This laboratory has developed new process quality management methods that use all-atom contact analysis (including the hydrogens) and updated torsion-angle criteria to guide procedures for improving experimental 3D structures of proteins and nucleic acids. Preliminary studies on a set of 15 crystal structures have achieved a truly dramatic improvement by factors of 5 to 10 in the average scores for all-atom clashes, poor sidechain rotamers, """"""""flipped"""""""" Asn/Gln/His, and Ramachandran outliers, while at the same time improving fit to the data (crystallographic Rfree down by typically 1 to 3%). Quality scores on RNA backbone and on NMR structures show serious but potentially correctable problems; as in protein crystal structures, even the most careful structural biologists did not previously have the independent information added by these new methods. This project, therefore, aims to achieve the benefits of breakthrough improvements in accuracy for both x-ray and NMR structures of both proteins and nucleic acids. For protein crystal structures, these techniques will be disseminated as widely as possible, applied to the correction and re-refinement of specific structures critical either for biomedical importance or for theoretical calculations, and enhanced further by researching optimal local backbone shifts and more feasible incorporation of multiple conformations of surface side chains. For RNAs and for NMR structures, the suitable set of evaluation criteria will be completed and effective techniques for their use in structure improvements and refinement will be developed and tested. The NMR ensembles will have """"""""outlier"""""""" conformations only where the experimental data requires them, rather than where the data is insufficient to rule them out. For the very best x-ray and NMR structures, the interiors already agree well with each other and with our quality criteria; the connected approaches being pursued here should make their results much more similar than at present, even for surface loops and sidechains. The validity of such a reconciliation must be proven by better agreement with experimental data (for x-ray, lower Rfree; for NMR, fewer constraint violations and better match to new additional types of data) and ultimately by improved performance of the structures for various biomedical uses by others.
Showing the most recent 10 out of 22 publications
