A novel technique of feature extraction with dual similarity measures for protein sequence classification

Abstract

In this article, a novel approach for extracting features from protein sequences is proposed. This approach extracts only six features corresponding to each protein sequence. These features are computed by globally considering the probabilities of occurrences of the amino acids in different positions within the superfamily which locally belongs to the six exchange groups. Then, these features are used as an input to the Neural Network formed by Boolean-Like Training Algorithm (BLTA). The BLTA is used to classify the protein sequences obtained from the Protein Information Resource (PIR). To investigate the efficacy of proposed feature extraction approach, the experimentation is performed on two superfamilies, namely Ras and Globin using tenfold cross validation. The highest Classification Accuracy achieved is 100.00±00.00 with Computational Time 170.49±70.87 (s) are remarkably better in comparison to the Classification Accuracies and Computational Time achieved by Mansouri, Bandyopadhyay and Wang. The experimental results demonstrate that the proposed approach extracts the most significant and lesser number of features for each protein sequence due to which it results in considerably potential improvement in Classification Accuracy and takes less Computational Time in comparison with other well-known feature extraction approaches. © 2015 The Authors. Published by Elsevier B.V.