Validation

From Media Wiki

Validation is the process of assessing the quality of an atomic model. Due to the uncertainties of Model building, atomic models are not always completely reliable. Validation can be performed against the experimental observations, or to a more limited extent on the basis of the features of the model itself.

A good model:

• Makes statistical sense:
  • The atomic model provides a statistically reasonable explanation of the experimental data (R-factor, etc.).
  • No meaningful difference map peaks.

• Makes chemical sense:
  • Model geometry is consistent with restraints, which are based on chemical knowledge.
  • Ramachandran plot has less than 1% outliers.
  • No chemical clashes.

• Makes biological sense:
  • Residues in a sensible chemical environment.
  • Is consistent with external biochemical observations (e.g. accessibility of the active site, etc.).

Paul Emsley, Kevin Cowtan