Genetic variation is currently being studied using chips mapping 1 million polymorphisms in populations of thousands of patients. While synergy among genetic factors contributes to disease, the number of combinations is enormous, presenting an urgent research challenge. Furthermore, the influences of many variants on disease risk is very small. In this project we will develop approaches to map interacting genetic combinations in a variety of common inflammatory, neurological and cardiovascular diseases.
The project will first reduce the search space of pairwise interactions from a computationally intractable thousand billion interactions. This will rely on existing methods (filtering out variants which do not even make a fractional contribution to disease on their own) and develop additional filters. The search space will also be reduced by focusing on interactions that are biologically plausible. For example, genes that regulate each-other (as predicted from known promoter motif distributions and evidence from expression patterns) and genes that are of similar function (e.g. known candidate genes for a disease) are more likely to interact to contribute to disease. We will also incorporate information from what is known about pairs of variants that interact to alter gene expression (RNA production from the DNA). We will apply these methods to a variety of diseases including Crohn's disease, autism, bipolar disorder (manic depression) and cardiovascular disease.
This project would suit a student with statistical, computational, or mathematical training and an interest in genetics.