Comparative Analysis of Bayesian Methods and GBLUP for Genomic Evaluations in Dairy Cattle

Abstract

In this study, the effects of Genomic Best Linear Unbiased Prediction (GBLUP) and Bayesian alphabet methods (A, B, C and Cp) were investigated on genomic predictions and indirect estimations in the Polish Holstein Friesian (PHF) dairy cattle population. The study analysed the milk yield data (MY, kg/lactation) and 13,481 single nucleotide polymorphism (SNP) genotype records from 534 Polish Holstein Friesian (PHF) dairy cattle raised on private farms in Poland. The quality control of the genotypic data included the removal of monomorphic loci and the exclusion of samples with SNP missing rates exceeding 10%. After the quality control, 493 animals and 13,250 SNPs were retained for the genomic prediction. Marker effects and genomic breeding values (GEBVs) were calculated using the Bayesian alphabet and GBLUP. The results indicated that, for the milk yield of PHF cows, the Bayes C method outperformed other approaches. This method achieved the highest prediction accuracy among the evaluated methods. Additionally, the Bayes C method required the shortest computational time, underscoring its efficiency.