Protein–Ligand Interactions: Knowledge‐Based Methods
J. Willem M. Nissink, Robin Taylor · Encyclopedia of Computational Chemistry · 1998
Abstract Crystallographic data provide a rich source of information on nonbonded interactions for use in drug design. Small‐molecule crystal data are available in abundance in the Cambridge Structural Database and can be exploited to assess preferences in intermolecular interactions. Likewise, the Protein Data Bank offers similar information at the macromolecular level, though usually at a lower crystallographic resolution. Such information can be used to understand preferred intermolecular interactions of common functional groups and can be applied to predict prevalent intermolecular interactions in protein‐binding sites. We discuss methodological and data‐related problems that arise in applying this knowledge toin vivosituations and describe a knowledge‐based algorithm that predicts favorable group interactions within protein‐binding sites. Detailed validation results are summarized and an example of the application of the method to neuraminidase is given.