Sir

As discussed in Editorials in Nature and Nature Structural Biology1,2, authors submitting research papers that describe molecular structures to Nature journals are required upon request to provide structural coordinates for reviewers to assess the quality of the work. But in the competitive field of structural biology, full disclosure of coordinates to anonymous peer review does raise serious concerns.

Nonetheless, the local nature of structure model quality — about which commonly provided global indicators such as R and Rfree reveal no detail3 — in fact requires reviewers to be able to access and assess the quality in relevant areas of the structure (in particular around sites with bound ligands), which is not possible without disclosure of coordinates and structure factors.

A solution to the conflict is the real-space correlation coefficient (RSCC) plot, introduced in 1990 (ref. 4). Such plots depict, residue by residue, the fit of the model and ligands to the electron density. Weak correlation indicates poor fit to electron density, suggesting genuine absence of ordered regions or building errors. Low real-space correlation is thus a general indication of lack of reliable information in that part of the structure.

RSCC plots for validation of deposited and released structures are available through the Electron Density Server EDS (http://eds.bmc.uu.se/). In addition, EDS plots are automatically generated as a part of the validation procedure during coordinate submission to the Protein Data Bank (PDB) at the European Bioinformatics Institute (EBI-MSD). Similar RSCC plots are returned by the program SFCHECK upon deposition to the US-based RCSB-PDB site.

We suggest that RSCC plots should be submitted with manuscripts to help the reviewers and users to assess the quality of structure models. Generated as a part of validation during structure deposition, these plots can be produced without any additional work by authors. The plots can be provided with the manuscript or as supplemental material to convince reviewers of the model quality in critical areas, without forcing authors to reveal coordinates and structure factors prematurely.

In view of the increasing frequency of publication of exciting and important structures of protein–ligand complexes, which depend crucially on the validity of the interpretation of ligand–protein interactions, the time is now appropriate to consider our proposal for mandatory provision of RSCC plots when submitting manuscripts to journals. Inspection of real-space correlation plots substantially lowers the risk of over-interpretation of poor local electron density, enhances information available for review, and provides authors with a strong means to demonstrate the quality of their structures without compromising their original data.