Protein Secondary Structure Reduction Methods Significantly Affect Prediction Accuracy

Abstract

Protein secondary structure prediction is a critical step in determining the 3D structure of a protein which in turns determines the function of a protein. Five methods that reduce the DSSP (Dictionary of Secondary Structure of Proteins) secondary structures from eight classes into three classes are attempted in this work. A protein secondary structure classifier from amino acid sequences has been used to evaluate the five reduction methods under the same hardware, platforms, and environments to allow stringent and reliable comparison between the methods and then arrive at clear conclusions. Researchers in the filed of protein secondary structure prediction use typical three states of secondary structures, namely: alpha helices (H) beta strands (E), and coils (C). The adopted reduction schemes from the DSSP eight classes (states) to the three classes (H, E, and C) are usually performed by using one of the five assignments or reduction methods implemented in this research. This paper explains and discusses the effect of these reduction methods on the prediction accuracy.