LVQ Approach Using AA Indices for Protein Subcellular Localisation Prediction

Abstract

Knowledge of subcellular localisation of proteins is important in determining their function and involvement in different pathways. A wide variety of methods has been proposed over the recent years in order to predict the subcellular localisation of proteins, mainly based on amino acid composition or single sequence inputs. We propose a Learning Vector Quantization (LVQ) method for protein subcellular localisation prediction based on N-terminal sorting signals by using the information derived from Amino Acid (AA) index database. The LVQ approach achieved overall prediction accuracies of 84.7% for 2427 eukaryotic protein sequences on Reinhardt and Hubbard dataset and upto 86.8% on the non-plant (eukaryotes) dataset of 2738 sequences from the TargetP website, which are comparable or better than the results of existing prediction methods.