Prediction of Protein Functional Class from Pseudo-Amino Acid Composition

Pseudo-amino acid composition approach has been successfully applied to prediction of protein subcellular localization and protein structure class and so on. In this paper, pseudo-amino acid composition is used to express a protein as a non-zero vector. The machine learning method K nearest neighbor method (K-NNA) is proposed for predicting protein functional classes from this information. The Euclidean distance and Mahalanobis distance are used to measure the similarity in K-NNA classification algorithm. Good results show that pseudo-amino acid composition of protein sequences used to predict other nature of protein can also be applied to the prediction of protein functional classes.