Weighting factors for genomic information used in Single-Step genomic selection in Australian beef

Abstract

Single-step genomic evaluation utilises all phenotypes, pedigree and genotypes and could significantly enhance beef cattle genetic evaluation. An appropriate weighting factor for genomic and pedigree information is required to predict single-step estimated breeding values (EBVs). This study assessed the optimal weighing factor lambda (λ, ranging between 0 and 1 for none to 100% weighing on genomic information) for a series of beef traits using an empirical approach. The optimal value of λ was identified from the maximum accuracies of genomic predictions by internal cross-validation. The estimated genomic accuracies for Brahman cattle ranged from 0.23 to 0.70 for traits with adequate numbers of genotypes and phenotypes. The accuracy of genomic predictions generally increased as the λ weighting factor increased for a range of traits and typically approached an asymptote towards the optimal λ. For traits with adequate numbers of records, the optimal λ values ranged from 0.4 to 0.8.