Hybrid biogeography based simultaneous feature selection and prediction of N-myristoylation substrate proteins using support vector machines and random forest classifiers

Abstract

Majority of proteins undergo important post-translational modifications (PTM) that may alter physical and chemical properties of the protein and mainly their functions. Laboratory processes of determining PTM sites in proteins are laborious and expensive. On the contrary, computational approaches are far swifter and economical; and the models for prediction of PTMs can be quite accurate too. Among the PTMs, Protein N- terminal N-myristoylation by myristoyl-CoA protein N-myristoyltransferase (NMT) is an important lipid anchor modification of eukaryotic and viral proteins; occurring in about 0.5% encoded NMT substrates. Reliable recognition of myristoylation capability from the substrate amino acid sequence is useful for proteomic functional annotation projects as also in building therapeutics targeting the NMT. Using computational techniques, prediction-based models can be developed and new functions of protein substrates can be identified. In this study, we employ Biogeography based Optimization (BBO) for feature selection along with Support Vector Machines (SVM) and Random Forest for classification of N-myristoylation sequences. The simulations indicate that N-myristoylation sites can be identified with high accuracy using hybrid BBO wrappers in combination with weighted filter methods. © 2012 Springer-Verlag.