A Bayesian Method of QTL Mapping for Multiple Traits in Recombinant Inbred Lines Derived from a Four-Way Cross

In the previous study (Hayashi et al.,PAG2012), we developed a Bayesian method for QTL mapping in recombinant inbred lines derived from a four-way cross, which are hereafter referred to as four-way RILs. We extended this Bayesian method for four-way RILs to simultaneous analysis of multiple traits in the current study. The phenotypes of multiple traits were expressed with a linear model incorporating all available markers as covariates. In the Bayesian framework, we treated the marker effects which corresponded to the regression coefficients in this model as random variables and assumed a multivariate normal distribution with zero mean and a locus-specific variance-covariance matrix for the prior distribution of the effects of each marker. Moreover, we assumed that the prior distribution of this variance-covariance matrix was an inverse-Wishart distribution. We applied the proposed method to the analysis of actual data of soluble solid content (SSC) and total fresh weight (TFW) of fruit in tomato RILs derived from a cross between two F₁ cultivars, which had the same genetic structure as four-way RILs except that the four founder parents were unknown. As a result of the analysis, the pleiotropic effects of QTLs affecting both SSC and TFW were detected.