| Identifying genetic loci responsible for variation in traits which are quantitative in nature poses a formidable challenge to geneticists and statisticians. We focus our attention on experimental crosses and complex traits, which are influenced by a multitude of genes in an interacting network. In a QTL experiment a lot of quantitative traits are measured but they are generally analyzed one at a time. However, joint analysis of multiple correlated traits can improve power and precision of estimates. More importantly, joint analysis allows the testing of biologically important hypotheses like, pleiotropy and pleiotropy vs. close linkage.;In this dissertation, we develop a Bayesian approach to jointly map multiple QTL for correlated traits. We consider two multivariate models, namely, the traditional multivariate model and the Seemingly Unrelated Regression or SUR model. The key difference is that the SUR model allows each trait to have a different set of genetic loci while the traditional model for all traits to have the same set of loci. We develop a computationally efficient Markov Chain Monte Carlo (MCMC) to generate posterior samples for the model parameters, number of QTL, their position and genetic effects. We also model gene-gene and gene-environment interactions. Extensive simulations are conducted to assess the performance of these methods in comparison to a univariate trait-by-trait analysis. Our methods have been implemented in the freely available package R/qtlbim (www.qtlbim.org), which greatly facilitates the general usage of the Bayesian methodology for unraveling the genetic architecture of complex traits. |
