Other important considerations for global assessment

In experimental structure determination (particularly X-ray and cryo-EM), the process of building and refining a model can introduce “model bias”. This means the resulting model might reflect the a priori knowledge or assumptions used during refinement (e.g., ideal bond lengths, preferred angles) more strongly than the experimental data itself, specifically if the data is weak or at a lower resolution. Global validation metrics, such as R-free and RSRZ, help identify potential areas where the model may be biased away from the actual experimental data.

The reported resolution is an average for the entire dataset or map. Some parts of the structure might be much better resolved than others. Global metrics give an average, but a local assessment is needed to check regions of specific interest

While validation metrics allow comparison of structures determined by the same method and a similar resolution, comparing metrics directly between structures from different methods (e.g. a high-resolution X-ray structure vs. a medium-resolution cryo-EM structure) requires caution. The strengths, weaknesses, and the nature of the data are different.

No single metric tells the whole story; always consider the biological question you are asking and the experimental method used. What might be acceptable quality for one type of study (e.g., overall shape) might be insufficient for another (e.g., drug design targeting a specific pocket).

Always refer to the original scientific publication associated with the PDB entry. The paper provides essential context about the biological question, the experimental conditions, the methods used, additional validation data (e.g., biochemical assays, spectroscopy), and the authors’ own assessment and interpretation of the structure. This is crucial for understanding the limitations and strengths of the model in its biological context and for interpreting global metrics in the proper context.