Abstract A Working Group consisting of the co-authors of this paper was established in 2020 to re-evaluate the standard valence geometry used for the validation of nucleic acid structure models in the Protein Data Bank (PDB). This Working Group re-examined the dependence of Cambridge Structural Database (CSD) derived targets on base and sugar type, sugar pucker, and phosphate and glycosidic conformation, before comparing those targets with the geometry of a quality-filtered reference set of nucleic acid crystal structural models held in the PDB. This revealed that the valence bond and angle mean values are close to the CSD targets, but many parameters have highly non-Gaussian or even multimodal distributions. One explanation is the inconsistency of restraints used over time and by different refinement programs. The Working Group recommends a new validation scheme for use by the PDB. For this purpose, we have developed a new three-tier scale for outlier detection—graded as Preferred, Allowed, and Of Concern intervals—based on a combination of quality-curated reference data from the CSD and the PDB. The proposed approach to validation should lead to improved nucleic acid models in (future) PDB-deposited macromolecular structures.
Černý et al. (Thu,) studied this question.