Assessment of genomic prediction of live weight traits using two validation methods in cattle

Abstract

The accuracy of the estimation (AE) of the genomic breeding values (GEBV) of live weight traits of registered Charolais cattle using two cross validation methods was assessed. A BLUP model and different genomic prediction (PG) methods were fitted: Genomic-based best linear unbiased prediction (GBLUP), Bayes C (BC) and Single-step Bayesian regression (SSBR). AE was compared using randomly formed validation groups (VGs) and CGs. The results showed that all three PG methods provided similar prediction accuracies among GVs but not higher prediction accuracies than BLUP. The prediction accuracy of GBLUP and BLUP was 0.35 and 0.37 for PN, and 0.30 and 0.41 for PD, respectively. The results show that the PG accuracies under the evaluated scenarios are low; Therefore, for its correct implementation, it is necessary to increase the number of animals and use de-returned genetic values as response variables.