141 Exploring Alternative Approaches for the National Argentine Angus Cattle Genomic Evaluation Program

Abstract

Abstract Beef cattle production is a relevant component of the Argentine economy. The Argentine Angus is the most common beef cattle breed, and this population has a large economic and cultural importance for the country. In 2019, a national genomic selection program was implemented in the Argentine Angus population. The objective of this study was to explore alternative approaches for the genomic selection program in Argentinean Angus so that it will better serve the stakeholders. For this study, a total of 9,944 phenotypes for weaning weight from Argentine Angus born in 2018 were used. All animals had pedigree information for at least 3 generations. Genomic information on 1,078 individuals for 150k SNPs was used. The genotyped individuals were connected to the population by pedigree. Weaning weight EPDs were calculated under an animal model considering contemporary group as fixed effects and additive direct and maternal genetic effects as random. Variance components were estimated using AIREMLF90, and EPDs were calculated using BLUPF90. The enhanced EPDs were the genomic estimated breeding values obtained by single-step GBLUP, by including the genomic information into the original model. Validation of the EPDs and enhanced EPDs were made by cross-validation, where a random sample of 70 genotyped individuals with more than 20 progeny was the validation population. The reference breeding values were the EPDs of the validation population with complete data. In a second analysis, the phenotype of the progeny of the validation population was removed. The correlation between the reduced and complete EPDs was the accuracy for this study, according to the LR method. The accuracy obtained by this method was overestimated due to potential relationships between animals in the training and validation population. However, the enhanced EPDs were 4% more accurate than the pedigree EPDs. The next steps of this research are exploring alternative approaches of the validation methodology, which will be accomplished by implementing a forward-in-time cross-validation. Such validation will be performed once more data is included in the model. Additionally, other models will be tested to identify the ideal combination of effects that will yield higher accuracies. Details regarding the construction of the genomic relationship matrix will be tested with the overarching goal of improving convergence without affecting accuracy. Moreover, a weighted single-step GBLUP approach will be tested to verify if prediction accuracy can improve even further. Finally, indirect prediction models will be tested so that the Argentine Angus Association can implement interim enhanced EPDs for young animals. In conclusion, including genomic data into this dataset yielded greater accuracy and such values are expected to improve even further after optimization.