Prediction of the disulphide bonding state of cysteines in proteins using conditional random fields

The formation of disulphide bonds between cysteines plays a major role in protein folding, structure, function and evolution. Many computational approaches have been used to predict the disulphide bonding state ofcysteines. In our work, we developed a novel method based on Conditional Random Fields (CRFs) to predict the disulphide bonding state from protein primary sequence, predicted secondary structures and predicted relative solvent accessibilities (all-state information). Our experiments obtain 84% accuracy, 88% precision and 94% recall, using all-state information. However, our results show essentially identical results when using protein sequence and predicted relative solvent accessibilities in the absence of secondary structure.