Genome-wide association study and its impact on the accuracy of genomic prediction for live body weights in Australian Merino sheep

Body weight traits such as birth weight (BW), post-weaning weight (PWW), and adult weight (AW) are economically important traits in sheep. While numerous quantitative trait loci (QTL) have been identified for production traits in other animals over the past decades, few QTL studies have been reported for sheep. This study aimed to elucidate the genetic architecture underlying body weight traits in Australian Merino sheep using medium (50K) and high-density (600K) SNP arrays as well as assessing the accuracy of predicting genetic merit for those traits in both density panels and the potential impact of explicitly incorporating QTL information on prediction accuracy. By conducting a Genome-Wide Association Study with 6890, 4169, and 2963 Merino sheep for BW, PWW and AW, respectively, we identified significant SNP associated with these traits. The single regression analysis revealed 15, 4, and 8 significant SNPs for BW, PWW, and AW, respectively, using the medium-density Chip, and 207, 53, and 53 significant SNPs using the high-density chip. We identified 27 genes located on OAR6 and OAR11 that were associated with body weight of sheep. Genomic predictions based on genomic best linear unbiased prediction demonstrated moderate to high accuracy (ranging from 0.46 to 0.63) for predicting genetic merit in body weight traits. Notably, we observed a 0.01 to 0.05 increase in prediction accuracy when utilizing the high-density panel compared to the medium-density panel. Moreover, incorporating the top two most significant SNPs as separate random effects into the prediction model led to an increase from 0.006 to 0.045 in the prediction accuracy of those traits. However, fitting a larger number of significant SNPs gave generally a decrease in prediction accuracy.