Do rare variants contribute to the genomic prediction accuracy?

Abstract Text: The major goal of this study is identification of SNPs with rare allelic variants i.e. with minor allele frequency lower than 1%, in data set of bulls from Polish Holstein-Friesian breed, and comparison of accuracy of breeding value prediction for data sets with and without rare alleles. Data set consisted of 3,100 proven and 1,968 young bulls. Each bull was genotyped using 50K Illumina BeadChip. In our analysis production, fertility and udder health traits were considered. Using SNP and SNP-BLUP model two evaluations were carried out: (1) with all available SNPs, including rare variants (53,862 SNPs); (2) with common SNPs only, for which minor allele frequency exceeds 1% (46,267 SNPs). Finally, statistical significance of SNP estimates and reliability of predicted breeding values were compared between two data sets. Results showed that including rare variants into analysis increase the accuracy of DGV and GEBV. Keywords: dairy cattle, rare variants, genetics