Abstract for presentation at 11th International Congress of Human Genetics

Genome-wide association studies have come to the forefront of statistical genetics. A number of analysis approaches for such studies have been suggested but have not yet proven successful. Van Steen et al (2005) suggested a testing strategy that allows the same data set to be used for the screening step and the replication step, bypassing the multiple testing problem and minimizing the effects of study heterogeneity. The approach has been successfully applied to identify a novel SNP for BMI that replicated in 5 additional studies (Herbert et al, in press). One limitation of the testing strategy by Van Steen et al (2005), however, is that linkage information is ignored, which, in practice, will lead to a substantial reduction in power. Our new methodology incorporates linkage information and can use the same data set for both the screening step and the replication step. The approach enables a more powerful analysis of genome-wide association studies in order to detect genetic factors for complex diseases. We assess the power of our approach by a series of simulation studies at a genome-wide level and show that the power gains are of practical relevance. We illustrate the approach by an application to an asthma study.