A Penalized Least Square Method for Identifying Protein Complexes in PPI Network


What is PLSMC?

PLSMC is a novel algorithm based on a penalized least square method to detect complex in PPI network, named PLSMC. It is on the basis of a natural assumption that interacting proteins are prone to participate in same complexes. PLSMC is to minimize the distances between the interaction and co-complex of protein pairs. By means of the optimization, the propensities of proteins to complexes are determined.


Proteins do not function in isolation, but interact together to form complexes. Protein complex plays an important role in cellular activities, such as signal transduction, cell cycle, DNA transcription, protein translation and so on . Identifying protein complexes are crucial for understanding molecular mechanism in cellular activities. It is important to develop computational methods for identifying complexes, since there are limitations in wet-lab techniques. Recent developments in high-throughput technologies have produced large amount of high-quality protein-protein interaction (PPI) data that can be represented as a PPI network, an undirected graph, in which nodes denote proteins and edges are interactions between pairs of proteins. Identifying protein complexes from PPI data can be formulated as finding dense regions in PPI network.

To examine the performance, PLSMC is applied on some public yeast PPI networks, and the predicted complexes are compared with other state-of-the-art methods. The results show that PLSMC outperforms other methods on complex identification. In particular, complexes predicted by PLSMC can match real complexes with higher accuracy, and are also with higher functional homogeneity. The proposed method provides a new way to computational identifying protein complexes.


Qiguo Dai, Maozu Guo, Yingjie Guo, Xiaoyan Liu, Yang Liu and Zhixia Teng. PLSMC: A Penalized Least Square Method for Identifying Protein Complexes in Protein-Protein Interaction Network. Submitted.