Importance of improving scoring methods in predicting protein free-energy changes

Improving the accuracy of protein structure prediction or sampling as well as the scoring function is the central problem in the current computational study of protein structure and function. Recently, a protocol in the protein design software ROSETTA, called “backrub”, was developed to sample protein backbone flexibility and shown to be able to improve the prediction accuracy of protein structural changes upon mutagenesis. Therefore, it should be very interesting to see if the improvement on structure sampling could as well improve the scoring results such as protein free-energy changes when mutations are introduced. In this paper, a protein database is created to evaluate the effect of this backrub protocol on the prediction of protein free-energy changes upon mutagenesis. The results showed that the improvement on the accuracy of protein structure prediction or sampling alone failed to improve the prediction of protein free-energy changes, and suggested that the further improvement in the current scoring function is the bottleneck.