Summary

Macromolecular models are hypotheses about atomic arrangements and must be validated against experimental data and fundamental chemical/physical principles.

Global quality assessment gives you an overall picture of a model’s reliability. It’s a crucial first step before looking at specific regions.

Look for these key global metrics, especially in the PDB Validation Slider and the PDB Validation Report:

• Resolution (X-ray, cryo-EM): Indicates the level of detail. Lower Ångström values mean higher resolution and generally more accurate atomic positions.
• R-free (X-ray): A key measure of how well the model fits unseen experimental data, helping detect overfitting. Lower values are better.
• Clashscore: Quantifies physically improbable atomic overlaps (atoms too close together). Lower values are better.
• Ramachandran outliers: Percentage of protein backbone residues with unfavourable angles. Lower percentages are better and can flag potential errors.
• Sidechain outliers: Percentage of residues with unusual sidechain arrangements. Lower percentages are better.
• RSRZ outliers (X-ray) / Residue inclusion outliers (cryo-EM): Percentage of residues that don’t fit well into the experimental density map. Lower percentages are better.
• Q-score (high-res cryo-EM): Measures how well atoms/residues are resolved in the EM map. Higher values indicate a better fit and more reliability in that area.
• Ligand RSCC and RSR (with ligands): Assess how well a ligand fits the experimental density in its binding site. High RSCC (near 1.0) and low RSR (near 0) indicate a good fit. If a ligand has an RSCC value below 0.8 and/or an RSR value above 0.4, it is highlighted in the validation report as a potential issue. This indicates that the experimental data may not fully support the modelled ligand, and its interpretation should be approached with caution.
• Assignment completeness (NMR): Percentage of expected chemical shifts that have been assigned. Higher percentages are better.
• Restraint violations (NMR): Number and magnitude of distance and dihedral angle restraints violated by the model. Lower numbers and magnitudes are better, especially for “consistently violated” restraints.

Use the PDB Validation Slider for a quick comparison of these metrics to other structures. The PDB Validation Report (PDF) provides a detailed summary on its first page and visual chain-by-chain summaries that highlight regions with geometric or density fit issues.

No single global metric is perfect; global scores can sometimes hide problems in specific regions of the model. Always consider the resolution, the experimental method, and the biological question you are trying to answer when evaluating quality. The original scientific publication can provide valuable context.