MATCHING CHEMISTRY AND SHAPE IN MOLECULAR DOCKING

We have added a chemical filter to the ligand placement algorithm of t he molecular docking program DOCK. DOCK places ligands in receptors us ing local shape features. Here we label these shape features by chemic al type and insist on complementary matches. We find fewer physically unrealistic complexes without reducing the number of complexes resembl ing the known ligand-receptor configurations. Approximately 10-fold fe wer complexes are calculated and the new algorithm is correspondingly 10-fold faster than the previous shape-only matching. We tested the ne w algorithm's ability to reproduce three known ligand - receptor compl exes: methotrexate in dihydrofolate reductase, deoxyuridine monophosph ate in thymidylate synthase and pancreatic trypsin inhibitor in trypsi n. The program found configurations within 1 angstrom of the crystallo graphic mode, with fewer non-native solutions compared with shape-only matching. We also tested the program's ability to retrieve known inhi bitors of thymidylate synthase and dihydrofolate reductase by screenin g molecular databases against the enzyme structures. Both algorithms r etrieved many known inhibitors preferentially to other compounds in th e database. The chemical matching algorithm generally ranks known inhi bitors better than does matching based on shape alone.