It is our greatest pleasure to announce the winner of the Summer 2025 edition of the Scientific Challenge: the winner is Dylan Capitti Fenton of California State University, Northridge (Van Nuys, United States) with his project ‘Structure-Based Drug Discovery of Bitter Taste Receptor Agonists for GLP-1 Release’.
Focusing on the therapeutic management of type 2 diabetes, Dylan developed an advanced tandem virtual screening workflow targeting bitter taste receptor 38 (TAS2R38). This target acts as an upstream stimulator for endogenous GLP-1 release but remains structurally unresolved with very few known actives. Dylan initiated his campaign by screening the Enamine Hit Locator Library of 460,160 compounds to identify 1,000 initial seed molecules. Using the ligand-based tools FTrees, SpaceLight, and SpaceMACS, these seeds were expanded against the massive Enamine REAL Space repository of 83 billion compounds to generate a target-optimized library of 300,000 make-on-demand analogs.
A standout feature of this research lay in the rigorous generation and validation of the active-state receptor model. Dylan constructed a homology model using TAS2R14 as a template, refining it within its native membrane environment through extensive physics-based molecular dynamics simulations, including 20 µs of apo and 4 µs of holo MD. This computationally intensive approach demonstrated clear superiority over deep learning alternatives, yielding a logAUC enrichment score 30.5% higher than AlphaFold2-based models and 93% higher than AlphaFold3. The optimized 300,000-compound library was then docked into this validated pocket, significantly shifting the score distribution toward stronger predicted binders and improving the top dock score by 13%.
Overall, Dylan impressed us with his systematic integration of physics-based protein refinement and ultra-large chemical space navigation. Following hit filtering, clustering, and strict pose inspection, 30 chemically diverse candidates were successfully synthesized by Enamine. The stability of the top four novel hits was further confirmed through 100-ns MD simulations. These promising candidates are currently undergoing in vitro validation via a TRUPATH BRET2 G-protein dissociation assay at The University of North Carolina at Chapel Hill.
We wish Dylan all the best for his professional and scientific future and look forward to reading more about his discoveries in the future!