Porcine SNP chip construction method based on gene regulatory network characteristics and application

Abstract

The invention discloses a pig SNP chip construction method based on gene regulatory network characteristics, and relates to the technical field of animal genetic breeding, and the pig SNP chip construction method comprises the following steps: constructing a pig tissue specificity multi-level gene regulatory network related to target traits and tissue types; performing function annotation and network comprehensive feature extraction on the whole genome SNP based on the gene regulation network; and carrying out score sorting and screening on the extracted network comprehensive characteristics, screening SNP variation sites with regulation function potential, constructing a pig SNP site set, and preparing the SNP chip. By introducing tissue / character related gene regulatory network information, functional priority ranking and screening are performed on SNP loci in a whole genome range, so that the character interpretation ability and breeding value estimation accuracy of SNP in a chip are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of animal genetic breeding technology, in particular to a pig SNP chip construction method based on gene regulatory network characteristics and application. BACKGROUND

[0002] In the research and molecular breeding practice of complex economic traits of pigs, how to accurately identify genetic variations with biological functions, especially functional SNPs in non-coding regions, is a key problem to achieve efficient genetic improvement. A large number of studies have shown that most of the trait-related variations discovered by whole-genome association studies or genome sequencing are not located in coding regions, but in regulatory regions or intergenic regions, which makes it difficult to explain their functional effects.

[0003] Non-coding variations do not change protein sequences, and affect phenotypes by regulating gene expression and disturbing the interaction between intergenic and non-coding RNAs. A method that relies solely on sequence annotation or chromatin state marker information cannot comprehensively analyze the mechanism of non-coding variations in complex biological systems. The occurrence of complex traits often involves the coordinated action of multiple genes and cross-pathway regulatory networks. The regulatory effect of variations usually depends on their nesting position in the regulatory network and the network propagation path, such as affecting the expression of key regulatory factors or disturbing the interaction between regulatory modules. Therefore, as an important model for revealing the functional dependence between genes, the gene regulatory network has become an important tool for the study of non-coding functional variations.

[0004] Traditional functional SNP prediction methods are mostly based on chromatin openness, transcription factor binding, and other epigenetic features for inference. Although they have reference value in specific cells, they lack overall modeling of the regulatory network structure between the whole tissue and multiple tissues, and are difficult to explain the synergistic or antagonistic effects of SNP variations on genes and signal pathways. Especially in agricultural animals such as pigs, due to the lack of complete gene regulatory network data analysis of trait phenotypes, annotating SNP variations based on chromatin features often has limitations. SUMMARY

[0005] To solve the above technical problems, the present application provides the following technical solutions: In a first aspect, the present application provides a pig SNP chip construction method based on gene regulatory network characteristics, comprising: constructing a pig tissue-specific multi-level gene regulatory network related to target traits and tissue types; functionally annotating whole-genome SNPs based on the gene regulatory network and extracting network comprehensive features; The extracted network comprehensive features are scored and screened, SNP variation sites with potential regulatory function are screened, and a pig SNP site set is constructed to prepare a SNP chip.

[0006] In a possible implementation, the constructing a pig tissue-specific multi-level gene regulatory network related to a target trait and a tissue type comprises: According to the breeding requirements, a target shape category is determined, and a main tissue or cell type regulating the trait is determined based on the physiological regulation mechanism of the selected trait. For subsequent network construction and feature extraction, integrate multi-source omics data; According to the target trait and its biological basis, an organization or cell type closely related to the trait is selected, and multi-source omics data is integrated to construct a multi-level gene regulatory network.

[0007] In a possible implementation, the gene regulatory network comprises: a protein-protein interaction network describing the functional connection between proteins; a transcription factor regulatory network revealing the regulatory relationship between transcription factors and target genes, constructed based on Motif prediction or ChIP-seq results; an miRNA-mRNA regulatory network established based on miRNA targeting prediction database and expression data inference; an eQTL co-regulation network constructed based on the regulatory mode of SNPs on multiple gene expressions; a WGCNA co-expression network establishing the synergistic expression structure between gene modules through weighted correlation analysis; and a functional module network forming the regulatory logic between modules through GO and KEGG biological databases.

