Comparative Analysis of Genomic Prediction for Production Traits Using Genomic Annotation and a Genome-Wide Association Study at Sequencing Levels in Beef Cattle

Leveraging whole-genome sequencing (WGS) that includes the full spectrum of genetic variation provides a better understanding of the biological mechanisms involved in the economically important traits of farm animals. However, the effectiveness of WGS in improving the accuracy of genomic prediction (GP) is limited. Recent genetic analyses of complex traits, such as genome-wide association study (GWAS), have identified numerous genomic regions and potential genes, which can provide valuable prior information for the improvement of genomic selection (GS). In this study, we applied different genome prediction methods to integrate GWAS results and gene feature annotations, which significantly improved the accuracy of GS for beef production traits. The Bayesian models incorporating genomic features showed the highest prediction accuracy, particularly for average daily gain (ADG) and bone weight (BW). Compared to prediction models based on WGS data, GP including biological prior can optimize the prediction accuracy by up to 11.56% for ADG and 14.60% for BW. Also, GP using GBLUP and Bayesian methods integrating biological priors for single-trait GWAS can significantly increase the prediction accuracy. Bayesian methods generally outperformed GBLUP models, with average improvements of 2.25% for ADG, 5.04% for BW, and 3.44% for live weight (LW). Our results indicate that leveraging biological prior knowledge can significantly refine GS models and underline the potential of combining WGS data with biological prior knowledge to further enhance the breeding process.