Bayesian association mapping for quantitative traits in a mixture of two populations

We introduce a novel Bayesian approach to estimate and account for populati on structure simultaneously with association mapping of multiple quantitati ve trait loci. The method is designed for an analysis of unrelated individu als from a mixture of two populations (no admixture), where the individual population memberships are unknown. In our approach, the population structu re is estimated and accounted for by using data on additional "grouping" ma rkers which are assumed to be in Hardy-Weinberg equilibrium within the popu lations but have different allele frequencies between the populations. We u se Bayesian hierarchical modeling and Markov chain Monte Carlo estimation, where we allow both population stratification and genetic heterogeneity. In our model the number of quantitative trait loci and their positions are tr eated as random variables, and we obtain their posterior distributions. Her e we select the candidate and the grouping markers based on results from a preliminary SOLAR analysis. (C) 2001 Wiley-Liss, Inc.