[0008] In a possible implementation, the functional annotation of the whole genome SNP based on the gene regulatory network and the network comprehensive feature extraction comprise: The constructed gene regulatory network is used as a structure background; The whole genome SNP is functionally annotated and mapped to the gene regulatory network; The topological structure attribute and expression regulation feature of each SNP-associated gene in the gene regulatory network are extracted.

[0009] In a possible implementation, the expression regulation feature comprises: network topological structure feature, sequence and functional annotation feature, and expression and pathway function feature.

[0010] In a possible implementation, the network topology features include: node degree, which represents the number of direct connections of the SNP-affected gene in the regulatory network, and the higher the degree, the more genes the gene can regulate or be affected by, and the wider the network influence range; intermediate centrality, which measures whether the gene is a transfer hub between multiple regulatory paths; page rank value, which is based on the hierarchy of the whole network, considers the number of connections and the weight of adjacent nodes, and is an important indicator for comprehensively evaluating the network propagation ability and status of the gene; closeness centrality, which represents the average path length reciprocal of the gene to all other genes; and module score, which divides the gene regulatory network into functional modules and evaluates whether the SNP target gene belongs to a module enriched in a target trait-related pathway.

[0011] In a possible implementation, the sequence and functional annotation features include: a conservation score, which evaluates the conservation degree of the region where the SNP is located in multiple species; a TF binding site disruption score, which predicts whether the SNP will destroy the DNA binding motif of a key transcription factor, and judges whether the SNP destroys an important transcription factor binding site by comparing the score difference of the base sequence before and after the SNP in the transcription factor recognition motif; enhancer activity, which judges whether the SNP falls into an actively verified enhancer region and evaluates whether the SNP falls into an enhancer region with H3K27ac+ / ATAC+; eQTL evidence, whether there is regulatory evidence related to expression quantitative traits; and distance from TSS, the distance between the SNP and the TSS of the potential target gene, and the closer the distance, the more likely it is a functional regulatory site.

[0012] In a possible implementation, the expression and pathway function features include: expression difference fold, which comes from the difference in expression between individuals with extreme phenotypes in the target tissue; significant differential expression state, which judges whether the gene regulated by the SNP reaches the difference standard in the target tissue; and pathway enrichment weight, whether the SNP-regulated gene is enriched in the KEGG / GO pathway related to the target traits of growth, reproduction and immunity.

[0013] In a possible implementation, the scoring and screening of the extracted network comprehensive features screen SNP variation sites with regulatory function potential, and a set of pig SNP sites is constructed, including: determining the function score of each SNP and scoring and ranking the whole genome SNPs; screening high-score and high-ranking SNPs to form a final chip site pool.

[0014] In a possible implementation, the preparation of the SNP chip includes: based on the sequencing effective read number statistical information of the probe target region, a deep learning method is used to establish a probe capture efficiency prediction model; The model input is composed of a nucleotide sequence feature set formed by extending the probe sequence of a preset length upstream and downstream, and the model parameters are optimized using ten-fold cross-validation; According to the prediction result, a probe is synthesized around the selected functional SNP site; The synthesized probe is mixed with pig genomic DNA in liquid form, and a hybridization reaction is performed under suitable temperature and salt ion conditions. After the probe forms a stable double-stranded structure with its complementary target region, streptavidin-coated magnetic beads are added to bind the biotin-labeled probe to the magnetic beads, thereby enriching the target region. Subsequently, non-specific binding products are removed by magnetic separation. The enriched product is eluted and purified, library amplification and high-throughput sequencing technology are used to obtain sequence information of the target region, and the genotype of the target SNP site is accurately analyzed according to the sequencing result, thereby completing the chip detection process.

[0015] In the embodiments of the present application, by introducing the information of the tissue / character-related gene regulatory network, the functional priority of the SNP sites in the whole genome is sorted and screened, thereby significantly improving the trait explanation ability and breeding value estimation accuracy of the SNP in the chip. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a pig SNP chip construction method based on gene regulatory network features provided in the embodiments of the present application is shown. Figure 2 A technical development roadmap provided in the embodiments of the present application is shown. Figure 3 A core SNP detection rate diagram provided in the embodiments of the present application is shown. Figure 4 A typing consistency diagram provided in the embodiments of the present application is shown. DETAILED DESCRIPTION

[0017] The present application will be described below in conjunction with the drawings and specific embodiments.

