ABSTRACT
Molecular interactions are typically represented as a graph in which nodes represent molecules and edges represent molecular interactions. Identifying hidden topological structures of molecular interaction networks often unveil biologically relevant functional groups and structural complexes. We have developed a heuristic algorithm for finding cliques and quasi-cliques in protein interaction networks. As highly connected subgraphs, the identified cliques and quasi-cliques can be used to predict the function and sub-cellular localization of uncharacterized proteins as well as to abstract complex molecular interaction networks.
