NIM, A Novel Computational Method for Predicting Nuclear-Encoded Chloroplast Proteins

Abstract

The identification of nuclear-encoded chloroplast proteins is important for the understanding of their functions and their interaction in chloroplasts. Despite various endeavors in predicting these proteins, there is still room for developing novel computational methods for further improving the prediction accuracy. Here we developed a novel computational method called NIM based on interpolated Markov chains to predict nuclear-encoded chloroplast proteins. By testing the method on real data, we show NIM has an average sensitivity larger than 92% and an average specificity larger than 97%. Compared with the state-of-the-art methods, we demonstrate that NIM performs better or is at least comparable with them. Our study thus provides a novel and useful tool for the prediction of nuclear-encoded chloroplast proteins. 