[0018] Referring to Figure 1 and Figure 2 , the pig SNP chip construction method based on gene regulatory network features provided in the embodiments of the present application includes: S101, constructing a pig tissue-specific multi-level gene regulatory network related to a target trait and a tissue type.

[0019] In this example, the selection of pig whole genome variation sites is carried out by obtaining pig whole genome variation information from public database resources, downloading SNP information from NCBI and ENSEMBL databases, introducing resequencing data of 52 core pig breeds in IAnimal database, and performing quality control on all variation sites. The SNP with high deletion rate, insufficient sequencing depth or low sample frequency (MAF < 0.05) is excluded, and finally 53,210,651 high-quality whole genome SNP sites are obtained.

[0020] Target traits and tissue definition, data resource preparation, provide basic information and data support for regulatory network construction and functional site prediction, which need to be analyzed comprehensively in combination with the economic trait characteristics of pigs and the characteristics of regulatory organization.

[0021] Trait definition and key organization determination. According to the actual breeding demand or research direction, the target trait category is determined, including the following types: reproductive traits: litter weight, total litter size, birth weight, etc.; growth traits: daily gain, weaning weight, feed conversion rate; carcass composition: backfat thickness, eye muscle area, intramuscular fat content, etc.; key regulatory organization determination. Based on the physiological regulation mechanism of the selected traits and the existing literature, the main organization or cell type that regulates the traits is determined, and the organization with transcriptome proteome data is preferred, such as muscle tissue (growth and carcass traits); reproductive system tissue (testis, ovary, etc., control reproductive traits); immune tissue (intestine, lung, spleen, etc., related to disease resistance traits).

[0022] Omics data collection. To provide basic data for subsequent network construction and feature extraction, multiple types of omics data need to be integrated, including but not limited to: whole genome SNP data: derived from public databases or self-resequencing, covering different breeds and trait populations; GWAS research results: significant signal regions and guide SNPs related to target traits; transcriptome data (RNA-seq): from different phenotypes (such as extreme traits) of target tissues, used to construct co-expression network, perform differential analysis and regulatory score; proteome data (Proteomics), proteome data of target tissues of extreme phenotype individuals. eQTL and co-expression information: assist in inferring the correlation between SNP and expression variation.

[0023] According to the target traits and their biological basis, the tissues or cell types closely related to the traits are selected, and multi-omics data is integrated to construct multi-level regulatory networks. The networks include but are not limited to: protein-protein interaction network (PPI Network), transcription factor-target gene regulatory network (TF-Gene Regulatory Network), miRNA-mRNA regulatory network, gene co-expression network (such as WGCNA module), eQTL co-regulation network, and pathway function module network. Network construction combines existing STRING database with RNA and protein expression profiles, variation information and pathway enrichment results of target tissues, and further introduces transcriptome and proteome data of trait extreme individuals to weight network nodes, in order to enhance the organization specificity and trait correlation of the network.

[0024] The protein-protein interaction network collects protein interaction information of differentially expressed genes in the target tissue. The data sources can include STRING, BioGRID, IntAct and other databases. By screening high-confidence (such as combined score>0.7) interaction pairs, a functional synergistic protein network is constructed to reflect the physical interaction of gene products and the synergistic structure of the pathway.

[0025] The transcription factor-target gene regulatory network uses transcription factor binding site prediction (such as JASPAR database) and literature annotation to establish the positive and negative regulation network of transcription factors on target genes. RNA-seq expression patterns and regularization regression methods can be combined to construct tissue-specific TF-Gene regulatory connections and capture the regulatory driving core.

[0026] The miRNA-mRNA regulatory network integrates miRNA-mRNA targeting information provided by miRBase, miRTarBase and other databases to construct the post-transcriptional regulation network mediated by pig miRNA. Further, co-expression / inverse expression pairs can be screened based on tissue expression data (miRNA-seq and mRNA-seq) to improve the credibility of the regulation prediction.

[0027] The gene co-expression network is based on RNA-seq data of multiple samples in the target tissue, and uses WGCNA and other methods to construct a gene co-expression module network. Each module represents a functionally related group of genes with coordinated expression, and the network nodes are genes and the edges are Pearson correlation weighted connections. At the same time, the module-trait correlation coefficient can be calculated to mark significant modules.

