BCGES seminar, 29 May 2012 at 1pm, Dr Chris Spencer & Dr Matti Pirinen, Bayesian methods for modelling effect heterogeneity in genetic association studies
1 May 2012
Venue: Room B15 – Biochemistry LT, Darwin building (UCL)
Dr Chris Spencer and Dr. Matti Pirinen, Wellcome Trust Centre for Human Genetics, Oxford University
Genome-wide association studies are aimed at detecting the effect of genetic variants on susceptibility to particular traits and diseases. When combining data from different studies or collections standard statistical methods often assume a genetic variant to have the same effect on risk. In many situations, for example when analysing data from different populations, or studies of different phenotypes, it may be beneficial to relax this assumption for the purpose of identify new associations. Moreover, it is often of interest to compare statistical models directly in order to quantify the evidence for similarities in and differences in effect between studies. We use examples from ischemic stroke, malaria susceptibility and auto-immune disease to illustrate the utility of the Bayesian approach.