Specific aspects of local assessment

Several factors influence the local quality and reliability of a model and require specific considerations during local assessment. Beyond visual inspection of density, other factors and specific metrics help assess local quality:

Stereochemical Outliers (Local geometry)

While we discussed global stereochemical metrics (like Ramachandran and sidechain outliers, clashscore) in the Section “Metrics based on stereochemistry“, these are calculated by summing up local issues. In a local assessment, you examine the specific residues or atoms flagged as outliers in the Validation Report to understand the nature and location of the problem.

• Ramachandran outliers: identify the specific residues that fall in disallowed regions. Are they in functionally important areas? Is there experimental data supporting this unusual conformation (e.g., clear electron density)? The PDB Validation Report’s “Residue-property plots” and “Model quality: Torsion angles” sections will pinpoint these specific residues.
• Sidechain outliers (rotamers): examine the side chains flagged as outliers. Do they fit the density? Do they clash with neighbouring atoms? The Validation Report’s “Model quality: Torsion angles” section provides lists of these.
• Bond length and angle outliers: the Validation Report lists specific bonds or angles that deviate significantly from ideal values. While minor deviations can occur, large deviations in critical regions should be viewed with suspicion, especially if not supported by density.

Clashes

Two atoms cannot occupy the same space. Not all clashes are created equal. Small, minor clashes might occur due to limitations in the resolution of the data or inherent flexibility in certain regions of the molecule. Sometimes, minor clashes are the result of poor refinement parameterisation. However, large, significant clashes, especially those in important regions, are more concerning.

Local assessment involves identifying where clashes occur in the 3D model. The Clashscore gives a global number, but the Validation Report and molecular viewers can show you the location of specific atomic clashes. Are there clashes in or near your region of interest (e.g., active site, protein-protein interface)? Significant clashes often indicate errors in local modelling, especially in areas with poor density.

Occupancy and multiple conformations

Biological macromolecules are dynamic. In some cases, a part of a molecule (like a side chain, a loop or a ligand) might exist in more than one distinct conformation or position within the sample. Or a specific chemical entity (like a bound metal ion) might only be present in a fraction of the molecules.

This dynamic behaviour or particle presence is represented in structural models using occupancy.

• An occupancy of 1 means the atom is modelled as being in that position in 100% of the molecules.
• An occupancy of less than 1 indicates that the atom or group is present in that position in only a fraction of the molecules.

When a region adopts two or more distinct conformations, these can sometimes be modelled as multiple conformations, each with an occupancy summing up to 1 (or less than 1 if the region is also partially absent).

Visually, regions with partial occupancy or multiple conformations may have weaker or split electron density that supports more than one position. The model will show the alternative positions modelled. High occupancy and clear, well-defined density for a single conformation generally indicate higher confidence in that local model. Regions with low occupancy, ambiguous density, or modelled multiple conformations might be less certain.

Often, in higher-resolution crystal structures, you might see residues, particularly side chains, modelled in multiple discrete conformations. These are typically indicated in the PDB file by alternative location identifiers (e.g., ‘A’ and ‘B’) preceding the residue name for atoms belonging to different conformations. Each of these conformations will have a partial occupancy, reflecting the fraction of molecules in the crystal that adopt that specific arrangement.

Why this matters: If a part of your molecule of interest (e.g., a ligand, a catalytic residue) has a low occupancy or is modelled with multiple conformations that are not clearly resolved, it means its precise position or presence is less certain. This can impact conclusions about binding affinity, specific interactions, or functional mechanisms.

Note that it will not be possible to distinguish between a heavy atom (an atom with more electrons) with partial occupancy and a lighter atom at full occupancy. This can lead to errors in atom assignments, particularly in metals; therefore, additional experimental data (such as mass spectrometry or x-ray fluorescence) and analysis of the coordination state are often required to produce an atomic assignment.