Protein fold recognition using n-Gram strict position specific scoring matrix and structural based feature extraction technique

Abstract

The decoding of tertiary structure of a protein is proved to be a difficult and important task in the study of biological science. In last few decades initiative has been taken care of predicting the tertiary structure by using different feature extracting techniques to discover the significant information from protein primary sequences and utilizing suitable classifiers to recognize different fold of unknown protein sequences. But in spite of several feature extraction techniques the limited accuracy has been achieved. In this study, we proposed a new feature extraction technique to discover the score matrix using evolutionary and structural information from primary sequence and utilize support vector machine classifier to classify protein folding. The proposed technique is compared with the recent feature extraction techniques such as; tri-gram, bi-gram, PSSM, and n-form with strict position specific scoring matrix achieved ~3.4-11.4% more accuracy.