Proceedings, 10 World Congress of Genetics Applied to Livestock Production Prediction of Genomic Breeding Values for Reproductive Traits in Nellore Heifers

The objective of this study was to assess the quality of genomic predictions for rebreeding success or failure after calving, age at first calving, and early pregnancy occurrence in Nellore females. There were 1,853 individuals that had 305,348 SNPs each used in genomic selection analyses. Three genomic prediction methods were used: GBLUP, a Bayesian method (BAYESCπ), and a lasso method (IBLASSO) were used to estimate SNP effects. The dependent variables were: adjusted phenotype (CPHEN), and deregressed breeding value (DEBV). Predictive abilities were assessed by the average correlation between CPHEN and genomic estimated breeding value (GEBV) and by the average correlations between DEBV and GEBV in the validation population. Regression coefficients of CPHEN on GEBVs in the validation population were indicators of prediction bias of GEBV. BAYESCπ was more suitable for estimation of SNP effects and GEBVs for all traits. Estimation of GEBV using CPHEN as dependent variable led to more accurate estimates.