Genomic prediction using high-density SNP markers in Nordic Holstein and Red

Abstract

This study investigated genomic prediction using medium density and high density markers, based on data from Nordic Holstein and Red (RDC). The Holstein data comprised 4,539 progeny-tested bulls, and the RDC data 4,403 bulls. The data were divided into reference data and test data using 2001-10-01 as a cut-off date (birth date of the bulls). This resulted in about 25% genotyped bulls in Nordic Holstein test data, and 20% in RDC test data. For each breed, three datasets of markers were used for predicting breeding values: 1) 50k dataset with some missing markers, 2) 50k dataset where missing markers were imputed, and 3) imputed HD dataset which was created by imputing the 50k data to HD data based on 557 bulls genotyped using 770k chip in Holstein, and 706 bulls in RDC. Based on the three marker datasets, direct genomic breeding values (DGV) for protein, fertility and udder health were predicted using a GBLUP model and a Bayesian mixture model with two normal distributions. Reliability of DGV was measured as squared correlations between de-regressed proofs (DRP) and DGV and then corrected for reliability of DRP, and unbiasedness was assessed by regression of DRP on DGV, based on the bulls in the test datasets. Averaged over the three traits, reliability of DGV based on the HD markers was 0.5% higher than that based on the 50k data in Holstein, and 1.0% higher in RDC. In addition, the HD markers led to an improvement on unbiasedness of DGV. The Bayesian mixture model led to 0.5% higher reliability than the GBLUP model in Holstein, but not in RDC. Compared with the raw 50k data, the imputed 50k data improved genomic prediction for protein in RDC.