Multivariate Genome-Wide Association Study Across Six Breeds for Tenderness and Other Carcass Traits

We seek to improve the power to identify quantitative trait loci (QTL) and train more accurate genomic predictions using multivariate analysis methods and combined data from six breeds of cattle. Using a multivariate genome-wide association analysis should have increased power in the presence of genetic correlations between the different traits.

We analyzed 651 Angus, 695 Charolais, 1,095 Hereford, 283 Limousin, 301 Maine-Anjou, and 516 Simmental samples with phenotypes and genotypes from the Carcass Merit Project (CMP). We used 3,993 Angus, 101 Charolais, 1,225 Hereford, 2,366 Limousin, 11 Maine-Anjou, and 1,913 Simmental purebred animals along with the CMP animals to phase the genotype data and impute the missing genotypes using BEAGLE v3. In GEMMA, we fit a univariate linear mixed model to test SNP associations with each phenotype individually. We also fit multivariate linear models for testing marker associations with multiple phenotypes simultaneously while controlling for population stratification and for estimating genetic correlations among complex phenotypes. We identified single nucleotide polymorphisms (SNPs) that have significant associations with tenderness and other carcass traits. We were able to identify the MSTN QTL in Limousin with the multivariate analyses, but not with univariate analyses. We also trained genomic prediction equations for tenderness using samples from a single breed and a multiple-breed training set. Accuracies of genomic predictions were evaluated using 3-fold cross validation. Correlations were moderate for within-breed training. The overall correlation from the multiple-breed prediction was moderate, but ranged from positive to negative when analyzed for a single breed.