I / Model Systems and Decoys can be used to test and ideally improve Molecular Docking Scoring Functions. Here we provide two types of decoys which address two major problems in molecular docking "Hit List" Decoys are experimentally determined non-binders that are predicted to bind, and "Geometric" Decoys are known ligands predicted to bind with an incorrect configuration-greater than 3 angstrom RMSD from the crystallographic pose-in the binding site.
Binders / Apparent non-binders:
Our "Hit List" Decoys come from Model Binding Sites that range in complexity from a small, hydrophobic cavity site in a T4 Lysozyme mutant (L99A), to a small, polar cavity site in a second T4 Lysozyme mutant (L99A/M102Q), to a small, negatively charged cavity site in a Cytochrome C Peroxidase mutant (W191G), to a solvent exposed drug target, AmpC beta-lactamase. The Model Systems, in particular the model cavity sites, are simple enough to allow one to separate the different energetic contributions via experimental measurements and to isolate modifications in a new scoring function from other aspects of binding. The decoys from these simple sites further allow one to elucidate the strengths and weaknesses of ones Molecular Docking algorithms and ideally direct their improvement.
Geometric decoys:
Our "Geometric Decoys" come from five well-characterized proteins each having several ligand-bound structures in the Protein Data Bank. We considered 19 complexes of DHFR, 25 complexes of thrombin, 12 complexes of PNP, and 13 complexes of TS, and eight complexes of AChE. DOCK3.5.54 was used to generate and score multiple conformations and orientations of each ligand in its cognate protein. In most cases, the best scoring ligand geometries matched to the crystallographic ligand geometries to within 2.0 angstrom RMSD; such geometries are considered to be native-like. We focused on ligands that had decoy geometries (> 3.0 angstrom RMSD from the native pose with better energy scores than any of the native-like dockings) to develop a test set of geometric decoys. DOCK predicted four geometric decoys for DHFR, five for thrombin, two for PNP, six for TS, and three for AChE.
II / Many promiscious inhibitors form aggregates causing unspecific inhibition. Here we provide a lists of smilesstrings of compounds for which aggregation has been measured (Aggregators) or for which there is no experimental evidence to form aggregates (Non Aggregators).
We are also developing high-throughput experimental methods and computational predictions to uncover aggregators.
The results of a high-throughput screen for aggregate-based inhibitors conducted by the NCGC. 70,563 molecules from the PubChem library were screened against AmpC beta-lactamase in the absence and presence of detergent.
III / The Shoichet Laboratory develops ZINC, a free database of small molecules for docking that are commercially available. ZINC is an acronym for "ZINC is not commercial". For more information, and to download ZINC for yourself, please visit the ZINC web site.