Predict protein subnuclear location with ensemble adaboost classifier

Protein function prediction with computational method is becoming an important research field in protein science and bioinformatics. In eukaryotic cells, the knowledge of subnuclear localization is essential for understanding the life function of nucleus. In this study, A novel ensemble classifier is designed incorporating three AdaBoost classifiers to predict protein subnuclear localization. The base classifier algorithms in AdaBoost classifier is fuzzy K nearest neighbors (FKNN). Three parts amino acid pair compositions with different spaces are computed to construct features vector for representing a protein sample. Jackknife cross-validation test are used to evaluate performance of proposed with two benchmark datasets. Compared with prior works, promising results obtained indicate that the proposed method is more effective and practical. Current approach may also be used to improve the prediction quality of other protein attributes. The software written in Matlab are available freely by contacting the corresponding author.