Model Construction Of Gene Functional Pathway Analysis For Livestock

With the rapid development of high-throughput technologies,huge amounts of data are produced including gene list,microarray and so on.The statistical models used in the gene functional analysis are urgently needed.The pathway analysis has contributed to construct the gene function network and to identify the associations between gene and phenotype,and among genes,even to evaluate their importance.It is well known that the pathways can cross and overlap because each gene has multiple functions and can act in more than one pathway.In the last decade,about one hundred statistics models were emerging for the pathway analysis.However,almost all these models had one common limitation.These models assumed that each pathway was independent of other pathways.The assumption of independence could lead to the unreliability of results,even erroneous results.The GO and Kyoto Encyclopedia of Genes and Genomes(KEGG)are the frequently-used important public pathway databases in the gene function analysis.This study aims to construct the Path analysis model and the decision analysis model of KEGG pathways by taking account into the correlation among pathways.Meanwhile,copy number variation(CNV)in the chicken genome was studied in order to construct the pathway Topology-based model according to IPA analysis result.The main results were listed as follows:1.By Path analysis model,the correlation coefficient of each KEGG subcategory pathway to the corresponding KEGG category pathway,and each KEGG secondary pathway to the corresponding KEGG subcategory pathway were subdivided.The subdivided direct effect and indirect effect could quantitatively display the complex regulation mechanisms among the KEGG subcategories and the secondary pathways.The correlation coefficient could be used to identify the most significant pathways as the total effect.The gradient method from principal component analysis was introduced to estimate preliminarily the impact direction of each KEGG pathway.The validation of data from the functional analysis of bovine mammary transcriptome during lactation demonstrated that the Path analysis model based on KEGG pathway could produce more biologically meaningful results.As a result,KEGG-PATH was an effective data mining of pathway analysis.2.The decision coefficient was constructed to reflect the comprehensive decision-making ability of each KEGG pathway from the angle of regression ’variation’,according to the subdivision of coefficient of determination.The absolute value of the decision coefficient could evaluate the importance of the pathways.The sign of decision coefficient could estimate the impact direction of each KEGG pathway.The direct determination factor and indirect determination factor from the decision coefficient could show the complex regulating relationship among the pathways.The decision tree was plotted according to the decision percentage,which could display visually the decision results.The validation of data from the analysis of bovine mammary transcriptome during peak lactation demonstrated that the decision analysis model could effectively identify the most important function pathways of bovine mammary.The decision analysis model of KEGG pathway was a deep level pathway analysis model.3.Identification of copy number variation in the chicken genome was studied using high-density 50K single nucleotide polymorphism(SNP)array and PennCNV algorithm.638 CNVs were identified in 207 chickens,and merged to 190 copy number variation regions(CNVRs).These CNVRs cover 20.3 megabases(～1.69%)of the entire chicken genome.About 39%(74/190)of these CNVRs were identified at least in two chickens.Selected 3 high frequent and 11 low frequent CNVRs were further experimentally validated by quantitative PCR.Overall almost 93%(13/14)of the CNVR regions got positive confirmation in our experiment.The association analysis of CNV and phenotype showed that the association relationship existed between CNV and phenotype.Still further,212 Marek’s disease resistant(MDR)chicken-specific genes and 69 Marek’s disease susceptible(MDS)chicken-specific genes were chosen for IPA network analysis.The results revealed that MDR-specific genes are involved in Gastrointestinal Disease,Skeletal and Muscular Disorders.In short,the complex regulation mechanisms among pathways were quantified within the Path analysis model and the decision analysis model of KEGG pathway in this study.In this way,the interaction among genes could be mastered easily by researchers.In other words,this quantitative analysis could provide theory evidence for deeper gene functional analysis.So the models proposed by this study were the development of pathway analysis methodology.Identification of chicken CNV was also very important in that it was a significant complement to chicken CNV mapping.The association study between CNV and phenotype had enlightenment function for seeking possible genes of Marek’s disease.More importantly,the results of IPA analysis can be used to construct the Topology-based pathway analysis model.