selection signal detection. After realizing the selection signal detection, the joint analysis of the Gamma-Poisson distribution model and the likelihood ratio test is applied to develop the detection of selection genes in the whole genome, and finally can realize the detection of selection signals and driver genes for mutations in coding and non-coding regions.
[0154] like Figure 3 As shown, the disclosed embodiment is based on the gene detection method of dNdS-Fun and the gene detection method of three other whole genome non-coding driver elements (in Figure 3 In this disclosure, 14,886 tumor samples were sequenced using three different gene detection methods (DIGDriver, DriverPower, and LARVA) for comparison, covering a total of 31 tumor subtypes. This disclosure uses COSMIC's CGC genes as the gold standard tumor driver genes. Figure 3 In the present study, the accuracy, enrichment, and (Jaccard) similarity of four different gene detection methods, dNdS-Fun, DIGDriver, DriverPower, and LARVA, in non-coding and coding regions were calculated.
[0155] exist Figure 3 It can be clearly seen that when all tissues and overlapping tissues of the cancer are counted, the gene detection method based on dNdS-Fun is superior to the gene detection methods of DIGDriver, DriverPower, and LARVA in terms of accuracy, enrichment, and Jaccard similarity.
[0156] In summary, the gene detection method in this embodiment, based on the development of the dNdS-Fun method, can improve the detection effect of coding and non-coding drivers in tumor whole genome sequencing data, and achieve the improvement of whole genome driver gene detection.
[0157] In this embodiment, a gene detection device is also provided, which is used to implement the above-mentioned embodiments and preferred embodiments, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0158] This embodiment provides a gene detection device, such as Figure 4 As shown, including:
[0159] A first model building module 41 is used to build a selection signal model of genes in whole genome sequencing data based on the whole genome functional impact score; the whole genome sequencing data includes coding regions and non-coding regions of the genome;
[0160] A second model building module 42 is used to build a Poisson regression model of whole genome sequencing data based on the gene selection signal model;
[0161] The third model building module 43 is used to build a triplet nucleotide mutation model based on a Poisson regression model of whole genome sequencing data;
[0162] A fourth model building module 44 is used to build a generalized linear model based on the triple nucleotide mutation model;
[0163] A fifth model building module 45 is used to build a gamma Poisson distribution model of the target selection gene based on a generalized linear model;
[0164] The whole genome detection module 46 is used to perform a likelihood ratio test on the target selection gene based on the gamma Poisson distribution model of the target selection gene to detect the target selection gene.
[0165] In an optional implementation, the first model building module 41 includes:
[0166] The first acquisition submodule is used to obtain the functional impact score value of each gene mutation in the whole genome sequencing data;
[0167] A first determination submodule, configured to determine that if the functional impact score value of each gene mutation is greater than or equal to a first preset threshold, the gene mutation belongs to a functional mutation group;
[0168] A second determination submodule is used for determining that if the functional impact score value of each gene mutation is less than a first preset threshold value, the gene mutation belongs to a non-functional mutation group;
[0169] The second acquisition submodule is used to obtain the number of functional mutations and the total number of functional sites of the functional mutation group observed during genome sequencing, as well as the number of non-functional mutations and the total number of non-functional mutation sites of the non-functional mutation group;
[0170] A calculation submodule, for calculating the selection signal ratio value of each gene based on the number of functional mutations and the total number of functional sites in the functional mutation group, the number of non-functional mutations and the total number of non-functional mutation sites in the non-functional mutation group;
[0171] The model building submodule is used to build a selection signal model of genes in whole genome sequencing data based on the selection signal ratio value of each gene.
[0172] In an optional embodiment, the selection signal ratio value of each gene is calculated using the above formula (1).
[0173] In an optional embodiment, the second model building module 42 is calculated using the above formulas (2)-(3).
[0174] In an optional implementation, the third model building module 43 is calculated using the above formulas (4)-(5).
[0175] In an optional implementation, the fourth model building module 44 includes:
[0176] A generalized linear equation creation submodule, used to create the above formulas (6)-(8);
[0177] A generalized linear equation calculation submodule, used for obtaining likelihood estimates of all parameters in the generalized linear equation using a Poisson regression method;
[0178] The gene selection signal confidence interval acquisition submodule is used to obtain the confidence interval of the selection signal ratio value of each gene in the selection signal ratio value model of whole genome sequencing.
[0179] In an optional implementation, the fifth model building module 45 includes:
[0180] A Gamma Poisson distribution equation creation submodule, used to create a Gamma Poisson distribution equation through the above formulas (8)-(9);
[0181] The submodule for calculating the number of non-functional mutations of genes is used to estimate the number of non-functional mutations of each gene in whole genome sequencing using a negative binomial regression model;
[0182] The selection coefficient solution submodule is used to obtain the solution of the selection coefficient by calculating the maximum likelihood estimation in Poisson regression based on the generalized linear model.
