Bioinformatics: Integrating physical interaction networks in the analysis of Complex Diseases

The traditional genome-wide association (GWA) studies have been largely unsuccessful for complex diseases, like cardiovascular disease. They often fail to replicate their results, and the DNA variants that have been found have a low effect on the disease. It is widely believed that this is due to non-additive genetic interactions, where the disease depends on the variance at two or more loci. Considering the vast number of possible genetic interactions, the traditional GWA methodology is not viable; as a consequence most studies ignore genetic interactions.

This project aims to develop methods that allow GWA studies to take into account a subset of all possible genetic interactions. The main idea is to restrict the genetic interactions with the help of physical interaction networks. By doing this we can reduce the number of statistical tests, and therefore reducing the accompanying noise. Although we might miss some interactions we believe that this is a good compromise between checking all interactions and those where one locus has a strong main effect.

The idea is based on the discovery that genetic interactions seem to be more common between protein complexes than within. This gives rise to the hypothesis that diseases are caused by disturbing proteins in different complexes, rather than proteins in the same complex. Under this hypothesis we can then reduce the number of potentially interacting loci by only considering loci from different protein complexes.

We will apply this method on data from cardiovascular disease.