Global Interbull EBV in domestic single step genomic evaluation


  • Josef Přibyl
  • Jiří Bauer
  • Jana Přibylová
  • Luboš Vostrý
  • Ludmila Zavadilová Institute of Animal Science, v.v.i , Prague, Czech Republic
  • Václav Čermák
  • Zdenek Růžička
  • Jiří Šplíchal
  • Martin Verner
  • Jiří Motyčka
  • Ladislav Vondrášek


genomic breeding value, single-step prediction, animal model, validated reliability


EBVs and GEBVs for milk production of young genotyped Holstein bulls were predicted using a conventional Animal Model, Ridge Regression, GBLUP, and ssGBLUP procedures. For prediction, the nation-wide database of domestic Czech production records was combined with deregressed proofs from Interbull files through 2008, which had been transformed by MACE to reflect domestic production conditions. 1,259 genotyped bulls had already been proven in 2008. Analyses were run that used Interbull values only for these genotyped bulls and used Interbull values for all available sires. Predictions were validated by comparing correlations of breeding value predictions with  EBV and DYD after progeny test in 2012 of 140 young genotyped bulls and their associated reliabilities. Combining domestic data with Interbull EBVs improved prediction of both EBV and GEBV. Prediction by Animal Model (traditional EBV) using only the domestic database had 0.29 validated reliability of prediction; whereas combining the nation-wide domestic database with all available deregressed proofs for genotyped and non-genotyped sires from Interbull resulted in reliability of 0.34, compared to 0.36 when using Interbull data only. Highest reliabilities were for predictions from the ssGBLUP procedure using combined data, or with all available deregressed proofs from Interbull only, which reached validated reliabilities for GEBV predictions 0.53 and 0.54 respectively.