A weekly genomic evaluation of newly genotyped selection candidates based on a single-step genomic model

Abstract

A single-step SNP BLUP model was developed for routine genomic evaluation of German Holstein. The current weekly genomic evaluation of young selection candidates based on a multi-step SNP BLUP model needed to be upgraded to optimally use the effect estimates from the single-step model. For indirect genomic prediction of newly genotyped selection candidates, two alternative statistical methods were assessed, an exact GRV method and a summation method. Both methods calculated direct genomic values using the SNP effect estimates from the full evaluation in the same way, but they differed in the computation of residual polygenic effects for the young candidates. GEBV of the candidates from the two methods were then compared to those from a single-step evaluation using phenotypic, genotypic and pedigree data from April 2023. To investigate the accuracy and bias of the two weekly evaluation methods, all 1,318,720 genotyped Holstein animals were divided into a reference set containing 1,169,502 animals born before April 2022 and a validation set of 149,218 animals born after April 2022. For all 69 evaluated traits in the German dairy cattle evaluation, correlation of GEBV of the weekly evaluation with the full evaluation was unity for the exact GRV method and ranged from 0.996 to 1 for the summation method. The regression coefficient of GEBV the full evaluation on the weekly evaluation was 1 for the exact GRV method and ranged from 0.988 to 1.002 for the second summation method. The two statistical methods for the indirect prediction of young candidates were shown to be accurate and unbiased.