Application of the Interbull genomic reliability method for single-step evaluations of test-day and conformation traits in German Holstein

Abstract

National single-step genomic evaluation required accurate genomic reliabilities, particularly for young, genotyped animals. The Interbull genomic reliability method was tested for single-step evaluation of four test-day traits as well as for 25 conformation traits in German Holstein. Genotypic, phenotypic and pedigree data were taken from the official genomic evaluation in April 2023. More than 1.3 million genotyped animals were considered jointly with non-genotyped animals, and the genomic reference population exceeded half a million animals for the test-day traits. Selecting fewer SNP markers in reliability calculation for direct genomic values (DGV) was proven to be an efficient way of decreasing computing time or memory usage while retaining a reasonable accuracy when at least 15,000 equidistant SNP markers were chosen. Due to the extremely large reference population, the level of DGV reliabilities was very high, also for the young, genotyped candidates. Adjusting the theoretical DGV reliabilities based on the Interbull reliability method seemed to be unavoidable, especially for the large reference population. Variation in the DGV reliabilities was shown to be small among animals born in the same year, especially among the young, genotyped animals without own phenotypic records. Therefore, a constant genomic effective daughter contribution could result in reasonably accurate genomic reliability values and at the same time may provide a computationally much less demanding way for routine genomic reliability calculation with several million genotyped animals included. The single-step genomic reliability values were compared to conventional reliabilities as well as genomic reliabilities from the current multi-step genomic model for diverse groups of animals of German Holstein. The single-step genomic reliabilities of the test-day and conformation traits seemed to be consistent with the variance of genomic breeding values.