Multi-trait genomic selection: comparison of methods

Genomic selection is becoming common practice in animal breeding. It uses genome-wide dense marker maps, to accurately predict the genetic ability of animals, without the need to record phenotypic performance from the animal itself or from close relatives. Therefore, it is particularly beneficial for selection of traits that are difficult or expensive to measure. Without genomic selection, selection depends on predictor traits in multi-trait breeding value estimation. To get the benefit of recording of predictor traits in genomic selection, our objective is to develop a multi-trait genomic selection method. Four different multi-trait models are considered: 1) a model with a traditional pedigree based relationship matrix, 2) a model where the traditional pedigree based relationship matrix is replaced by a genomic relationship matrix based on markers, 3) a model that both includes polygenic effects related through pedigree information and SNP effects without considering the covariance between traits when estimating the SNP effects, and 4) the same as model 3, with considering the covariance between traits when estimating the SNP effects. The second model assumes equal contribution of each SNP to the total additive genetic (co)variance, while model 3 and 4 allow unequal contributions per SNP. Those 4 models will be compared for 2 simulated traits, having a genetic correlation of 0.2, 0.5 or 0.8 between them. Scenarios are included where some animals have phenotypes for one trait, but not for the other.