A density-periphery based graph clustering software developed for detection of protein complexes in interaction networks

In this paper we present a graph clustering software called DPClus, which we have developed based on a clustering algorithm. The algorithm tends to isolate densely connected regions of a graph as clusters. Users have freedom to choose two input parameters within reasonable range and thus to affect the outcome of the clustering up to certain extent. Though this software can be used for graph clustering in general but it mainly focuses on detection of protein complexes in interaction networks. The proposed software makes it possible to detect and visualize clusters of proteins in interaction networks which mostly represent molecular biological functional units. We believe that the present software can be applied not only to other biological networks but also to networks in many other applications where finding cohesive group is an agenda.