Hello Charles and others, I have a question regarding the usage of RSCA as validation dataset and hoping you could shed some light on this.
I have followed the details presented in "Validation of Protein Structure from Anisotropic Carbonyl Chemical Shifts in a Dilute Liquid Crystalline Phase" in Jacs 1998, 120, 6836 which used chemical shift anisotropies as validation data for 1d3z nmr structure of Ubiquitin and also the recent paper introducing 2MJB. The chemical shift tensor orientation for Carbonyl atoms are dependent upon the beta angle it makes to the CN bond vector while N and HN depend upon the bond vector orientation of NH. (Jacs 2000, 122, 10143). Since RDC's also contain information about bond vector orientations, any structure/ensemble that is optimized with CN, NH RDC data could also optimize the RCSA.This along with the observation that alignment tensors of Ubiquitin are very similar (they take one of the 2 orientations in my calculations and one of your papers), makes RCSA a very weak validation dataset. Am I missing something in my observations? How can one justify RSCA as cross-validation data? I have observed one small glitch in the documentation of XPLOR-NIH, in varTensorTools. normalizedScalarProduct(t1,t2) : t1, t2 should be the tensor objects instead of RDCpot. ( http://nmr.cit.nih.gov/xplor-nih/doc/current/python/ref/varTensorTools.html). Sorry if it was obvious. Thanks Santhosh
_______________________________________________ Xplor-nih mailing list [email protected] https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
