Using RNAseq data to improve genomic selection in dairy cattle

Abstract

Genomic selection is playing an increasingly important role in animal breeding worldwide. For reasons of cost and computational feasibility, it is useful to select variant sets of maximum predictive ability, while minimising the number of variants required. RNA sequencing has the potential to aid variant selection in two ways: first, by enabling the discovery of variants of strong biological relevance (by virtue of their expression in tissues of interest); and second, by empowering the discovery of expression QTL, enabling enrichment of variant data with loci of demonstrable, modulatory effect. In this study, RNAseq was performed on lactating mammary gland from 373 New Zealand dairy cattle, followed by variant calling and eQTL discovery. Significant eQTL were identified at 3,738 genes, yielding 3,695 distinct tag variants, which were subsequently tested for their ability to predict milk volume, fat and protein phenotypes in a genomic selection context. We show that variants selected in this manner show good predictive abilities for these phenotypes, and that RNAseq is a potentially useful approach for enhancing variant selection.