Genomic prediction of growth traits in Brangus beef cattle

Abstract Text:

The availability of a variety of single nucleotide polymorphism (SNP) panels could influence the cost-effectiveness or the accuracy of genomic prediction in cattle. The objective of this study was to evaluate the prediction accuracy of direct genomic breeding values (DGV) for growth traits in Brangus beef cattle. Estimated breeding values (EBV) of the three studied traits (birth weight, weaning weight, and yearling weight) in 1495 progeny–tested Brangus cattle were obtained from the American Brangus Breeder Association along with their assumed heritabilities (Table 1). The cattle were genotyped using one or other of different SNP panels (Bovine50K, GGPHD77K, or BovineHD770K), then all animals were imputed to BovineHD770K using Fimpute. After quality control, 736,053 segregating SNPs remained. Deregressed estimated breeding values (DEBV) were derived from EBV and corresponding accuracies and then fitted in a weighted Bayesian analysis to estimate SNP effects using GENSEL. BayesB and BayesC models were both applied in this study. The population was divided into 5 mutually exclusive groups using K-means clustering methods, in order to perform cross-validation. The accuracy of prediction was assessed as the genetic correlation between DEBV and DGV. The prediction accuracies were sizeable in these three traits, ranging from 0.52 to 0.71 (Table 1). The accuracies from BayesB were slightly higher than those from BayesC. The results indicated that it is promising to apply DGV for routine evaluation in Brangus cattle. Breeders could improve prediction accuracy in young animals and therefore increase selection response by selecting on DGV.

Table 1. Heritability and prediction accuracy (correlation) from BayesC and BayesB.

Trait	h2	BayesC	BayesB
Birth weight	0.42	0.65	0.71
Weaning weight	0.24	0.52	0.54
Yearling weight	0.27	0.52	0.54

Keywords: deregressed estimated breeding value, genomic prediction, Brangus