Identification of Important Proteins in Protein Interaction Network Based on SVM

• Published in: Applied Mechanics and Materials Vol. 644-650; pp. 5202 - 5206
• Main Authors: Luo, Wan Cheng, Zha, Yan Li
• Format: Journal Article
• Language: English
• Published: Zurich Trans Tech Publications Ltd 01.09.2014
• Subjects: Topological Feature; Protein Interaction Network; Support Vector Machine (SVM); Essential Protein
• ISBN: 9783038352464; 3038352462
• ISSN: 1660-9336; 1662-7482; 1662-7482
• DOI: 10.4028/www.scientific.net/AMM.644-650.5202

Importance of proteins are different to perform functions of cells in living organisms according to the relevant experiment results, and more essential proteins is the most important kind of proteins. There are recently many computational approaches proposed to predict essential proteins in network level through network topologies combined with biological information of proteins. However it is still hard to identify them because of limitations of topological centralities and bioinformatic sources. And more it is the challenge is to perform better with less resources. Therefore in this paper, we first examine the correlation between common topological centralities and essential proteins and choose a few particular centralities, and then to build a SVM model, names as TC-SVM, for predicting the essential proteins. The new method has been applied to a yeast protein interaction networks, which are obtained from the BioGRID database. The ten folds experimental results show that the performance of predicting essential proteins by TC-SVM is excellent.