consensus decision for protein structure classification共识决定蛋白质结构分类.pdfVIP

Journal of Intelligent Learning Systems and Applications, 2012, 4, 216-222 / 10.4236/jilsa.2012.43022 Published Online August 2012 (http://www.SciRP.org/journal/jilsa) Consensus Decision for Protein Structure Classification Khaddouja Boujenfa, Mohamed Limam Institut Supérieur de Gestion, LARODEC Laboratory, University of Tunis, Tunis, Tunisia. Email: Khaddouja.Boujenfa@isg.rnu.tn, Mohamed.Limam@isg.rnu.tn Received January 2nd st th , 2012; revised June 1 , 2012; accepted June 8 , 2012 ABSTRACT The fundamental aim of protein classification is to recognize the family of a given protein and determine its biological function. In the literature, the most common approaches are based on sequence or structure similarity comparisons. Other methods use evolutionary distances between proteins. In order to increase classification performance, this work proposes a novel method, namely Consensus, which combines the decisions of several sequence and structure com- parison tools to classify a given structure. Additionally, Consensus uses the evolutionary information of the compared structures. Our method is tested on three databases and evaluated based on different criteria. Performance evaluation of our method shows that it outperforms the different classifiers used separately and gives higher classification perform- ance than a free-alignment method, namely ProtClass. Keywords: Bioinformatics Databases; Classification; Mining Methods and Algorithms; Similarity Measures 1. Introduction curate classification of proteins. The knowledge of the 3D structure of a given protein is useful to improve its Protein classification is one of the fundamental problems