[0028] eQTL co-regulation network: eQTL co-regulation network incorporates expression quantitative trait loci (eQTL) information from QTLdb, GTEx or self-constructed, to construct SNP-gene regulatory relationship network. If multiple SNPs regulate the same gene or multiple genes are regulated by the same eQTL, a co-regulation unit is formed. eQTL network can be used to infer the long-range regulatory effect of SNPs.

[0029] Pathway functional module network: According to the metabolic / signal module defined by KEGG, GO, and biological pathway database, the known trait-related genes are projected into the functional pathway atlas to extract modular network units. It is used to assist in evaluating whether the SNP-regulated genes are clustered in key pathways, and to realize the integration of functional annotation and pathway focusing.

[0030] S102, functional annotation and network comprehensive feature extraction of whole genome SNP based on the gene regulatory network.

[0031] Map the candidate SNPs in the whole genome to the constructed regulatory network, infer the potential regulatory path through their relationship with functional genes, and extract the network and expression characteristics corresponding to each SNP. The characteristics include: Network topology features: This type of feature is based on mapping non-coding SNPs to their target genes and extracting the network properties of the target genes using the tissue-specific constructed gene regulatory network. The network is constructed based on the integration of enhancer-promoter interaction relationships and gene-gene functional interaction relationships, reflecting the real structure of multiple levels of regulation in cells.

[0032] The network topology features include: node degree: represents the number of direct connections of the gene in the regulatory network. The higher the degree, the more genes the gene may regulate or be affected by, and the wider the network influence range. Betweenness centrality: measures whether the gene is a transit hub between multiple regulatory pathways. Genes with high centrality often play a bottleneck control role in network signal transmission and are easily affected by regulatory disturbances. The higher the value, the stronger the control ability of the regulatory pathway. PageRank value: based on the hierarchy of the whole network, considering the number of connections and the weight of adjacent nodes, it is an important indicator to comprehensively evaluate the network propagation power and status of the gene. Closeness centrality: represents the average path length reciprocal of the gene to all other genes. This index reflects the potential of the node as a regulator of integrated information, affecting the speed and amplitude of the regulatory response. Module score: the gene regulatory network is divided into functional modules (such as the WGCNA method), and it is evaluated whether the SNP target gene belongs to the module enriched in the target trait-related pathway, such as the reproduction pathway (such as GnRH signaling), the growth pathway (such as mTOR), the immune pathway (such as NF-κB), etc. The module score can be obtained by enrichment statistical significance (such as Fisher's test P value) or intra-module density calculation.

[0033] Sequence structure and functional annotation features: This layer of features reflects the biological attributes of the DNA sequence where the SNP is located in terms of structure and function, including evolutionary conservation, regulatory element location, transcription factor binding disruption, etc.

[0034] Sequence structure and functional annotation features include: Conservation score: Using cross-species conservation indicators (such as phyloP, phastCons), the evolutionary stability of the sequence region where the SNP is located is evaluated. Sites with high conservation are more likely to have key biological functions. TF binding site disruption score: By comparing the scoring difference (ΔPWM) of the base sequence before and after the SNP in the transcription factor recognition motif, it is determined whether the SNP disrupts important transcription factor binding sites (such as GATA4, FOXL2, MYOD, etc.). Enhancer activity: Determine whether the SNP falls within an experimentally verified active enhancer region. Such regions usually have epigenetic modifications such as H3K27ac or H3K4me1, and there are ATAC-seq peaks. Enhancer activity can be represented by signal intensity integration or peak width weighting. eQTL evidence: Based on known or self-built expression quantitative trait locus analysis, it is determined whether the SNP is significantly associated with the expression level of the downstream gene. Preferably, eQTL evidence from the same tissue / cell type is considered. Distance from TSS: Represents the linear distance between the SNP and the nearest gene transcription start site. This indicator is used to capture the spatial proximity rules of cis-regulatory SNPs, with common thresholds including 5kb, 10kb, etc.

[0035] Expression and biological pathway features: This type of feature reflects the potential regulatory targets of the SNP in terms of tissue expression and their activity in biological processes, emphasizing the relevance to the target trait.

