133
Searching for Causal Networks Involving Latent Variables in Complex Traits: An Application to Growth, Carcass, and Meat Quality Traits in Pig
Structural equation models (SEM) are used to model causal relationships between multiple traits. Among the strengths of SEM is the ability to consider latent variables. It is worth to note that in a quantitative genetics context, causal inference cannot be performed directly on the joint distribution of the traits under study because causal links can be masked by genetic covariances. Here, we describe a method for assessing causal networks involving latent variables conditionally to unobservable genetic effects. We applied this method to a dataset with 413 F2 pigs for which several phenotypes were recorded over time. Causal relationships involving growth, carcass and quality traits, modeled using five latent variables, were evaluated. Interestingly, we found that growth and carcass traits have a negative causal effect on quality traits. Overall, the proposed method allows further learning regarding phenotypic causal structures underlying complex traits.
Keywords: Causal inference, Complex traits, Latent variables
