Prioritization of SNP markers for genomic prediction in closed beef cattle populations

With the advances in high-throughput technologies, genomic information is becoming readily available. This has led to whole genome sequences and denser single nucleotide polymorphism (SNP) panels being generated for more individuals. However, the increase in genomic information has shown little benefit in improving the prediction accuracy of genomic estimated breeding values (GEBV). One method to best utilize the increased amount of SNP information is to optimize the selection of informative SNP markers. In this study, genomic prediction of growth traits in two closed beef cattle populations using various prioritization techniques was evaluated. The first population used is Line 1 Hereford. The data consisted of 1192 animals with genotypes and phenotypes. The second population is a composite breed (50% Red Angus, 25% Charolais, 25% Tarentaise) and included of 2776 genotypes and phenotypes. The SNP prioritization methods adopted in this study were based on fixation index (Fst) and GWAS based SNP marker effects. Using a subset of prioritized SNP markers increased the accuracy for all three traits for the Line 1 Hereford population. On the other hand, using a weighted G matrix based on Fst and SNP effects did not increase the accuracy and in some instances decreased. Furthermore, the predication accuracy was higher in Line 1 Hereford which is an inbred population compared to the composite population. The study showed that prediction accuracy of GEBV can be improved with SNP prioritization, however it is population specific, trait specific and model specific. Moreover, this study highlights the importance of population structure in the prediction accuracy of GEBV.