[0036] Expression and pathway function features include: Expression fold change: In the extreme performance population of the target trait, the SNP-regulated gene expression is analyzed for differences, resulting in a log2-transformed expression difference value as a quantitative measure of functional impact. Significant differential expression status: Determine whether the SNP target gene reaches a significant standard (such as FDR < 0.05) in the differential expression analysis, which is used to mark the possibility of functional regulation. Pathway enrichment weight: Statistics on whether SNP target genes are significantly enriched in certain functional pathways or GO terms. Methods such as hypergeometric test or GSEA can be used to evaluate their representation in reproduction, development, immunity, etc. pathways, and assign weight scores according to the degree of enrichment.

[0037] By joint analysis of the above three types of features and assigning weights to the scores, the functional effect size and importance of SNP sites in the whole genome can be predicted, and a batch of high-potential regulatory SNPs can be selected for subsequent chip design, as shown in Table 1. The scoring system is not only suitable for molecular breeding of pigs and other domestic animals, but also has the ability to be extended to other species and other complex traits. The above features are used together to describe the functional background and biological importance of each site in the gene regulatory network.

[0038] Table 1. Network feature extraction scoring criteria for whole genome SNPs S103, scoring and screening the extracted network comprehensive features, screening SNP variation sites with regulatory function potential, constructing a set of pig SNP sites, and preparing a SNP chip.

[0039] Based on the comprehensive features extracted in S102, scoring and screening are performed: the model outputs the functional score of each SNP, scores and ranks the whole genome SNPs, screens high-score and top-ranked SNPs, forms a final chip site pool, and obtains a chip composed of 75,019 variation sites. Table 2 gives some of the sites, and the chromosome distribution of 5,019 variation sites is shown in Table 3.

[0040] Table 2. Part of the SNP chip sites Table 3. Chromosome distribution To construct a liquid-phase pig SNP chip with high capture efficiency and site functionality, a capture efficiency optimization strategy based on sequence composition is introduced in the probe design and hybridization capture process. First, based on the sequencing effective read number statistics of the probe targeting region, a probe capture efficiency prediction model is established using a deep learning method. The model input consists of a nucleotide sequence feature set formed by extending the 120 bp long probe sequence 5 bp, 10 bp, 15 bp and 20 bp upstream and downstream, and the model parameters are optimized using 10-fold cross-validation. The results show that when the probe sequence is extended 15 bp upstream and downstream, the prediction performance is optimal, which can effectively guide the subsequent probe screening.

[0041] According to the prediction results, around the selected 75,019 functional SNP sites, the 120 bp nucleotide sequence of 60 bp before and after the site is extracted for probe synthesis. All probes are modified at the 5' end by biotin to facilitate capture after hybridization.

[0042] In the liquid-phase hybridization process, the modified probes are mixed with pig genomic DNA in liquid form, and the hybridization reaction is carried out under suitable temperature and salt ion conditions. After the probe forms a stable double-stranded structure with its complementary target region, streptavidin-coated magnetic beads are added to bind the biotin-labeled probe to the magnetic beads, achieving enrichment of the target region. Subsequently, non-specific binding products are removed by magnetic separation.

[0043] The enriched products are eluted and purified, and library amplification and high-throughput sequencing technology is used to obtain sequence information of the target region, and the genotype of the target SNP site is accurately analyzed according to the sequencing results, completing the chip detection process.

[0044] To verify the genotyping accuracy and platform compatibility of the functional site liquid-phase SNP chip in practical application, 16 samples were randomly selected from the target pig population for comparison and analysis. The sequencing data of each sample is 1.5 GB, and the same DNA sample is detected by parallel genotyping using the liquid-phase capture chip of the present application and the existing 25X data of second-generation resequencing, respectively.

[0045] After chip detection, the detection performance of the core SNP site is statistically analyzed. The results show that the detection rate (Call Rate) of the core SNP of the chip in all samples is higher than 0.993, showing good detection sensitivity and genotyping stability. In the genotype comparison with the solid-phase chip, the genotype consistency coefficient between the two platforms is more than 0.995, indicating that the genotyping results have high consistency and accuracy, as shown in Figure 3

[0046] Further consistency evaluation of the chip on the intersection region of the SNP site is carried out, as shown in Figure 4 ​As shown, the average concordance of the co-coverage SNP sites reached above 0.994, indicating that the liquid chip has excellent compatibility between the cross-platform typing results.

