Assessing Metal Ion Assignment Accuracy in Protein Data Bank Models via Elemental Spectroscopy
Edward H Snell1,2, Geoffrey W Grime3, Samuel M Webb4
1The Department of Materials Design and Innovation, SUNY University at Buffalo, 120 Bonner Hall, Buffalo, New York14260, United States.
Journal of Chemical Information and Modeling
|June 15, 2026
Summary
Accurate metal ion representation in protein structures is crucial. This study found that metals in Protein Data Bank (PDB) entries often mismatch experimental data from protein samples, indicating widespread data integrity issues.
Area of Science:
- Biophysics
- Structural Biology
- Analytical Chemistry
Background:
- Accurate metal ion representation in macromolecular structures is vital for chemical interpretation, computational modeling, and machine learning.
- The elemental identity of metals in crystallographic structures is often inferred and lacks experimental validation.
Purpose of the Study:
- To experimentally determine the elemental composition of protein samples used for metalloprotein crystal structures.
- To assess the consistency between modeled metal ions in Protein Data Bank (PDB) entries and the actual elemental content of protein samples before crystallization.
- To identify widespread data integrity issues in deposited macromolecular structures.
Main Methods:
- Utilized Particle Induced X-ray Emission (PIXE) and X-ray Fluorescence Spectroscopy (XRFS) to analyze the elemental composition of original protein materials.
- Integrated spectroscopic results with automated crystallographic validation metrics like real-space Z-difference (RSZD) analysis.
- Performed systematic rerefinement to evaluate atomic-number mismatch at metal sites.
Main Results:
- In a majority of cases, modeled metal ions in PDB entries were inconsistent with experimentally detected elements in protein samples.
- Additional metals were often present in protein samples but not represented in the corresponding structural models.
- PIXE and XRFS demonstrated strong agreement and provided complementary methods for identifying suspect metal assignments.
Conclusions:
- A widespread data integrity issue exists in deposited macromolecular structures, affecting PDB-wide entries.
- Established an experimentally corroborated link between elemental identity and crystallographic validation metrics.
- Developed a scalable approach for detecting chemically inconsistent annotations in structural databases used for computational modeling and machine learning.
