Finding groups of proteins that interacts to understand disease

Developed methods for finding communities in social networks constructed using uncertain information.

Complex networks is a mathematical model of now different agents interacts with each other. One such network is a social network, which e.g., could consist of your Facebook friends. Two of your friends would be connected if they are friends on Facebook, i.e., if they are mutual friends. This information is usually presented as a so-called graph. One interesting problem is to find sub-groups of your friends corresponding to e.g., your family, classmates from University and acquaintances from different organisations. The key observation here is that these groups often are densely interconnected, i.e., all of your classmates are probably friends with each other on Facebook and therefore form a so-called cluster.

Finding communities have important applications in finding out which proteins interact with each other in cells, to find hidden criminal networks and to construct food webs (maps of which animals that feed on each other) to better understand the ecological system. This project aims to develop methods that can find these clusters of communities from noisy and imperfect observations. The estimates of the clusters therefore need to be accompanied by an uncertainty quantification and be presented to the user at the same time to facilitate better analysis and decision making.

Image is used under Creative Commons with credits to Simon Cockell on FlickR.