Accurate Discrimination of Outer Membrane Proteins Using Secondary Structure Element Alignment and Support Vector Machine

Abstract

Outer membrane proteins (OMPs) play critical roles in many cellular processes and discriminating OMPs from other types of proteins is very important for OMPs identification in bacterial genomic proteins. In this study, a method SSEA_SVM is developed using secondary structure element alignment and support vector machine. Moreover, a novel kernel function is designed to utilize secondary structure information in the support vector machine classifier. A benchmark dataset, which consists of 208 OMPs, 673 globular proteins, and 206 α-helical membrane proteins, is used to evaluate the performance of SSEA_SVM. A high accuracy of 97.7% with 0.926 MCC is achieved while SSEA_SVM is applied to discriminating OMPs and non-OMPs. In comparison with existing methods in the literature, SSEA_SVM is also highly competitive. We suggest that SSEA_SVM is a much more promising method to identify OMPs in genomic proteins. A web server that implements SSEA_SVM is freely available at http://bioinfo.tmmu.edu.cn/SSEA_SVM/.