Generate molecules
Series, motifs & R-groups
Concept simulation
Human-guided medicinal chemistry
Design. Test. Learn.
Decide what changes next.
A fictional orchestration surface for a closed medicinal-chemistry loop: molecular generation, ADMET profiling, prioritization, biophysical assessment, and simulation-informed feedback—always steered by the chemist.
02 / orchestrate
Closed design loop
Standing by
Profile ADMET
Exposure & liability signals
Prioritize
Trade-offs & queue selection
Assess binding
Alchemistry, FEP & MD
Redirect generation
Learn, decide & iterate
LOOPCONTEXT
04 / tune each handoff
Every stage has a
human-adjustable strategy.
Choose what the conceptual loop should optimize, protect, test, and send back to molecular generation.
01 / generationChemist sets
Exploration posture
Shape the next molecular batch around the retained hypothesis rather than a generic score.
02 / admetChemist sets
Property guardrail
Decide whether the conceptual profile favors early liability removal or a wider learning set.
03 / prioritizeChemist sets
Queue logic
Make the trade-off explicit before limited biophysical-compute capacity is allocated.
04 / biophysicsReview required
Alchemistry assessment
Scope an illustrative alchemical free-energy calculation and MD simulation plan; approve before queueing.
05 / feedbackChemist directs
Feed learning back
Translate simulation observations into the next generation brief, retaining uncertainty and rationale.
Illustrative evidence queue
Not an automated verdict.
A reviewable record.
G-14Queue candidate
Property-balanced branch
Illustrative ADMET profile clears the selected guardrail; needs binding evidence before promotion.
Design fit0.82
A-22Trade-off
Potency-led branch
Illustrative potency signal is stronger; the chemist may keep it as an MD comparison case.
Design fit0.71
S-09Learn
Simulation-retest branch
Illustrative MD observation raises pose uncertainty—use it to redirect the next molecular generation.
Learning value0.76
How to read this concept
Orchestration is a
chemist-led conversation.
Specialized work stays distinct
Generation, ADMET profiling, prioritization, and alchemical/MD assessment contribute different kinds of evidence.
Evidence travels with uncertainty
The conceptual router carries the brief, decisions, retained constraints, and reasons for every handoff.
Simulation changes the next question
Alchemistry and MD observations inform—not dictate—the chemist’s next molecular-generation strategy.