Protein Tertiary Structure Classification based on its Physicochemical property using Neural Network and KPCA-SVM: A Comparative Study

Proteins are one of the most important molecules in living organisms so they play a vital structural role in the cells of living organism. They are constructed of several polypeptide chains of amino acids, which fold into complex tertiary Structure. The knowledge of the protein function is directly dependent on its three dimensional (tertiary) structure. The Physicochemical properties of proteins always guide to determine the quality of the protein tertiary structure. Therefore it has been rigorously used to distinguish native or native like structure from other predicted structure. The experiments were conducted on the CASP dataset to classify RMSD (target class) near to native protein tertiary structure or not. Kernel principal component analysis (KPCA) is used for feature extraction since it performs better than PCA on protein tertiary structure dataset due to their nonlinear structures. The proposed model compare with neural network classification method. The experiments conducted shows that support vector machine combined with KPCA feature extraction performs better than neural network classifier. More than our results show better performance in Gaussian KPCA feature extraction with respect to other kernels.