Use of principal component approach to predict direct genomic breeding values for meat traits in Italian Simmental Bulls

In the current study, principal component (PC) analysis was used to reduce the number of predictors in the estimation of direct genomic breeding values (DGV) for meat traits in a sample of 457 Italian Sim- mental bulls. SNP marker genotypes were determined with the 54K Illumina beadchip. After edits, 40,179 SNPs were retained. PC extraction was carried out separately for each chromosome and 2,466 new variables able to explain 70% of total variance were obtained. Bulls were divided into reference and validation population. Three scenarios of the ratio reference/validation were tested: 70:30, 80:20, 90:10. Effect of PC scores on polygenic EBVs was estimated in the reference population with a BLUP model. Traits analyzed were daily live weight gain, size score, muscularity score, feet and legs score, beef index (economic index), calving ease direct effect, and cow muscularity. Accuracy was calcu- lated as correlation between DGV and polygenic EBV in the validation bulls. Muscularity, feet and legs, and the beef index show the highest accuracies (Table 1), calving ease the lowest. In general, accuracies are slightly higher (Table 1) when reference animals are selected at random and the best scenario is 90:10.