A deregression method for single-step genomic model using all genotype data

Abstract

For various applications in dairy cattle evaluation, pseudo-phenotype data are needed for cows with own records or bulls with daughters. EBV deregression using pedigree is performed routinely to generate deregressed proofs (DRP) for the bull MACE evaluation. In some countries DRP for cows with own data or bulls with daughters are used as pseudo-phenotype in the current multi-step genomic step for dairy cattle genomic evaluations. When more and more countries upgrade their current genomic evaluation to a single-step model, genomic-free EBV must be guaranteed for Interbull’s conventional MACE evaluation. Statistical methods were proposed to deregress genomic breeding values of the single-step evaluation using the inverse of the genomic relationship matrix H of the single-step GBLUP model. A high number of genotyped female animals in some countries may lead to a H matrix too large to be inverted, new GEBV deregression methods are therefore needed that are feasible for using genotypes of millions of animals. The purpose of this paper was to develop a GEBV deregression method for the single-step model using all genotype data. A special single-step SNP BLUP model was applied to the GEBV deregression. All animals with own phenotype data and all genotyped animals including young candidates were included in the GEBV deregression. The same pedigree file as well as the same genotype data were considered in the GEBV deregression process as in the original single-step evaluation. Thanks to the efficient single-step SNP BLUP model, the proposed GEBV deregression should be feasible for processing millions of genotyped animals. Methodological and technical issues were addressed, and a validation procedure was proposed for the GEBV deregression method of the single-step evaluation. Analysis of real data will be required to verify the developed GEBV deregression method.