Predicting Protein Subcelluar Localizations Using Weighted Euclidian Distance

Predicting subcellular localizations of proteins is very important for the determination of protein functions. In this paper, we present a K-nearest neighbor (K-NN) method for predicting the subcellular localizations of proteins in Gram-negative bacteria. The method makes predictions based on a weighted Euclidian distance computed from amino acid composition. The method achieves 81.4% accuracy in assigning proteins into five subcellular locations. Comparisons show that using the weighted Euclidian distance developed in this study can achieves better performance in predicting subcellular localization than using the standard Euclidian distance. We also compare our method with CELLO II, one of the best methods in subcellular localization prediction. The comparisons show the performances of the two methods are comparable, while our method is much simpler and faster.