Comparison of BLUPF90IOD3 and MiXBLUP implementations of the single-step model applied to the Polish national dairy cattle evaluation

The integration of phenotypic, genomic and pedigree data into a single-step model for predicting genomically enhanced estimated breeding values (GEBVs) has become crucial for the accurate genetic evaluation of dairy cattle. This study compared two widely used software implementations, MiXBLUP and BLUPF90IOD3, for the prediction of breeding values using the single-step G-BLUP model based on data from the Polish national evaluation for stature. Four core animal sets were tested, which differed in the selection of bulls and cows. The GEBVs were predicted and validated using different subsets of the population. Both software packages resulted in high correlations (0.89 and 0.97) between full and truncated dataset predictions and similar validation performance, with MiXBLUP exhibiting slightly greater consistency across different sets of core animals. The ranking of the top 50 bulls remained stable across the implementations. This study concludes that both software implementations provide comparable GEBV predictions, suggesting that software choice should consider computational efficiency, cost and modeling flexibility, with MiXBLUP offering additional options for GEBV estimation.