A hybrid approach to combine HMM and SVM methods for the prediction of the transmembrane spanning region

Abstract

Transmembrane proteins are the primary targets for the development of new drugs, and a number of algorithms that predict transmembrane topology are publicly available on the Web. In this paper, we present a novel approach using both SVM and HMM methods and we demonstrate that our system outperform the previous systems which only use either HMM methods or SVM methods alone. © Springer-Verlag Berlin Heidelberg 2005.