## Wednesday, October 9, 2013

### Physics-based method to validate and repair flaws in protein structures

Osvaldo A. Martin, Yelena A. Arnautovac, Alejandro A. Icazattia, Harold A. Scheraga, and Jorge A. Vila. Proceedings of the National Academy of Sciences USA, early edition
Contributed by +Jan Jensen

Scheraga, Vila, and co-workers have developed a method, and associated web-server CheShift-2, to validate and repair errors in the side chain conformations of amino acids based on the difference in measured and computed $^{13}\text{C}^\alpha$ and $^{13}\text{C}^\beta$ chemical shifts.

This is a basically a two-step process: In the first step the $^{13}\text{C}^\alpha$ and $^{13}\text{C}^\beta$ chemical shifts are predicted using the CheShift method based on an input protein structure.  CheShift is based on a database of roughly 1,200,000 DFT-predicted $^{13}\text{C}^\alpha$ and $^{13}\text{C}^\beta$ values. The extension to $^{13}\text{C}^\beta$ chemical shifts is part of the presented work.  The predicted values are compared to the corresponding experimental values and amino acids for which the deviation is large are identified.  For a structural ensemble average chemical shifts are used to identify problematic residues.

In the second step, $\chi1$ and $\chi2$ angles that give predicted chemical shifts in better agreement with experiment are identified and the angles that result in the least steric clash are selected.

The method is applied to 42 protein structural ensembles determined by NMR and the repaired structures are compared to the corresponding X-ray structures.  For ca 90% of the chosen residues at least one of the chosen ($\chi1$, $\chi2$) angles is seen in the corresponding X-ray structure.