Incorporating Functional Annotation To Develop Genomic Selection And Genome-wide Association Study Method

Genome-wide association study(GWAS)and genomic selection(GS)have become important tools to study the genetic basis of complex traits and predict the phenotype in human,animal and plant.With the development of sequencing technology,the genome of human,animal and plant are increasingly annotated.Incorporating the genomic annotation in GWAS and GS has the potential to speed the study on the genetic basis of complex traits and improve the predicting performance.Based on the linear mixed model(LMM),we proposed the strategies and models to incorporate genomic information in GWAS and GS,and developed two methods for genetic analysis: incorporate Prior biological information in Genomic Best Linear Unbiased Predictor(pGBLUP)and Scalable Multiple Annotation integration for trait-Relevant Tissue identification and usage(SMART).The main research results in this study are as follows:(1)The model and method for pGBLUP has been developed.By comparing with the traditional GS methods(GBLUP and Bayes R)in simulations,pGBLUP incorporating the annotation information showed the potential to improve the performance of genomic performance,which depended on the function of annotation information.(2)The model of pGBLUP could be extended to estimate the heritability enrichment,which were able to estimate the folds of enrichment of different functional annotation information accurately and measure the importance of annotation for the traits.(3)The application of pGBLUP on buffalo milk traits genomic selection,the cattle milk related QTL regions had medium and week heritability enrichments for buffalo milk traits.Owning to the limited buffalo sample size(412),the improvement of genomic selection on buffalo milk traits by incorporating cattle milk related QTL information was not significant.While pGBLUP was still an important strategy to incorporate prior biologically functional annotation to improve genomic selection performance in small population.(4)The application of pGBLUP on functional annotation enrichment analysis of piglet splay leg syndrome,identified 7 KEGG pathways had very large fold of enrichments(fe>100),including pathway KEGG_MTOR_SIGNALING_PATHWAY which controls the muscle developments and deserved the further research.(5)The model and method of SMART has been developed.By simulations and real data applications,SMART incorporating multiple annotations showed the high accuracy in inferring the trait-relevant tissues(cell types).What's more,the power of SNP-set test has been improved,thus detecting more significant loci in GWAS when weighted the SNP based on the annotation information of trait-relevant tissue(cell types)inferred by SMART approach.(6)The application of SMART on human traits and genomic functional annotation in public database: based on the relevance between traits and tissues(cell types),the 43 traits were clustered into 5 groups and the relevant relationships of pair-wise traits were also discovered.The histone marks H3K4me1 and H3K4me3 have large effect indicating their ability to predict the genomic function.In summary,we have developed genomic selection method(pGBLUP)and genome-wide association method(SMART)that both incorporated functional annotation and improved the performances.The two methods will provide new ideas and tools to study the basis of complex traits and predict the phenotype.Incorporating animal multi-omic functional annotation with our methods will also speed the animal genetic selection and breeding.