Genomic prediction and genetic parameter estimation for unsaturated and saturated fatty acids in Canadian dairy cattle.

The current study aimed to obtain the estimates of heritabilities and genetic correlations and the prediction abilities and accuracy of Bayesian GBLUP and Bayesian alphabet (BayesA, BayesB, BayesC) models for total and individual monounsaturated, polyunsaturated and saturated fatty acids from Canadian Holstein cows by using genome-wide SNP markers from genotyping-by-sequencing method. The heritability estimates were obtained from Bayesian GBLUP and Bayesian alphabet models. They ranged from 0.61 to 0.67 for total monounsaturated, from 0.35 to 0.45 for polyunsaturated and from 0.51 to 0.60 for saturated fatty acids, respectively. For thirty-three individual monounsaturated, polyunsaturated and saturated fatty acids, the heritability estimates ranged from 0.27 to 0.69 for individual monounsaturated, from 0.27 to 0.68 for individual polyunsaturated and from 0.35 to 0.69 for individual saturated fatty acids. These results indicated that total and individual monounsaturated, polyunsaturated and saturated fatty acids were under moderate genetic control and can be improved through genomic selection. The estimates of genetic correlations among total and individual monounsaturated, polyunsaturated and saturated fatty acids showed a moderate to high genetic relationships and pointed out the need for consideration of genetic relationships in successful genomic selection for fatty acids traits. The accuracies of BayesC and BayesA models were similar and better than that of GBLUP and BayesB models which indicated that fatty acids were determined by many genes having non-null effects, which are assumed to follow a univariate or multivariate Student's t distribution.