The Sixth CACHE Challenge – A Comprehensive Drug Discovery Workflow to Discover Potential Inhibitors of the Triple Tudor Domain of SETDB1
Bibliographic record
Abstract
Our main objective for the CACHE Challenge #6 was to identify novel inhibitors of SETDB1 that bind to its distinctive triple Tudor domain (TTD). As CACHE challenges are inherently collaborative, we partnered with HTuO Biosciences Inc., a Vancouver-based company developing advanced physics-based computational drug discovery technology, to share knowledge and expertise. Together, we developed a comprehensive drug discovery workflow focused on a structure-based approach. First, we carried out in-depth structural analysis on the existing SETDB1 crystal structures to identify key structural elements that we could leverage in a prospective screening campaign. Then, we carried out a retrospective benchmarking study (i.e., self-, cross-docking) of our docking program Fitted on a select number of structures, followed by extensive pharmacophore modeling. We implemented the learnings from previous CACHE challenges by considering a relatively large library of small molecules (Enamine REAL Diversity Set, ~67M compounds) for screening. We filtered this library according to CACHE and medicinal chemistry guidelines and screened it against our pharmacophore models, obtaining ~263K hits. We used Fitted to dock and score these compounds against several SETDB1 structures and our in-house protein-ligand analysis platform to identify those compounds that interacted with key amino acids identified in our structural analysis (~26K compounds). We further trimmed down the list using a combination of ranking by dock score and clustering, to arrive at a final list of 629 compounds for visualization. We developed a thorough visual inspection scheme that focused on aspects such as key protein-ligand interactions, ligand conformation, favorable and unfavorable contacts, and overall pocket fit. Compounds were scored by each team member on a scale of 1-10 and statistics were gathered; the compounds were assigned to different confidence tiers (low, medium-low, medium-high, high), depending on the standard deviation of the visual scores. The compounds from the (medium-)high tiers with a mean score ≥ 5.5 were automatically advanced to rescoring with AtomForge, a highly accurate polarizable general-purpose forcefield developed by HTuO Biosciences for use in drug discovery. Compounds in the (medium-)low tiers were debated as a group and included in the selection for rescoring if arguments for their inclusion were persuasive. Overall, 186 compounds were rescored. The selection of the preliminary list of 150 compounds focused primarily on compounds with good docking/visual scores and AtomForge affinity ranking better than positive controls. However, we also sampled compounds with different combinations of docking scores, visual scores, and AtomForge affinity ranking, which will allow us to critique our different evaluation metrics. From this list, we selected 100 compounds for purchase and testing based on practical considerations of cost, availability, and synthetic feasibility.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.004 | 0.005 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.003 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.002 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.003 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".