Multitrait Across Country Genomic Evaluations for EuroGenomics Countries ─ Daughter information approximation method

Abstract

EuroGenomics countries share bull genotypes, enabling multitrait across country SNP-BLUP evaluations using pseudo phenotypes from all countries directly. We have previously shown that such model is feasible and benefits the participants. However, in the EuroGenomics countries there is a vast amount of cow data and cow genotypes that the countries would like to be included into the model, but without sharing the actual genotypes. The Daughter information approximation method is our attempt to accomplish this. Our method would require sharing only the national full reference population SNP- solutions, and the prediction error variances of the already shared bulls. The method is pilot tested by partitioning the currently shared EuroGenomics data into subsets mimicking the shared bulls and the unshared parts of the national reference populations.