Multi-Dimensional Scaling based grouping of known complexes and intelligent protein complex detection. (June 2018)

• Record Type: Journal Article
• Authors: Rehman, Zia ur; Idris, Adnan; Khan, Asifullah
• Is Part Of: Computational biology and chemistry. Volume 74(2018)
• Page Start: 149
• Page End: 156
• Publication Date: 2018-06
• Subjects: Protein complex detection -- Genetic algorithm -- Multi-dimensional scaling -- Bagging based ensemble; Chemistry -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; Biochemistry -- Data processing; Biology -- Data processing; Molecular biology -- Data processing; Periodicals; Electronic journals; 542.85
• DOI: 10.1016/j.compbiolchem.2018.03.023
• Languages: English
• ISSNs: 1476-9271

Abstract

Abstract: Protein–Protein Interactions (PPI) play a vital role in cellular processes and are formed because of thousands of interactions among proteins. Advancements in proteomics technologies have resulted in huge PPI datasets that need to be systematically analyzed. Protein complexes are the locally dense regions in PPI networks, which extend important role in metabolic pathways and gene regulation. In this work, a novel two-phase protein complex detection and grouping mechanism is proposed. In the first phase, topological and biological features are extracted for each complex, and prediction performance is investigated using Bagging based Ensemble classifier (PCD-BEns). Performance evaluation through cross validation shows improvement in comparison to CDIP, MCode, CFinder and PLSMC methods Second phase employs Multi-Dimensional Scaling (MDS) for the grouping of known complexes by exploring inter complex relations. It is experimentally observed that the combination of topological and biological features in the proposed approach has greatly enhanced prediction performance for protein complex detection, which may help to understand various biological processes, whereas application of MDS based exploration may assist in grouping potentially similar complexes.