Molecular docking is the prediction of the structure of a complex of two molecules. Usually, one of the molecules is a protein or polypeptide; the other is typically smaller, although it may also be a protein.

Docking is used:

• To locate the catalytic sites or allosteric binding sites of enzymes

• To study the geometry and strength of the interaction of an enzyme with a substrate or inhibitor

• To develop leads for new targeted pharmaceuticals, by screening "in silico" rather than in experimental animals

• To understand the interaction between two polypeptide chains that may lead to the formation of an active enzyme

Reproducing the conformation space available to a macromolecule is a very difficult task, and always requires approximation. Three levels of approximation are employed in docking:

• rigid body docking - both protein and ligand are treated as having no conformational freedom

• semi-flexible docking - the ligand is given some degrees of freedom, while the protein is maintained in a single rigid conformation

• fully flexible docking - both components are given at least some degree of conformational freedom

Conformational freedom may be added to the components of the dock is a subsequent step, by conducting a molecular dynamics minimization.

The input to a docking experiment typically consists of two structures, usually in Protein DataBank format.

• The protein structure typically is downloaded from the PDB.

• The small molecule structures may either be built with modeling software or downloaded from any of several databases (see below).

A force field is needed to represent the interactions between the protein and ligand; i.e., this is a molecular mechanics methodology.

• These include steric interactions, electrostatics, and sometimes hydrogen bonds as a separate type of electrostatic interaction.

• Consideration of electrostatics requires that the input structure files include information on atomic charges.

• Typically, the force field is adapted from one of those used in molecular dynamics calculations

A scheme for searching the conformation space of both protein and ligand must be chosen. Several schemes are described on the next page.

Finally, some method for scoring - selecting the best fit - must be used.

• The free energy of association would be the best choice; however estimating the entropy change on association is problematic

• The enthalpy of association is most commonly used and is available directly from the force field calculations

• Numerous empirical schemes have been developed for use in drug design