CPSNF: Classification of Protein Structure with Novel Features

Abstract

Protein structures play key roles in many fields of biology. However, identification of protein structural types from protein sequences seems to be a challenge issue. In this study, several novel reconstructed features have been proposed and employed to be the features to deal with the machine learning issue. So as to demonstrate the performance of these features, 10-fold has been utilized in two benchmark datasets, including 1189 and 25PDB. Our proposed features can be effective to deal with the four types of protein tertiary structure than other art-of-the-state methods.