The importance of automatically annotating the subcellular attributes of uncharacterized proteins and its timely utilization in drug discovery is self-evident. This accurate information about protein locations in a cell facilitates in the understanding of the function of a protein and further interaction in the cellular environment. We proposed a novel GNeg-CEF approach for predicting gram-negative bacterial subcellular locations. In the proposed scheme, we exploited diversity both in feature and decision spaces. In order to exploit diversity in feature space, we used six feature extraction strategies; Amino Acid Composition (AAC), Split Amino Acid Composition (SAAC), Pseudo Amino Acid Composition (PseAAC) parallel, PseAAC Series, Dipeptide Composition (DC), and Sequential Evolution (PseEvo). Diversity in decision space is exploited using three state of the art classification models; Support Vector Machine, k-Nearest Neighbor, and Back Propagation Neural Network. First, the performance of individual ensemble classifiers for single feature extraction technique is evaluated. Next, the improved performance of the composite ensemble GNeg-CEF of all individual ensembles is investigated using majority voting scheme for gram-negative bacterial protein dataset.
展开▼
