Biophysical properties of small molecules binding to proteins

Binding MOAD (Mother of All Databases) is the largest collection of high-quality, protein-ligand complexes. Binding MOAD contains 13,138 protein-ligand complexes comprised of 4078 unique protein families and 6210 unique ligands. We have compiled binding data for 4146 of the protein-ligand complexes. The creation of this database and three studies mining the database for biophysical properties of protein-small molecule binding are discussed in this thesis. An additional study is included in the appendix which investigates flexibility upon small molecule binding to MDM2.;First, we present the development of GoCav, which allows us to mine properties of the whole database. We have determined that most complexes have well buried binding sites (70--85%), which fits the idea that a large degree of contact between the ligand and protein is significant in molecular recognition.;Secondly, we investigate the differences in biophysical properties of binding to enzymes versus non-enzymes. Differences in the sizes of weak versus tight ligands indicate that the addition of complementary functional groups may improve the affinity of an enzyme inhibitor, but the process may not be as fruitful for ligands of non-enzymes. Non-enzymes were found to have greater ligand efficiencies than enzymes, which supports the feasibility of non-enzymes as druggable targets. This has significant ramifications for target selection in drug design. Most importantly, the differences in ligand efficiencies appear to come from the pockets which yield different amino acid compositions, despite similar overall distributions of amino acids.;We then investigate the biophysical properties of the most efficient protein-ligand complexes. All highly efficient small molecules contain one or more charge and are found in binding sites with at least one charge, challenging previous thoughts that hydrophobic properties of ligands lead to the better binding. Lastly, it is known that affinity for complexes rarely exceeds -15 kcal/mol, and we suggest that ligands do not exceed this value because there is no evolutionary pressure to drive tighter binding.