Accounting for Missing Pedigree Information with Single-Step Random Regression Test-Day Models

Genomic selection is widely used in dairy cattle breeding, but still, single-step models are rarely national evaluations. New computing methods have allowed the utilization of very large genomic data sets. However, an unsolved model problem how to build genomic- (G) and pedigree- (A22) relationship matrices that satisfy theoretical assumptions about same scale equal base populations. Incompatibility issues also been observed manner which genetic groups included model. In this study, we compared three approaches for accounting missing pedigree information: (1) GT_H full Quaas Pollak (QP) transformation groups, including both pedigree-based genomic-relationship matrices, (2) GT_A22 partial QP omitted G−1, (3) GT_MF metafounder approach. addition models, (4) official animal with a unknown parent (UPG) from (5) approach were comparison. These tested Nordic Holstein test-day production models. The 8.5 million cows total 173.7 records 10.9 animals pedigree, there 274,145 genotyped animals. All VanRaden method 1 G had 30% residual polygenic proportion (RPG). scaled average diagonal A22. Comparisons between based on Mendelian sampling terms forward prediction validation using linear regression solutions full- reduced-data Models gave similar results overprediction. MF showed lowest bias.