Protein Dihedral Angle Prediction: The State of the Art

Abstract

AbstractThe three dimensional native structure plays an important role in determining the function of a protein. However, structure determination is tedious and costly, so prediction of protein three dimensional structures is a very important as well as a challenging task in computational biophysics. Prediction of dihedral angle is particularly helpful for predicting tertiary structure of proteins as knowledge of backbone torsion angles significantly narrow down the conformational search space for tertiary structure prediction. Dihedral angles provide a detailed description of local conformation of a protein. With the advancement of machine learning and other relevant techniques, dihedral angle prediction may establish itself as a fascinating supplement to secondary structure prediction. Over the last two decades, research in this direction has led to development of several dihedral angle prediction methods. In this article we critically review available methods for protein dihedral angle prediction with an emphasis on deep learning based real value angle prediction methods. We believe this review will provide important insights into the state of the art of protein dihedral angle prediction.