[0047] In summary, the functional site liquid chip of the present application meets the technical requirements of large-scale application in terms of genotyping accuracy, detection consistency and cross-platform compatibility, has the feasibility of replacing traditional solid-phase chips and second-generation resequencing in breeding practice, and provides reliable typing technical support for large-scale genome selection.

[0048] To verify the application effect of the pig 75K liquid chip based on functional sites constructed by the present application in genome selection, 567 Large White pigs and Landrace pigs raised in a pig farm were selected as the evaluation population. The genotypes of the individuals in the population were detected, and the phenotypic data of two important economic traits, average daily gain (ADG) and backfat thickness (BF), were simultaneously collected.

[0049] In terms of genotype detection, the following three SNP chip platforms were used for comparison and evaluation: the functional site liquid capture chip (75K) constructed by the present application; a commonly used commercial pig 50K chip (chip A); and a 75K site set randomly extracted from the whole genome SNP database (named “random site-70K”).

[0050] Among them, the chip design of the present application is based on the capture strategy of regulatory functional regions, which enriches the regions containing functional annotation SNPs through targeted capture probes; the random site-75K is used as a control to evaluate the performance of SNP sites without functional site enrichment characteristics in genome prediction.

[0051] Genome prediction analysis uses the ssGBLUP software to construct a single-trait linear mixed model and performs 50 five-fold cross-validation to simulate the genome breeding value prediction ability under the condition of missing phenotypes. The prediction accuracy is evaluated by the correlation coefficient between the breeding value and the true phenotype, and is used to compare the performance difference of different chip platforms in economic trait prediction, see Table 4.

[0052] Table 4 Reliability of evaluated breeding values for average daily gain and backfat thickness The liquid chip constructed based on functional annotation sites showed high genome prediction accuracy in two core economic traits. Especially in the backfat thickness trait, the prediction ability of the chip of the present application was higher than that of the random site set and the commercial chip, verifying the practical value of functional site design in improving the efficiency of trait prediction.

[0053] In the embodiments of the present application, "at least one" refers to one or more, and "multiple" refers to two or two or more. The "and / or" describes the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Wherein A, B can be singular or plural. The character " / " generally represents that the associated objects before and after are in an "or" relationship. "At least one of the following" and the like means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b and c can represent: a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, c can be single or multiple.

[0054] The above is only a specific embodiment of the present application, and any skilled person in the art can easily think of changes or replacements within the technical scope disclosed by the present application, which should be covered within the protection scope of the present application. The protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for constructing a pig SNP chip based on gene regulatory network features, characterized in that, The application relates to a method for constructing a pig tissue-specific multi-level gene regulation network related to target traits and tissue types. The method comprises the following steps: constructing a pig tissue-specific multi-level gene regulation network related to target traits and tissue types; functionally annotating whole genome SNPs based on the gene regulation network and extracting network comprehensive features; 2. The method of claim 1, wherein the method is based on a gene regulatory network feature of a pig SNP chip. ranking and screening the extracted network comprehensive features, screening SNP variation sites with regulation function potential, constructing a pig SNP site set, and preparing an SNP chip. The method for constructing a pig tissue-specific multi-level gene regulation network related to target traits and tissue types comprises the following steps: clearly defining a target shape category according to breeding requirements and determining main tissues or cell types for regulating the traits based on the physiological regulation mechanism of the selected traits; integrating multi-source omics data for subsequent network construction and feature extraction; 3. The method of claim 2, wherein the method is based on a gene regulatory network feature. selecting tissues or cell types closely related to the traits according to the target traits and biological bases, and integrating multi-source omics data to construct a multi-level gene regulation network.

