A post-hoc genome-wide association study using matched samples

Abstract

Genome-wide association studies have identified many causal candidate loci associated with common complex phenotypes, such as type-2 diabetes and obesity. However, most of these studies have been drawn from non-randomised case/control experiments, where the units exposed to one group generally differ from those exposed to the other group. The aim of this study was to address the issues arising from non-randomised case/control experiments. In order to achieve this, we have proposed a post-hoc association analysis using subsets of samples selected by the proposed matching technique. This method was applied to two different binary traits, type-2 diabetes and obesity, in Korean subjects. It identified nine and two additional variants for type-2 diabetes and obesity, respectively, which were not identified using the total dataset. Our study demonstrates that the proposed a post-hoc genome-wide association analysis can determine additional candidate causal variants responsible for common complex phenotypes