The aim of this study was to explore the possibility of applying GP to important economic traits in the domesticated yak, thus providing theoretical support for its molecular breeding. A reference population was constructed consisting of 354 polled yaks, measuring four growth traits and eight hematological traits related to resistance to disease (involved in immune response and phagocytosis). The Illumina bovine HD 770k chip was used to obtain SNP information of all the individuals. With these genotypes and phenotypes, GBLUP, Bayes B and Bayes Cπ methods were used to predict genomic estimated breeding values (GEBV) and assess prediction capability. The correlation coefficient of the association of GEBV with estimated breeding value (EBV) was used as PA for each trait. The prediction accuracy varied from 0.043 to 0.281 for different traits. Each trait displayed similar PAs when using the three methods. Lymphocyte counts (LYM) exhibited the highest predictive accuracy (0.319) during all GP, while chest girth (CG) provided the lowest predictive accuracy (0.043). Our results showed moderate PA in most traits such as body length (0.212) and hematocrit (0.23). Those traits with lower PA could be improved by using SNP chips designed specifically for yak, a better optimized reference group structure, and more efficient statistical algorithms and tools.
Keywords: disease resistance; domesticated yak (Bos grunniens); genomic prediction; growth performance; prediction accuracy.