Identifying Protein Complexes By Integrating Gene Expression Profiles And PPI Networks

With the completing of genome sequencing, more and more focus have been paid to proteomics. Proteins are the main taker of life activities, but they seldom work alone to carry out cellular functions while they bind together into protein complexes to perform these cellular functions. Identification of protein complexes from protein-protein interaction network (PPI Network) is very important and significant for understanding protein functions and Specific biological processes. The main works and contributions in this paper are as follows.This paper develops a novel method, named TACD, to identify protein complexes based on time course association by integrating PPI network and gene expression profiles. The method TACD, firstly constructs the dynamic PPI network according to the research on the threshold of gene expression; then uses the known clustering method to identify clusters on the PPI Network in each time course; then adopts the time course association to integrate the above clusters; finally takes the the process of filtering false positive. The method TACD is applied on PPI network and gene expression profiles of Saccharomyces cerevisiae and the experimental results show that the method TACD has better ability of purifying redundancy for the protein complexes from dynamic PPI network.According to the analysis of the inherent organization in protein complexes, this paper proposes a new method, named DPC, to identify protein complexes based on core-attachment. The protein complex contains two parts:static core consisting of always active proteins and dynamic attachments short-lived. The always active proteins refer to the proteins which are expressed in all the molecular cycle. The proposed algorithm DPC was applied on the data of Saccharomyces cerevisiae and the experimental results show that the method DPC, not only can identify more accurate cores compared with the algorithms COACH and Core-Attachment, but also outperforms CMC, MCL, SPICi and HC-PIN on the validation of matching with known protein complexes, and the predicted protein complexes of DPC have good functional enrichments.