Statistical methods for the study of molecular mechanisms of disease (360G-Wellcome-091388_Z_10_Z)

Genome-wide association (GWA) studies have been very successful in pointing to genetic loci associated with risk of type 1 diabetes (T1D) and other diseases. We knew only two or three such loci for T1D when I took up my fellowship at the end of 2000 while now we have 45. The pace at which new data has been generated in the last two to three years has been such that some analysis remains to be done on existing datasets. Examples include pathway?based analyses and meta-analysis of results for multiple autoimmune diseases. But GWA studies cannot identify all disease susceptibility loci. In particular, low frequency variants with larger effects are certain to occur and may prove more valuable for further study of disease mechanisms. Advances in high-throughput sequencing and genotyping arrays promise to allow us to extend the spectrum of frequency and types of disease variants which we can identify. But statistical problems will be faced in the design and analysis of such studies. I would hope to contribute to the solution of these problems.