Protein Three-Dimensional Structure Validation

The wealth of structural information about proteins that is now available in the Protein Data Bank (PDB) comes at a price. When using protein structure models for modeling and drug design purposes, one must often choose between multiple, seemingly equivalent coordinate sets. Protein validation is an important step in this selection process. It is the key to assessing the quality of protein structures and thereby to the quality of all subsequent computational steps. In this chapter we shall explain the importance of structure validation for drug design and many other applications of protein structural data. Different types of validation are discussed; from the detection of simple administrative errors, to calculations of amino acid packing and hydrogen bond networks. We show that even seemingly unimportant errors can have serious implications when left undetected. Structure validation, and the large number of software tools that help in the process, has developed over time to keep up with the demands of structural biologists and drug designers. We discuss the development of the quality of protein structures since the beginning of the PDB based on a number of validation tests. Validation has a long future ahead of it because new classes of errors are still discovered frequently and many problems still await resolution. Generally, there is a trend towards better structures over time and that trend should continue. We therefore encourage the validation of protein structures before use for drug design and homology modeling, but also during the process of structure calculation and refinement in nuclear magnetic resonance (NMR) spectroscopy and x-ray crystallography.