Discrimination of β-Barrel Membrane Proteins Using Machine Learning Techniques

β-barrel membrane proteins (TMBs) perform a variety of functions in living organisms and these proteins contain β-strands as their membrane spanning segments. The membrane spanning segments of TMBs contain several charged and polar residues in contrast with a stretch of hydrophobic amino acid residues in transmembrane helical (TMH) proteins. Hence, most predictive schemes, which are successful in discriminating TMH proteins, fail to discriminate β-barrel membrane proteins. Discriminating β-barrel membrane proteins from other folding types of globular and membrane proteins is an important task both for identifying β-barrel membrane proteins from genomic sequences and for the successful prediction of their secondary and tertiary structures. In recent years, machine learning techniques are widely used in different areas of Bioinformatics, such as, predicting protein secondary structures, solvent accessibility, fold recognition etc. In this work we have analyzed several machine learning techniques for discriminating β-barrel membrane proteins and utilized the methods, neural networks and support vector machines based on the composition of amino acid residues, residue pair preferences and amino acid properties for discriminating TMBs. We obtained an accuracy of 94% for correctly discriminating TMBs and globular/TMH proteins, which is better than other methods in the literature.