Accounting for Disease Model Uncertainty in Mapping Heterogeneous Traits: A Bayesian Model Averaging Approach

• Activity Number: 520
• Type: Contributed
• Date/Time: Wednesday, August 5, 2009 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Swati Biswas*+
• Companies: University of North Texas Health Science Center
• Keywords: statistical genetics ; genetic epidemiology ; linkage analysis ; Reversible Jump MCMC ; complex traits

Abstract

Locus heterogeneity presents a major challenge in mapping genes. Recently, a Bayesian approach has been proposed to account for variable rates of heterogeneity across families. Its application, however, requires specification of disease model. To address this limitation, we propose a Bayesian Model Averaging (BMA) approach. We implement BMA by considering a finite number of disease models and treating "model" as an unknown parameter. Our simulations under single-locus disease models show that BMA retains at least 80% of the power that is obtained by analyzing under the true disease model. Next we investigate the properties of BMA when the true model is extremely complex consisting of multiple interacting loci by analyzing all 100 replicates of Genetic Analysis Workshop 13 simulated data. Similar results are obtained illustrating the effectiveness of BMA in mapping heterogeneous traits.