American Statistical Association New York City Metropolitan Area Chapter Mailman School of Public Health Columbia University Department of Biostatistics Colloquium

Bayesian methods have been proposed for the analysis of genetic association studies (Stephens and Balding, 2009, “Nature Reviews Genetics”). The approximate Bayes factors (ABFs) of Wakefield (2007, AJHG; 2009, Genet Epidemiol) have closed forms and are easy to use for genome-wide association studies (GWAS). In ABFs, the data in the conventional Bayes factors are replaced by the maximum likelihood estimates of the parameter of interest. The ABFs require a known genetic model, e.g., the additive model. In this talk, we present results of using ABFs incorporating Hardy-Weinberg disequilibrium and genetic model uncertainty. The proposed methods are more robust than the existing ones as shown in simulation studies. Applications to real GWAS data from Wellcome Trust Case-Control Consortium are used to demonstrate the usefulness of our approaches.

Originally from Shanghai, China, Dr. Zheng came to the U.S. in 1994 and received his PhD in statistics from George Washington University in 2000. Since then, he has been working as a mathematical statistician at the Office of Biostatistics Research, National Heart, Lung and Blood Institute. His research interests include inference using order statistics, legal statistics, robust methods, and statistical genetics. Dr. Zheng has over 80 publications in statistical methods and statistical applications since he received his PhD.