Protein Secondary Structure Prediction Accuracy versus Reduction Methods

Predicting protein secondary structure is a key step in determining the 3D structure of a protein that determines its function. The Dictionary of Secondary Structure of Proteins (DSSP) uses eight classes to describe a protein. The DSSP database is a database of secondary structure assignments for all protein entries in the Protein Data Bank (PDB) with an algorithm designed to standardize these secondary structure assignments. Five methods that reduce theses eight classes into the adopted three classes: alpha helices (H) beta strands (E), and coils (C) are implemented in this research. A protein secondary structure classifier (NN-GORV-II) has been used to evaluate the five reduction methods under the same hardware, platforms, and environment to allow stringent and reliable comparison of these methods and then arrive at a clear conclusion. This paper explains and discusses the effect of these reduction methods on the prediction accuracy and quality.