Genomic selection for longitudinal traits of Holstein cattle

Abstract

Longitudinal traits are of great significance for animal production, however, when it comes to genomic selection several methods are being evaluated for specific traits, such as milk production in age-specific or total production. Thus, the application of these methods to complete trajectories of the phenotype of interest over time can be an important tool for decision making the best time to select the animals. In this study data will be used for milk production, protein and lactose, and somatic cell count of dairy cattle in a two-stage method, where first the nonlinear Wood (1967), Cobby & Le Du (1978), Wilmink (1987) and Pollott (2000) will be adjusted. The best model will be chosen by different criteria such as mean square residual adjustment (QMR), number of iterations, prediction error, Akaike information criterion (AIC) and Bayesian information criterion Schuwarz (BIC). This model has estimates of its parameters subsequently used as phenotypes. In the second stage will be tested different methods for estimating the genomic values (GBV), such as: RR-BLUP, Bayes A, B and Bayes LASSO, which will consider how phenotypes parameter estimates obtained previously. The matching of phenotypes and GBV will be verified by cross-validation, which will enable the selection of the most appropriate method for assessing longitudinal traits in dairy cattle.