Genomic prediction for beef fatty acid profile in Nellore cattle

Abstract

The objective of this study was to compare SNP-BLUP, BayesCπ, BayesC and Bayesian Lasso methodologies to predict the direct genomic value for saturated, monounsaturated, and polyunsaturated fatty acid profile, omega 3 and 6 in the Longissimus thoracis muscle of Nellore cattle finished in feedlot. A total of 963 Nellore bulls with phenotype for fatty acid profiles, were genotyped using the Illumina BovineHD BeadChip (Illumina, San Diego, CA) with 777,962 SNP. The predictive ability was evaluated using cross validation. To compare the methodologies, the correlation between DGV and pseudo-phenotypes was calculated. The accuracy varied from − 0.40 to 0.62. Our results indicate that none of the methods excelled in terms of accuracy, however, the SNP-BLUP method allows obtaining less biased genomic evaluations, thereby; this method is more feasible when taking into account the analyses' operating cost. Despite the lowest bias observed for EBV, the adjusted phenotype is the preferred pseudophenotype considering the genomic prediction accuracies regarding the context of the present study.