[0183] In an optional embodiment, the whole genome detection module 46 includes:
[0184] The test submodule is used to conduct null hypothesis and alternative hypothesis through the above formulas (13)-(14);
[0185] A correction submodule is used to perform multiple test corrections on each gene in whole genome sequencing using a multiple weight correction method;
[0186] A first determination submodule is used to determine that if the significance value of any gene in whole genome sequencing is less than a second preset threshold after multiple test correction, the gene is a target selection gene;
[0187] The second determination submodule is used to determine that if the significance value of any gene in the whole genome sequencing is greater than or equal to a second preset threshold after multiple test correction, the gene is not a target selection gene.
[0188] In an optional embodiment, the positive target selection gene or the negative target selection gene is identified by the above formula (15).
[0189] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.
[0190] The gene detection device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.
[0191] An embodiment of the present invention further provides a computer device having the above-mentioned gene detection device.
[0192] See also Figure 5 , Figure 5 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.
[0193] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.
[0194] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.
[0195] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0196] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.
[0197] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 5 The example of connecting through bus is taken in the following.
[0198] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.
[0199] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.
[0200] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.
[0201] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.
Claims
1. A gene detection method, characterized in that: The method comprises: Based on the whole genome functional impact score, constructing a selection signal model of genes in whole genome sequencing data; the whole genome sequencing data includes coding regions and non-coding regions of the genome; Based on the gene selection signal model, a Poisson regression model of whole genome sequencing data was constructed; constructing a triplet nucleotide mutation model based on a Poisson regression model of the whole genome sequencing data; Based on the triple nucleotide mutation model, a generalized linear model is constructed; Based on the generalized linear model, a gamma Poisson distribution model of the target selection gene is constructed; Based on the gamma Poisson distribution model of the target selection gene, a likelihood ratio test is performed on the target selection gene to detect the target selection gene.
2. The gene detection method according to claim 1, characterized in that: Based on the whole genome functional impact score, a selection signal model of genes in whole genome sequencing data is constructed, including: Obtain the functional impact score of each gene mutation in whole genome sequencing; If the functional impact score value of each gene mutation is greater than or equal to the first preset threshold, the gene mutation belongs to the functional mutation group; If the functional impact score of each gene mutation is less than the first preset threshold, the gene mutation belongs to the non-functional mutation group; Obtaining the number of functional mutations and the total number of functional sites of the functional mutation group observed during genome sequencing, as well as the number of non-functional mutations and the total number of non-functional mutation sites of the non-functional mutation group; Calculating the selection signal ratio value of each gene based on the number of functional mutations and the total number of functional sites in the functional mutation group, the number of non-functional mutations and the total number of non-functional mutation sites in the non-functional mutation group; Based on the selection signal ratio value of each gene, a selection signal model of genes in the whole genome sequencing data is constructed.
3. The gene detection method according to claim 2, characterized in that: The selection signal ratio value of each gene is calculated by the following formula: ω is the selection signal ratio value of each gene, dN is the first ratio value of the number of functional mutations in the functional mutation group observed during genome sequencing to the total number of functional mutation sites, ds is the second ratio value of the number of non-functional mutations in the non-functional mutation group to the total number of non-functional mutation sites, is the normalized ratio between the first ratio value and the second ratio value, n is the number of functional mutations observed during genome sequencing, N is the number of functional mutations observed during genome sequencing, s is the number of non-functional mutations observed during genome sequencing, and S is the total number of non-functional mutation sites observed during genome sequencing.
4. The gene detection method according to claim 1, characterized in that: Based on the gene selection signal model, a Poisson regression model of whole genome sequencing data is constructed and executed by the following formula: s C>T ~Poisson(λ=d1r C>T S C>T ,S) n C>T ~Poisson(λ=d1r C>T *ωN C>T ,N) Where d1 is the density of mutations at each site in the non-functional mutation group, S is the total number of non-functional mutation sites observed during genome sequencing, and S C>T is the total number of C>T sites in the non-functional mutation group on the reference genome, r C>T is the mutation rate of C>T sites in non-functional mutation sites, s C>T is the number of C>T mutations in the non-functional mutation group in the actual observed data during gene sequencing, and λ is the expected value in the Poisson regression process; Where N is the total number of functional mutation sites observed during genome sequencing, ω is the selection signal ratio value of each gene, and r C>T *ω is the mutation rate in the functional mutation group, n C>T It is the number of C>T mutations in the non-functional mutation group in the actual observed data during gene sequencing.
