Computer Algorithms for Discriminating Protein Folds and Predicting Protein Folding Rates Based on Contact Information

Inter-residue interactions play an important role in governing the folding and stability of protein structures. In this work, we have analyzed the contacts between amino acid residues in different folding types of globular proteins and various ranges of folding rates. Based on amino acid contacts a novel parameter, multiple contact index has been developed for understanding protein folding rates. We have computed short, medium and long-range contacts in proteins of different folding types and rates from their sequences and structures. Utilizing the information, we have developed models for recognizing protein folds and predicting their folding rates, using machine learning techniques. Our methods showed an accuracy of 55% for discriminating 1612 globular proteins belonging to 30 different folds and an accuracy of 96% to distinguish the fast and slow folding proteins using 5-fold cross-validation method.