Increasing the Accuracy of Genomic Estimated Breeding Values of Milk Traits in New Zealand Dairy Cattle Using DGAT Genotypes

Accurate genomic estimated breeding values (GEBVs) are required to prevent accumulation of inaccuracies in pedigrees with parents that were unproven when selected. This study quantified the effects of including the causal mutation for DGAT, the largest QTL affecting milk traits, in calculating GEBVs. Data on 5,661 Holstein cows were provided by LIC, a New Zealand dairy cattle company, and consisted of Illumina BovineSNP50 genotypes (50k) on all cows, 1,133 actual DGAT genotypes, and deregressed estimated breeding values (DEBV) for milk fat. DGAT genotypes were imputed for the remaining 4,528 cows using BEAGLE. Three Bayes C models with 97.5% of SNPs assumed to have zero effect were run in GenSel with 5-fold cross-validation: 1) one model relying on associations between markers in the SNP panel and causal mutations, 2) one with the causal DGAT mutation fit as a random effect like any other SNP, and 3) one fitting DGAT genotypes as a fixed class effect. The accuracy of GEBVs, measured as the correlation between GEBVs and DEBVs, was 1) 0.377 using 50k SNPs, 2) 0.381 with DGAT as a random effect, and 3) 0.389 fitting DGAT as a fixed effect. Adjusting for non-genetic variation in DEBV, by dividing by root heritability, correlations between GEBV and BV would be about twice that magnitude. The results demonstrate the usefulness of fitting DGAT genotypes as a fixed effect when the QTL genotypes are readily available by genotyping or imputation.