Initially, compound databases were prepared, and screening methodologies were optimized. A total of 125 sEH crystals from the PDB were prepared. Records from the vendors were prepared for screening using our CompoundUniqifier [1] tool to standardize and filter datasets for virtual screening (VS). Similarity filtering against co-crystallized sEH ligands reduced compounds to millions, enhancing computational feasibility. Bioactivity data was curated with our SARUniqifier [2] tool. Cross-docking studies using AI and physics-based tools identified proteins best suited for simulations, with heatmaps to visualize the results. Structure-based pharmacophore models were developed. Their statistical analysis led to the selection of the best performing model for VS. Chemical space docking (CSD) campaign was completed for 2 commercial billion scale spaces, with analysis ongoing and while 4 more spaces remaining to be screened. Next steps include molecular dynamics (MD) and experimental validation.
After 3 months, Emine Yekta has achieved the following milestones:
- The compound library of small molecules was reduced to 5,519,235 through PAINS, REOS, and similarity filtering against co-crystallized ligands from sEH crystal structures, optimizing computational feasibility. The bioactivity dataset was refined to 8,040 molecules. HPC-enabled CSD for GalaXi and CHEMriya spaces was completed, and analysis is ongoing to identify top candidates. Future plans include screening the remaining spaces REALspace, AMBrosia, FreedomSpace, and eXplore, expanding the scope of compound exploration. Cross-docking and re-docking studies with AI- and physics-based tools identified reliable protein structures for downstream analyses & heatmaps highlighted top performers. Refined bioactivity data were clustered, actives identified, and decoys generated, which were used to validate pharmacophore models. ROC curves and AUC values confirmed the best-performing pharmacophore models for VS, which were integrated into the screening pipelines.
- The groundwork for Milestone 2 is actively underway, with VS efforts generating a focused list of potential active compounds. The integration of validated pharmacophore models into the workflow has ensured the selection of high-confidence candidates from an initial pool of millions to billions of compounds. Preparations for molecular dynamics (MD) simulations are underway. Protein-ligand complexes will be assembled using the shortlisted active compounds from the VS results. These simulations aim to evaluate the binding stability, interaction dynamics, and conformational flexibility of the complexes. Upon completing the initial MD simulations, data will be analyzed to further refine the selection criteria, ensuring that only the most promising candidates advance to experimental validation. The MD simulation phase is expected to provide critical insights into the molecular behavior of the selected compounds, strengthening their potential as lead candidates.
- Enzyme inhibition and kinetic assays are being optimized using known inhibitors to support in vitro validation. Current efforts focus on establishing robust experimental workflows to ensure accuracy and reproducibility. Additionally, an activity assay method is under development to test compounds identified through VS. Preliminary results indicate an IC50 value of 1 nM at an sEH concentration of 1.2 µg/L. Following the completion of VS and MD simulations, selected compounds will undergo in vitro assays to validate their activity against the target enzyme. This phase will assess their inhibitory potential and elucidate their mechanisms of action. Subsequently, crystallization studies will be conducted by our collaborators at Koç University using Turkish DeLight [3]. References are provided within the attached figure.