An ANFIS approach to transmembrane protein prediction

This paper is concerned with transmembrane prediction analysis. Most of novel drug design requires the use of Membrane proteins. Transmembrane protein structure allows pharmaceutical industry to design new drugs based on structural layout. However, laboratory experimental structure determination by X-ray crystallography is difficult to be achieved as the hydrophobic molecules do not crystalize easily. Moreover, the sheer number of proteins demands a computational solution to transmembrane regions identifications. This research therefore presents a novel Adaptive Neural Fuzzy Inference System (ANFIS) approach to predict and analyze of membrane helices in amino acid sequences. The ANFIS technique is implemented to predict membrane helices using sliding window data capturing. The paper uses hydrophobicity and propensity to encode the datasets using the conventional one letter symbol of amino acid residues. The computer simulation results show that the offered ANFIS methodology predicts transmembrane regions with high accuracy for randomly selected proteins.