Performance of weighted genomic BLUP and Bayesian methods for Hanwoo carcass traits.

Abstract

To improve the quality and yield of the Korean beef industry, selection criteria often focus on estimated breeding values for carcass weight (CWT), eye muscle area (EMA), backfat thickness (BF), and marbling score (MS). This study estimated genetic parameters and assessed the accuracy of genomic estimated breeding values (GEBVs) using SNP weighting methods. We compared the accuracy of these methods with the genomic best linear unbiased prediction (GBLUP) and various Bayesian approaches (BayesA, BayesB, BayesC, and BayesCPi) for the specified traits. The study used single-trait animal models, including GBLUP, weighted GBLUP (WGBLUP), and the Bayesian methods to predict genomic breeding values in a population of Hanwoo steers. A total of 19154 phenotypes were collected with all animals genotyped using the Illumina Bovine 50 K SNP chip. The average heritability for the carcass traits was 0.33 (GBLUP) and 0.35 (Bayesian), with Bayesian methods yielding heritability estimates that were on average 0.02 points (6.1%) higher than GBLUP. The accuracy of genomic predictions ranged from 0.7-0.83 (GBLUP), 0.83-0.87 (WGBLUP), and 0.81-0.87 across the Bayesian methods. WGBLUP accuracies for the carcass traits were, on average 8.97% higher than the GBLUP accuracies and 1.80% higher than the Bayesian alphabets. The Bayesian alphabet's accuracy is also, on average 6.00% higher than the GBLUP accuracy. According to these findings, the weighting GBLUP approach provides higher prediction accuracy for Hanwoo carcass traits than the Bayesian alphabet. Therefore, WGBLUP can be used for genomic selection in the Hanwoo evaluation program.