Challenge #6

The design of molecules targeting the histone binding groove of the SETDB1 triple Tudor domain (TTD) presents a unique opportunity to discover novel therapeutics. We propose two complementary strategies: active learning and pharmacophore modeling, to identify and optimize potential inhibitors for SETDB1 TTD. 

In the first strategy, we employ an active learning approach utilizing MOLPAL, an open-source package, on the PDB structure 7CJT, which is complexed with a top active compound exhibiting a dissociation constant (kD) of 100 nM. We will screen the Enamine Real subset library, consisting of 48 million compounds, and select the top 15,000 compounds based on their predicted scores for further docking studies using smina. Subsequent analysis will involve pose validation, interaction assessment, and filtering based on solubility and other physicochemical properties. From this pool, we aim to procure and test 70 compounds.

The second strategy involves generating multiple interaction-based pharmacophore models for all subcavities of the SETDB1 TTD using Pharmit, an open-source pharmacophore modeling and screening platform. These models will facilitate virtual screening against the Enamine Real library. We will select the top 2,000 molecules from each pharmacophore screening based on RMSD and predicted docking scores obtained from Pharmit's minimization function. The resulting hit list will undergo docking against the respective subpockets using smina, followed by pose review, interaction analysis, and solubility and property filtering. From this approach, we plan to procure and test 30 compounds.

By integrating these diverse strategies, focusing on both the top active compound and the least active compounds binding to different subpockets, we aim to identify better binders with a ratio of 70:30 (active learning to pharmacophore). This dual approach is expected to enhance our ability to discover potent inhibitors for the SETDB1 TTD.