Accuracy of genomic prediction of dry matter intake in Dutch Holsteins using sequence variants from meta-analyses

We evaluated the accuracy of biology informed genomic prediction for dry matter intake in 2,162 Dutch Holstein cows. Sequence variants were selected from meta-analyses including GWAS summary statistics for QTL and metabolomic QTL in several dairy and crossbred beef populations. Selected variants were prioritized in GBLUP models in a five-fold cross-validation. The accuracies were compared to genomic prediction based on routine 50k genotype data. The average accuracy for the 50k scenario was 0.683. Adding selected sequence variants in the GBLUP model did not improve the accuracies for dry matter intake. Next steps will include testing Bayesian variable selection methods to prioritize variants in genomic prediction for dry matter intake.