The relative importance of information on unrelated individuals on the prediction of genomic breeding values

Abstract

The theory of genomic selection is based on the prediction of the effects of genetic markers in linkage disequilibrium (LD) with quantitative trait loci (QTL). However, there is increasing evidence that genomic selection also relies on relationships between individuals or the patterns of LD associated with these relationships to accurately predict genetic value. This study aimed to examine the relative importance of information on essentially unrelated individuals on the estimation of breeding value when using gBLUP and BLUP. Analysis was undertaken using a simulated population of 2000 animals. Two reference populations were formed from 1750 animals and the accuracy of prediction was assessed for the remaining 250 animals that formed the test population. Two test populations were constructed such that one included 10 families that had no family members in the reference population and the other included 5 half siblings from 50 families. The gBLUP method more accurately predicted breeding value than BLUP in both test populations. The highest accuracy was achieved when gBLUP was used to predict the breeding value of closely related animals. However, gBLUP was still able to predict breeding value accurately even when animals were distantly related.