A Composite Genome-wide Association Study Strategy For Bone Weight And Carcass Weight In Simmental Cattle

Carcass weight is one of the most importantly economic trait in beef,which is the most direct profit of beef industries.Bone weight,commonly referred to as the sum weight of the bone in beef carcass,can reflect some body conformation traits and skeletal diseases related to bone mineral density.Currently,Genome-wide association study(GWAS)is applied to identify the causal variants of importantly economic traits in livestock.The most popular method of GWAS now is linear mixed model(LMM),whereas the LMM has four main pitfalls:(i)LMM cannot consider the joint effects of multiple markers.(ii)The Bonferroni correction is too conservative.(iii)Double-fitting of the testing marker will cause the increase of type I error.(iv)LMM cannot calculate the effects of rare variants.In order to avoid the above four pitfalls and gain a better understanding of the underlying genetic mechanism of these two traits,we performed a GWAS experiment with a composite strategy which includes three models.(i)Single marker regression model:We used LMM there,which is the most popular GWAS method in livestock.(ii)Multi-regression model(LMM-Lasso):SNPs with P-values less than 0.05 from LMM were selected to perform Lasso in the second stage.(iii)Rare variants association method(gene-based SKAT):Genes containing two or more rare variants were examined performing Sequence Kernel Association Test(gene-based SKAT).Our experimental population includes 1225 Simmental cattle,all of which were genotyped by Illumina Bovine HD BeadChip which contained 770,000 SNPs.After the procedures of quality-control,1217 individuals together with 608,696 common SNPs and 105,787 rare SNPs(with 0.001<minor allele frequency<0.05)remained in the sample for subsequent analysis.For the bone weight trait in Simmental cattle,our composite GWAS strategy detected ten significant genes including RIMS2、CHSY1、PRKAR2B、LCORL、LAP3、MAP2K6、NR2F2、CHD7 and LARP4.Among them,LMM successfully identified three candidate genes,all of which had been reported before.LMM-lasso identified not only all three candidate genes in LMM,but also other six significant genes.Gene-based SKAT only detected one significant gene.For the carcass weight in Simmental cattle,our composite GWAS strategy detected six significant genes including SNORA76、ARVCF、TMEMI82、ANGPT4、RIMS2 and TXNDC11.LMM successfully identified three significant genes,while LMM-lasso detected four significant genes.The most significant gene in LMM-lasso is same with the most significant genes in LMM.However,gene-based SKAT did not detect any significant genes.The results of our experiments revealed that the results of LMM and LMM-lasso are quite similar,and LMM-lasso could detect more significant SNPs than LMM.Apart from that,rare variants association in this paper has a relative low power,which only detected only one significant genes associated with bone weight of Simmental cattle.We think this phenomenon mainly because of the limited number of rare variants in SNP array.With the development of sequence techniques and the increasement of SNP array,rare variants association could become an essential part of GWAS.Our composite GWAS strategy offers the opportunity to gain a more deep and more rounded understanding of the genetic mechanism of importantly economic traits in livestock.Moreover,this kind of composite GWAS strategy provides experience for subsequent GWAS experiments.