Large-scale comparison between the diffraction-component precision indexes favors Cruickshank’s Rfree function

Abstract

This study aims to provide a first large-scale comparison between the various diffraction-component precision index (DPI) equations, assess the applicability of the parameter, and make recommendations on DPI computation. The DPI estimates the average accuracy of the atomic coordinates obtained by the structural refinement of protein diffraction data, with application in crystallography and cheminformatics. Although, Cruickshank and Blow proposed DPI equations based on R and Rfree in order to calculate DPI values, which remain scarcely employed in the quality assessment of the Protein Data Base (PDB) files, due to the unclear data extraction protocols (to assign variables), the complex equations, the lack of extensive applicability studies and the limited access to automated computations. In order to address these shortcomings, the entire RCSB PDB database was evaluated using Cruickshank’s and Blow’s R and Rfree DPI variations. Computations of 143070 X-ray structures indicate that Rfree-based DPI equations apply to 30 % more protein structures compared to R-based DPI equations, with Cruickshank Rfree-based DPI (CRF) exceeding the number of successful Blow’s Rfree-based DPI (BRF) computations. Although our results indicate that, in general, the resolutions < 2 Å assure consistency among the various DPIs computations (differences <0.05 Å), we recommend the use of CRF DPI because of its wider applicability.

Keywords

• cheminformatics
• drug discovery
• protein structure
• docking
• crystallography