Prediction of Plasma Membrane Spanning Region and Topology Using Hidden Markov Model and Neural Network

Abstract

Unlike bacteria, which generally consist of a single intracellular compartment surrounded by a plasma membrane, a eukaryotic cell is elaborately subdivided into functionally distinct, membrane-enclosed intracellular compartments that are composed of the nucleus, mitochondria, and chloroplast. Although transmembrane spanning region and topology prediction tools are available, such software cannot distinguish plasma membrane from intracellular membrane. Moreover, the presence of signal peptide, which has information of intracellular targeting, complicates the transmembrane topology prediction because the hydrophobic composite of signal peptide is considered to be a putative transmembrane region. By immediately detecting a signal peptide and transmembrane topology in a query sequence, we can discriminate plasma membrane spanning proteins from intracellular membrane spanning proteins. Moreover, the prediction performance significantly increases when signal peptide is contained in queries. Transmembrane region prediction algorithm based on the Hidden Markov Model and ER signal peptide detection architecture for neural networks has been used for the actual implementation of the software.