The Process of Validating National Genomic Predictions

Abstract

In last years many countries have implemented genomic evaluation for dairy cattle breeding. To validate national genomic models, Interbull introduced a GEBV test (Mäntysaari et al. 2010) in August 2010. Almost all countries or populations have passed the GEBV test for protein yield, many of them also for milk or fat yields. However, an extension of the GEBV test to functional traits seemed to be more difficult, with unsatisfactory test results. The purpose of this study was, therefore, to describe the process of validating national genomic predictions in details using German Holstein population as an example. Several ways for improving genomic prediction were discussed with regard to genomic validation and passing the GEBV test. Increasing residual polygenic variance in SNP effect estimation was shown to be effective to make genomic prediction less biased, and it increased regression slope of the GEBV test, even for female fertility traits with low heritabilities. Using truncated national and MACE conventional evaluations for genomic validation should guarantee the validity of current national conventional evaluation model and the right time frame of phenotypic data, which is especially important for countries sharing a common genomic reference population. The current GEBV test was shown to be an important and valuable test for validating national genomic predictions.