Local quality assessment

In the previous section, we explored global metrics that give you an overall sense of a model’s quality. However, a good global score doesn’t guarantee that every part of the model is equally reliable. Flexible regions, surface loops, or specific binding pockets might have lower accuracy than the more rigid core of the molecule. This is why local quality assessment is essential; it allows you to evaluate the reliability of specific regions relevant to your research question.

Different biological questions require different levels of model accuracy. For example, if you are only interested in the overall fold of a protein or how two large domains are arranged, a lower-resolution model with some local inaccuracies might be sufficient. However, if you are studying how a ligand binds to a specific pocket, designing mutations in an active site, or analysing the dynamics of a flexible loop, you need to be confident in the accuracy of the model in that specific region.

Local quality assessment provides the tools to determine if the model details are trustworthy for your intended use.

In this section, we will focus on evaluating specific regions of the structure, primarily using visual inspection with molecular graphics software and by examining local indicators derived from experimental data and stereochemical principles.

By the end of this section, you will be able to:

• Use visual inspection in the molecular viewer, Mol*, to assess the quality of specific regions in a structure.
• Interpret electron density maps (for X-ray and cryo-EM) to evaluate how well the model fits the experimental data at a local level.
• Identify common indicators of local modelling problems, such as poor fit to density, stereochemical outliers, and atomic clashes.
• Understand how factors like occupancy and multiple conformations affect local model reliability.
• Assess the quality of bound ligands, which is particularly important for drug discovery studies.