Our goal with this competition is to evaluate our released tools (rather than new methods). Pharmit (http://pharmit.csb.pitt.edu) will be used to perform pharmacophore screens of purchasable compounds. Pharmacophores will be constructed in a mostly manual process through inspection of known binders and docked fragments. We will screen all the commercial libraries in Pharmit (MolPort, MCULE, Chemspace, LabNetwork). Hit compounds will be pose optimized and ranked using the default deep learning scoring function of gnina (https://github.com/gnina). Multiple receptor structures will be used for pose optimization and ranking. Molecular dynamics simulations may be used to generate an ensemble of receptor structures. Consensus ranked hits will be clustered by chemical similarity to maximize the diversity of scaffolds and filtered by desirable properties (e.g. solubility, molecular weight, rotatable bonds). We also check that proposed compounds still have advertised availability from their vendors. Time permitting we may also do a high-throughput "blind" docking of the target using gnina. Default options will be used. Most likely this screen will be limited to the MolPort library and use only a single rigid receptor structure. We will merge the ranked lists from Pharmit and blind docking, but also ensure that compounds from both methods are selected.
