The current state of the art of molecular docking software is much better at predicting poses of ligands in receptors than at predicting accurate values for the free energy of binding.

Hydrophobic interactions are the main driving force for complex formation, mainly because of the increase of solvent entropy upon burial of hydrophobic surfaces upon complex formation. Although hydrogen bonds contribute significant interaction energy, you actually have to break a hydrogen bonds to water in order to form hydrogen bonds between receptor and ligand, yielding a low net contribution to the free energy of complex formation. As a result, hydrogen bonds are mainly important for specificity, as burying a hydrogen bond acceptor or donor in the interface without forming a hydrogen bond is energetically very unfavourable.

Just enumerating and counting the interactions is a very poor measure of the interaction energy, as the contribution of each type of interaction is also very dependent on the exact distance between the interaction partners.

Certainly for ranking of the different docked poses, docking scores are better than just counting the number of interactions shown in LigPlot, especially since LigPlot does not highlight repulsive interactions.

Hi Kevin,

In addition to the excellent comments by Dr Richardson and Honegger let me also comment. In my experience with structure-based virtual screening, many interactions between polar groups, such as hydrogen bonds, in the molecular docking results does not determine whether the ligand actually binds strongly or not. If you are not sure that your ligands will bind strongly experimentally, you should focus on ligands that form only one or two hydrogen bonds and on their binding energies through molecular docking. This is because it is highly unlikely that many specific bonds like hydrogen-bonding between structurally un-optimized ligands and amino acids in a protein pocket will be conveniently formed.