Multiple-trait Bayesian Regression Methods with Mixture Priors for Genomic Prediction

Abstract

Bayesian multiple-regression methods incorporating different mixture priors for marker effects are widely used in genomic prediction. Improvement in prediction accuracies from using those methods, such as BayesB, BayesC and BayesC{pi}, have been shown in single-trait analyses with both simulated data and real data. These methods have been extended to multi-trait analyses, but only under a specific limited circumstance that assumes a locus affects all the traits or none of them. In this paper, we develop and implement the most general multi-trait BayesC{Pi} and BayesB methods allowing a broader range of mixture priors. Further, we compare them to single-trait methods and the \"restricted\" multi-trait formulation using real data. In those data analyses, significant higher prediction accuracies were sometimes observed from these new broad-based multi-trait Bayesian multiple-regression methods. The software tool JWAS offers routines to perform the analyses.