Genomic growth curves of an outbred pig population

Abstract

The success of pig production systems, including the evaluation of alternative management and marketing strategies, requires knowledge of the body weight behavior over time, commonly referred to as the growth curve. This knowledge allows the assessment of growth characteristics in actual production situations and translates this information into economic decisions. Differences among animal growth curves partly reflect genetic influences, with multiple genes contributing at different levels to the overall phenotype. Hence, selection strategies that attempt to modify the growth curve shape to meet demands of the pork market are very relevant. In the current post-genomic era, understanding the genomic basis of pig growth cannot be limited to simply estimating marker effects using body weight at a specific time as a phenotype, but must also consider changes in body weight over time. According to Pong-Wong and Hadjipavlou (2010) and Ibáñez-Escriche and Blasco (2011) this can be done by estimating the marker effects for parameters of nonlinear regression models that are widely used to describe growth curves. Regardless of the phenotype used, a major challenge in genome-wide selection (GS) is to identify the most powerful statistical methods for predicting phenotypic values based on estimates of marker effects. Since the seminal GS paper by Meuwissen et al. (2001), several studies have compared the efficiency of simple methods, such as the RR-BLUP (Random Regression Blup) (Meuwissen et al., 2001), with more sophisticated methods, such as Bayesian LASSO (BL) (de los Campos et al., 2009). The main difference between these two very popular GS methods is that the first one assumes, a priori, that all loci explain an equal amount of genetic variation, while the second one allows the assumption that each locus explains its own amount of this variation. Although these two methods have already been compared in other studies, so far there has been no comparison of these methods using a major gene, such as the halothane gene in pigs (Fujii et al., 1991), as a marker. In addition, these methods have not yet been applied to the analysis of growth curves in conjunction with nonlinear regression models. In this study, we compared the accuracies of RR-BLUP and BL for predicting genetic merit in an empirical application using weight-age data from an outbred F2 (Brazilian Piau X commercial) pig population (Silva et al., 2011). In this approach, the phenotypes were defined by parameter estimates obtained with a nonlinear logistic regression model and the halothane gene was considered a single nucleotide polymorphism (SNP) marker in order to evaluate the assumptions of the GS methods in relation to the genetic variation explained by each locus. Genomic growth curves based on genomic estimated breeding values were constructed and the most relevant SNPs associated with growth parameters were identified.