174. Genomic prediction using breed origin of alleles model accounting for probabilities in the assignment of the alleles

Abstract

This study investigates genomic prediction using a breed origin of alleles (BOA) model which accounts for BOA with a (i) definitive assignment to a breed or (ii) probabilities of assignments to each breed. Our BOA model estimates breed-specific marker effects based on genotypic and phenotypic information from the purebred and crossbred animals. A traditional combined analysis of all breeds' data implicitly assumes a correlation of one between the marker effects of the breeds, whereas our BOA model allows the estimation of these correlations. We used de-regresed proofs of production traits from the admixed Nordic Red Cattle population to evaluate the model and performed the analysis assuming marker effects between breeds are correlated or assuming they are uncorrelated. We found that using probabilities in the BOA model outperformed the BOA model, assuming the origin of each marker is known with certainty, especially when the marker effects were assumed uncorrelated between breeds.