Identification of functional modules in protein networks by near-clique detection

概要

論文の詳細を見る
In analysis of protein-protein interaction (PPI) networks, detection of functional module is one of the most important problems for understanding of cellular function of uncharacterized proteins. Identification of functional modules has been mainly done by searching densely connected subgraph, and some methods have been proposed to identify modules by using different criteria of densely connected subgraph. Here, we propose a new method NCMine to detect functional modules by extracting near-clique subgraphs aiming to get better identification of functional modules. We tested NCMine and other methods by using human PPI network from HPRD. When NCMine is applied to the network, it extracts about 2000 modules and 55% of them have at least one enriched GO term that is shared among members of a module. On the other hand, the percentage of GO-enriched modules extracted by other methods was lower than that of NCMine. This indicates that NCMine is superior to other methods in identification of biologically meaningful modules.
2013-03-14