The Study Of Using Bin Model For Genomic Selection In Simmental Beef Cattle

With the availability of high-throughput genotyping and sequencing technologies,genomic prediction based on high-density(HD)and whole-genome sequencing(WGS)data is now feasible in animal breeding schemes.We face the challenge for dealing with the large data.However,few dimensional reduction algorithms are applied in GS because of worrying about the loss of causal variants.So,we proposed a Bin Model to reduce the data dimension by combining neighboring SNPs into bins,which helps to incorporate genomic information and take advantage of the LD information.Our objective was to compare prediction ability with high-density array data and imputed whole genome sequencing data for economically traits in Simmental beef cattle using various approaches with or without Bin Models.Firstly,Genomic estimated breeding values(GEBVs)for two traits with HD data were predicted using genome best linear unbiased prediction model(GBLUP),elastic net(EN),BayesB and BinGBLUP/EN/BayesB.The results have shown that the Bin-Methods can reduce the computing time of genomic prediction while the prediction accuracy remain the same in most cases.There was the same trend in mean(MSPE)of genomic prediction.Also,we estimated the GEBV for four traits with WGS data using Original-GBLUP and Bin-GBLUP.A total of 1217 Simmental beef cattle were genotyped with Illumina BovineHD Beadchip genotypes.The BovineHD genotypes(658,513 SNPs)of each bull were used to impute whole-genome sequence genotypes(10,550,707 SNPs)using the Beagle software.The results have shown that the prediction accuracy ranged from 0.188 to 0.284.Accuracies of genomic predictions for both sequence data and BovineHD chip data were equal or slightly higher with them of Bin-GBLUP,when compared with Original-GBLUP.However,our results showed that no benefit was gained when using imputed WGS data to perform genomic prediction compared to using HD array data.Bin imputed sequence data than with BovineHD chip data.To investigate the putative advantage of genomic prediction using imputed sequence data,a training set with a larger number of individuals that are distantly related to each other and Bin-BayesR should be considered in genomic prediction models with sequence data.