266. On the advantage of identifying causal genetic variants for genomic prediction

Abstract

Genome Wide Association Studies have been successful in identifying Quantitative Trait Locus regions. However, the identification of causal variants (CVs) has proven difficult in many cases. One may hypothesize that knowing the CV would not improve the (genomic) prediction of phenotypes much since otherwise phenotypes would clearly point to the CV. The effect of (not) knowing the CV on local estimated breeding values (LEBV) in 200 kb regions surrounding the CV was estimated for 3 well-known CVs (in ABCG2, DGAT and GHR) in a large whole genome sequence dataset. Knowing the CV improved the accuracy of LEBV for all three CVs: for ABCG2 it was instrumental in pointing to the correct CV; the DGAT-CV was poorly imputed, and knowing the CV would result in improved genotyping; and GHR was poorly predicted by other SNPs in the region. In addition, knowing CVs is important for understanding trait-biology and for targeted gene-editing.