Genomic validation software: USA update including truncated MACE

Authors: MOTA, RODRIGO - Council On Dairy Cattle Breeding; SULLIVAN, PETER - Collaborator; NICOLAZZI, EZEQUIEL - Council On Dairy Cattle Breeding; MCWHORTER, TAYLOR - Council On Dairy Cattle Breeding; LEGARRA, ANDRES - Council On Dairy Cattle Breeding; Vanraden, Paul

Submitted to: Interbull Annual Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/28/2024
Publication Date: 9/4/2024
Citation: Mota, R., Sullivan, P., Nicolazzi, E., McWhorter, T.M., Legarra, A., Van Raden, P.M. 2024. Genomic validation software: USA update including truncated MACE. Interbull Bulletin. 60:200-206.

Interpretive Summary:

Technical Abstract: With the need to establish a reliable method to validate genomic breeding values (GEBV) to meet the requirements for marketing the semen of young bulls in Europe, the Interbull GEBV Test has routinely added new features to the GEBVtest software. In 2023, the US conducted a GEBV validation and reported that large population breeds and traits with high heritability were more stable, whereas smaller populations and complex traits often failed due to several reasons. In addition, the use of Truncated MACE (TMACE)-based genomic evaluations was recommended to verify if this model would outperform 4-year-old official results. A new version of the GEBVtest software will become the standard for GEBV validation in 2024. This new version adds bootstrapping to improve and expand significance, with better tests for slopes, validation accuracy, and bias and does not allow bulls with GEBV foreign proof to be included as candidates. In this study, GEBV validation was performed using the newest version of the GEBVtest software while validating truncated domestic plus TMACE instead of using official US predictions from 4 years ago and applied to US dairy cattle populations. Nine traits were tested: milk yield (MIL), fat yield (FAT), protein yield (PRO), longevity (DLO), somatic cell score (SCS), heifer conception rate (HCO), cow recycling (CC1), and calving interval (INT). All available traits were tested in all five US breeds receiving genomic evaluations: Holstein (HOL), Jersey (JER), Brown Swiss (BSW), Red Dairy Cattle (RDC), and Guernsey (GUE). GEBVs from August 2023 were used as the full dataset, whereas TMACE-based GEBV were used as the truncated dataset. The use of TMACE-based model allows to accommodate model or data changes over time as well as to validate traits that were not even implemented four years ago such as mastitis for JER and BSW, implemented respectively in 2020 and 2022. In general, the inclusion of TMACE improved results for all breeds. In HOL, all traits except for HCO passed validation. HCO failed with a b1 of 1.31 (>1.2). The slope standard error was 0.02, which confirms an underestimation of this trait. In JER, traits CC1 and INT failed validation and passed for all other traits. There was a clear improvement of the b1 including TMACE for CC1 and INT but not enough to pass the validation test. In BSW, traits FAT and SCS failed validation due to a b1 < 0.8, validation was inconclusive for PRO and passed for the other traits. There were no bulls to validate mastitis at this time. Finally, the smaller breeds RDC and GUE showed several inconclusive passes and much fewer failures compared to a previous study. The results may be again due to the complexity of traits and the small number of candidate bulls. The use of TMACE-based genomic evaluations improves the validation test and is a tool to be considered as standard when performing GEBV validation, especially for smaller breeds.