4. The method of claim 1, wherein the method is based on a gene regulatory network feature. The gene regulation network comprises a protein-protein interaction network describing the functional connection between proteins, a transcription factor regulation network revealing the regulation relationship between transcription factors and target genes based on motif prediction or ChIP-seq results, an miRNA-mRNA regulation network established based on miRNA targeting prediction databases and expression data, an eQTL co-regulation network constructed based on the regulation mode of SNPs on multiple gene expressions, a WGCNA co-expression network establishing the synergistic expression structure between gene modules through weighted correlation analysis, and a functional module network forming the regulation logic between modules through GO and KEGG biological databases. The method for functionally annotating whole genome SNPs based on the gene regulation network and extracting network comprehensive features comprises the following steps: taking the constructed gene regulation network as a structure background; functionally annotating whole genome SNPs and mapping the SNPs to the gene regulation network; 5. The method of claim 4, wherein the method is based on a gene regulatory network feature. extracting the topological structure attributes and expression regulation features of each SNP-associated gene in the gene regulation network.

6. The method of claim 5, wherein the method is based on a gene regulatory network feature. The expression regulation features comprise network topological structure features, sequence and function annotation features, and expression and pathway function features. The network topological structure features comprise node degree, intermediate centrality, web page ranking value, proximity centrality and module score. The node degree represents the number of direct connections of a SNP-affected gene in the regulation network, and the higher the degree, the more genes that can be regulated or affected by the gene, and the gene has a wider network influence range. The intermediate centrality measures whether the gene is a transfer hub between multiple regulation paths. The web page ranking value is based on the hierarchy of the whole network, considers the number of connections and the weight of adjacent nodes, and is an important indicator for comprehensively evaluating the network propagation force and status of the gene. The proximity centrality represents the average path length reciprocal of the gene to all other genes. The module score divides the gene regulation network into functional modules, and evaluates whether the module to which the SNP target gene belongs is enriched in the target trait-related pathway.

7. The method of claim 5, wherein the method is characterized by, The sequence and functional annotation features include: conservation score: evaluating the conservation degree of the SNP region in multiple species; TF binding site disruption score: predicting whether the SNP will destroy the DNA binding motif of a key transcription factor, by comparing the scoring difference of the base sequence before and after the SNP in the transcription factor recognition motif, to determine whether the SNP destroys the binding site of an important transcription factor; enhancer activity: determining whether the SNP falls into an active enhancer region verified by experiment, and evaluating whether the SNP falls into an enhancer region with H3K27ac+ / ATAC+; eQTL evidence: whether there is evidence of quantitative trait expression regulation; distance from TSS: distance from the transcription start site, the distance between the SNP and its potential target gene TSS, the closer the more likely to be a functional regulatory site.

8. The method of claim 5, wherein the method is characterized by, The expression and pathway function features include: expression difference fold: difference in expression between individuals with extreme phenotypes in the target tissue; significant differential expression state: determining whether the SNP-regulated gene reaches the difference standard in the target tissue; pathway enrichment weight: whether the SNP-regulated gene is enriched in growth, reproduction, and immune-related KEGG / GO pathways of target traits.

9. The method of claim 1, wherein the method is based on a gene regulatory network feature of a pig SNP chip. The scoring and screening of the extracted network comprehensive features screen SNP variation sites with regulatory function potential, and construct a set of pig SNP sites, including: Determine the functional score of each SNP and score and rank the whole genome SNPs; Screen high-score and top-ranked SNPs to form a final chip site pool. 10.The method of constructing a pig SNP chip based on a gene regulatory network feature according to claim 9, wherein, The preparation of the SNP chip includes: Based on the sequencing effective read number statistics information of the probe target region, a probe capture efficiency prediction model is established using a deep learning method; The model input consists of nucleotide sequence features formed by extending the probe sequence of a predetermined length upstream and downstream, and the model parameters are optimized using ten-fold cross-validation; According to the prediction results, probes are synthesized around the selected functional SNP sites; Mix the synthesized probes with pig genomic DNA in liquid form, and perform hybridization reaction under suitable temperature and salt ion conditions. After the probe and its complementary target region form a stable double-stranded structure, add streptavidin-coated magnetic beads to bind the biotin-labeled probe to the magnetic beads, achieve the enrichment of the target region, and then remove the non-specific binding products by magnetic separation; The enrichment products are eluted and purified, the library is amplified, and high-throughput sequencing technology is used to obtain sequence information of the target region, and the genotype of the target SNP site is accurately analyzed according to the sequencing results, and the chip detection process is completed.

Images