Using Pseudo-observations to combine Genomic and Conventional Data in the Dutch National Evaluation

Abstract Text: Since 2010, the Dutch-Flemish evaluation use a post-processing step to estimate Genomically Enhanced Breeding Values in their national evaluation. To optimize the use of genomic data, this paper describes a method where Direct Genomic Values (DGV) are transformed into equivalent daughter performances (pseudo-records), and analyzed in a multi-trait animal model as pseudo-trait to the conventional trait-of-interest with a heritability close to unity, and a genetic correlation set to the predictive value of the DGV. This method can account for pre-selection of data, allows for an animal model in which data can flow to relatives, is easy to incorporate technically, and avoids problems of genotype ownership, by directly using DGV. Results for overall conformation traits are shown as application of this method.

Keywords: dairy cattle, genomic prediction, breeding value estimation, genetics