Pain & inflammation
LoadingEvaluating targets…
Connecting ChEMBL target similarity models.
Molecular exploration / release A preview
Design with evidence,
not assumptions.
Explore structure-led variations, compare calculated properties, and preserve the reasoning behind every candidate.
01
Research tracksChoose the questions this project will organize.
Active structure
Cannabidiol reference
Analyzing
Computed conformerWaiting for RDKit analysis…
Waiting for coordinatesRDKit atom and bond counts will appear here.
drag to rotate · scroll to zoom
—Formula
—Heavy atoms
—Formal charge
—Stereochemistry
04
Structure-led exploration
Generate a focused set
Apply named, deterministic transformations to the active structure. Every output keeps its parent and method.
Advanced constraints +
Ready to rungenerator/rdkit-graph-rules@3
0%
Choose constraints, then start a bounded RDKit run.
—valid—unique—retained
05
Candidate set
Compare trade-offs
Run the generator to create candidates
| Select | Candidate | Change from parent | Formula | Mol. weight | Est. logP | Drug-likeness | SAScore | MPO | CNS / ESOL | Similarity | Status |
|---|---|---|---|---|---|---|---|---|---|---|---|
| ⌁No candidate run yetYour generated structures will appear here with transparent lineage. | |||||||||||
Select up to four candidates to compare.
06
Target Prediction & ChEMBL Profiling
Endpoint catalog & targets
Evidence-based target similarity and bioactivity profiling against ChEMBL benchmark ligands. Morgan circular fingerprint (radius 2) Tanimoto nearest-neighbors across research tracks.
Anxiety
LoadingEvaluating targets…
Connecting ChEMBL target similarity models.
Sleep
LoadingEvaluating targets…
Connecting ChEMBL target similarity models.
Seizure disorders
LoadingEvaluating targets…
Connecting ChEMBL target similarity models.
07
Preparation hypotheses
Enumerate molecular states
Generate bounded protonation and tautomer hypotheses for a selected pH range, then build a separate 3D ensemble for each state.
i
Preparation, not population predictionStates are plausible hypotheses. The app does not calculate microscopic pKa values or state populations, and energies are never compared between states.
Analyze the active structure, then enumerate its molecular states.