Approximating the accuracy of single step EBVs

Abstract

To accompany the implementation of multi-trait Single Step Genomic BLUP (SS-GBLUP) in the BREEDPLAN and OVIS genetic evaluation systems, an algorithm to approximate accuracy with genomic information has been developed and is presented in this paper. Data from full terminal sire OVIS and Brahman BREEDPLAN runs were processed using this new method. Results demonstrate that the approximated accuracy of SS-GBLUP estimated breed values (EBVs) is highly correlated (R² >0.96) with exact accuracies in several small example analyses for both beef and sheep. SS-GBLUP EBV accuracies increase more for traits with a larger reference population and for traits with higher heritabilities. Animals with low pedigree-only (ABLUP) EBV accuracies benefit more from genomic information than animals with high ABLUP EBV accuracies.