Our approach combines expertise of Kozakov Lab at Stony Brook and Tropsha Lab at UNC. Our workflow uses several complimentary modules for identification of high affinity hits for a given protein target with a known 3D structure. Identification of binding site hot-spot information together with conventional structure-based virtual screening methods are enabling components of our hit selection approach. First, we will use FTMap: a computational mapping algorithm that identifies binding regions on the surface of the target protein with major contributions to the ligand binding free energy. FTMap samples all possible positions of small organic molecule probes and score them using a physical energy function. The binding site regions that bind multiple probes identify the binding hot spots and the corresponding favorable chemical groups. In addition to hypothesis free FTMap, we will use a different approach towards hot spot identification, LigTBM, that was inspired by structural similarity search methods. The basic idea is to match physico-chemical environment of the protein to the micro pockets containing FTMap style small organic molecule probes extracted from PDB structures containing bound ligands. Early version of LigTBM, was top performer by docking model accuracy at NIH sponsored D3R competition. This matching procedure will also provide us with possible fragment placement within the target protein, so the data will be presented in the same form as FTMap data, which will facilitate their comparative analysis and identification of consensus hot spots. The hot spot information will be used to create a pharmacophore model for the next stage of the virtual screening. Using this model, we then will perform a pharmacophore-based screening of the entire Enamine REAL library (~40B with tautomers) to select a subset of the target specific compounds based on the fitting to our pharmacophore hypothesis. In addition, we will use pharmacophore models to bias our deep and reinforcement learning method termed ReLeaSE to generate target-specific novel hit compounds. The combined set of virtual screening and de novo generated hit compounds (usually ~1M) then will be docked into the binding site using Glide by Schrödinger, as well as LigTBM type approach. The top scored docking hits then will be additionally prioritized using the hot spot information. These consensus hits will be nominated for the experimental testing.

Glide by Schrödinger

FTMap server (https://ftmap.bu.edu/), RDKit

Kozakov D, Grove LE, Hall DR, Bohnuud T, Mottarella SE, Luo L, Xia B, Beglov D, Vajda S. The FTMap family of web servers for determining and characterizing ligand-binding hot spots of proteins. Nature Protocols. 2015 Popova M, Isayev O, Tropsha A.* Deep reinforcement learning for de novo drug design. Sci Adv. 2018 Jul 25;4(7):eaap7885. doi: 10.1126/sciadv.aap7885 Alekseenko, A.; Kotelnikov, S.; Ignatov, M.; Egbert, M.; Kholodov, Y.; Vajda, S.; Kozakov, D. ClusPro LigTBM: Automated Template-Based Small Molecule Docking. J. Mol. Biol. 2019. https://doi.org/10.1016/j.jmb.2019.12.011.