5. The gene detection method according to claim 1, characterized in that: Based on the Poisson regression model of the whole genome sequencing data, a triple nucleotide mutation model was constructed and calculated by the following formula: s i ~Poisson(λ=d1 r i S i ,S) n i ~Poisson(λ=tr i *ωN i ,N) Among them, S is the total number of non-functional mutation sites observed during genome sequencing, i is the type of 192 triplet nucleotide mutations, d1 is the density of mutations at each site in the non-functional mutation group, and r i The mutation rate of non-functional mutation sites containing 192 triplet nucleotide mutation types, s i is the number of mutations in the non-functional mutation group containing 192 triplet nucleotide mutation types in the actual observed data during gene sequencing, λ is the expected value in the Poisson regression process, and n i is the number of mutations in the functional mutation group in the actual observed data, which contains 192 triplet nucleotide mutation types, ω is the selection signal ratio value of each gene, t is the classification variable of functional mutation and non-functional mutation, S i is the total number of sites containing 192 triplet nucleotide mutation types in the non-functional mutation group on the reference genome, N i It is the total number of sites containing 192 triplet nucleotide mutation types in the functional mutation group on the reference genome.
6. The gene detection method according to claim 1, characterized in that: Based on the triple nucleotide mutation model, a generalized linear model is constructed, including: The generalized linear equation is created by the following formula: Log(μ i )=βX i ^T+e β a =[β t ,β1,…,β 191 ] ω=exp(β t ) Log(μ i ) is the log link function used to relate the linear predictor to the Poisson parameter used in the statistical analysis, μ i is the expected probability that the mutation observed during gene sequencing falls into the functional mutation group or the non-functional mutation group, X is the number of 192 triplet nucleotide mutations observed during gene sequencing, T is the operation matrix in the generalized linear equation, ε is the residual value of the linear equation, i is the 192 triplet nucleotide mutation types, ω is the selection signal ratio value of each gene, β t is the effect size of the t variable, β a =[β1,…,β 191 ] is the effect size of 191 trinucleotide covariates; Obtaining likelihood estimates of all parameters in the generalized linear equation using Poisson regression; The confidence interval of the selection signal ratio value of each gene in the selection signal model of genes in whole genome sequencing was obtained using the maximum likelihood estimation method.
7. The gene detection method according to claim 1, characterized in that: Based on the generalized linear model, a gamma Poisson distribution model of target selection genes is constructed, including: The gamma Poisson distribution equation is created by the following formula; s i ~Poisson((λ)) λ~Gamma(α,β) s i is the total number of sites containing 192 triplet nucleotide mutation types in the non-functional mutation group on the reference genome, λ is the expected value in the Poisson regression process, α is the shape parameter of the Gamma distribution, and β is the rate parameter; The negative binomial regression model was used to estimate the number of non-functional mutations of each gene in whole genome sequencing; based on the generalized linear model, the solution of the selection coefficient was obtained by calculating the maximum likelihood estimate in Poisson regression.
8. The gene detection method according to claim 1, characterized in that: Based on the gamma Poisson distribution model of the target selection gene, a likelihood ratio test is performed on the target selection gene to detect the target selection gene, comprising: The null hypothesis and alternative hypothesis are made by the following formula; Null hypothesis (H0):ω j =1 Alternative hypothesis (H1):ω j ≠1 Among them, ω j For the jth gene, the test statistic of the likelihood ratio test follows a chi-square distribution with 1 degree of freedom, and the significance value associated with the test statistic is combined to assess whether the gene is selective; Multiple weight correction was used to perform multiple test correction on each gene in whole genome sequencing; If, after multiple testing correction, the significance value of any gene in whole genome sequencing is less than the second preset threshold, the gene is the target selection gene; If, after correction for multiple tests, the significance value of any gene in whole genome sequencing is greater than or equal to the second preset threshold, the gene is not a target selection gene.
9. The gene detection method according to claim 8, characterized in that: When it is based on the target selection gene, the following formula is used to identify whether it is a positive target selection gene or a negative target selection gene; Among them, ω j Genes with values greater than 1 are defined as positive target selection genes, whereas genes with values greater than 1 are defined as positive target selection genes. j Values less than 1 are defined as negative target selection genes, j is the subscript of the gene, i refers to 192 triplet nucleotides, n i,j refers to the number of functional mutations of the i-th triplet of the j-th gene, s i,j Refers to the number of non-functional mutations of the ith triplet of the jth gene.
10. A gene detection device, characterized in that: The device comprises: A first model building module is used to build a selection signal model of genes in whole genome sequencing data based on the whole genome functional impact score; the whole genome sequencing data includes coding regions and non-coding regions of genes; The second model building module is used to build a Poisson regression model for whole genome sequencing data based on the gene selection signal model; A third model building module is used to build a triplet nucleotide mutation model based on the Poisson regression model of the whole genome sequencing data; A fourth model building module, used to build a generalized linear model based on the triple nucleotide mutation model; A fifth model building module is used to build a gamma Poisson distribution model of the target selection gene based on the generalized linear model; The whole genome detection module is used to perform a likelihood ratio test on the target selection gene based on the gamma Poisson distribution model of the target selection gene to detect the target selection gene.
11. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the gene detection method according to any one of claims 1 to 9 by executing the computer instructions.
12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the gene detection method according to any one of claims 1 to 9.
13. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the gene detection method according to any one of claims 1 to 9.
