Genomic prediction of milk-production traits and somatic cell score using single-step genomic best linear unbiased predictor with random regression test-day model in Thai dairy cattle

摘要

Cow genotypes are expected to improve the accuracyof genomic estimated breeding values (GEBV)for young bulls in relatively small populations such asThai Holstein-Friesian crossbred dairy cattle in Thailand.The objective of this study was to investigate theeffect of cow genotypes on the predictive ability andindividual accuracies of GEBV for young dairy bullsin Thailand. Test-day data included milk yield (n =170,666), milk component traits (fat yield, proteinyield, total solids yield, fat percentage, protein percentage,and total solids percentage; n = 160,526), andsomatic cell score (n = 82,378) from 23,201, 82,378,and 13,737 (for milk yield, milk component traits,and SCS, respectively) cows calving between 1993and 2017, respectively. Pedigree information included51,128; 48,834; and 32,743 animals for milk yield, milkcomponent traits, and somatic cell score, respectively.Additionally, 876, 868, and 632 pedigreed animals(for milk yield, milk component traits, and SCS, respectively)were genotyped (152 bulls and 724 cows),respectively, using Illumina Bovine SNP50 BeadChip.We cut off the data in the last 6 yr, and the validationanimals were defined as genotyped bulls with nodaughters in the truncated set. We calculated GEBVusing a single-step random regression test-day model(SS-RR-TDM), in comparison with estimated breedvalue (EBV) based on the pedigree-based model usedas the official method in Thailand (RR-TDM). Individualaccuracies of GEBV were obtained by invertingthe coefficient matrix of the mixed model equations,whereas validation accuracies were measured by thePearson correlation between deregressed EBV from thefull data set and (G)EBV predicted with the reduceddata set. When only bull genotypes were used, on average,SS-RR-TDM increased individual accuracies by0.22 and validation accuracies by 0.07, compared withRR-TDM. With cow genotypes, the additional increasewas 0.02 for individual accuracies and 0.06 for validationaccuracies. The inflation of GEBV tended to bereduced using cow genotypes. Genomic evaluation bySS-RR-TDM is feasible to select young bulls for thelongitudinal traits in Thai dairy cattle, and the accuracyof selection is expected to be increased with moregenotypes. Genomic selection using the SS-RR-TDMshould be implemented in the routine genetic evaluationof the Thai dairy cattle population. The geneticevaluation should consider including genotypes of bothsires and cows.