QurieGen and Rayca Precision redefine drug discovery by connecting novel target discovery with AI-powered drug design
Partnership Nov 3, 2026
Building the operating intelligence for the molecular world.
Rayca learns why therapeutic candidates succeed or fail, then uses that knowledge to guide design, simulation and experiment across modalities.
Each program expands what the system can understand, design and test.
Rayca agentRunning
- Read the biology A deep pocket with a mutant cysteine
- Match the modality Oral small molecule, covalent
- Score Pocket occupancy
- Score Covalent reactivity
- Score Absorption and clearance
- Return ranked designs With the reasoning exposed
Candidate · iteration 1
RC-0512
covalent KRAS G12C inhibitor
MPO
0.42
Starting designCurrent design
- Pocket occupancy–
- Warhead kinact/Ki–
- cLogP–
- Caco-2 Papp–
Small molecule
KRAS G12C with a covalent inhibitor, PDB 6OIM
Advancing molecular intelligence with institutions at the scientific frontier.
Our collaborators connect molecular intelligence to frontier biology, translational science and therapeutic creation. Rayca is part of the DayOne Accelerator in Basel and Health Incubator Helsinki.
01
Therapeutic co-development partners
02
Accelerator and ecosystem partners
Too little of it is available when the next candidate is designed.
Evidence is scattered across experiments, models, reports and teams. When a candidate fails, the underlying reason rarely informs future design across programs or modalities.
Failure analysis and design constraints
Representative oncology programMolecular design question
What did every prior molecule prove this design must avoid?
Assays
Assay systems and LIMS
18,240 molecular outcomes
Chemistry
Registries and notebooks
3,120 synthesized compounds
Structures
Experimental and modelled
46 conformational records
Developability
ADME and safety reports
212 liability reports
Decisions
Reviews and meeting notes
390 program reviews
Scientific expertise
Distributed across teams
14 domain specialists
The failure is documented in three systems. Its design principle exists in none.
15 years
Typical target-to-approval timeline
$2.6B
Cost per approved drug
90%
Of clinical candidates lost
Commonly cited industry estimates. Each cell is one clinical candidate.
Why it matters
Rayca preserves supported failure patterns as design principles for future candidates.
Rayca connects molecular evidence, models, simulations and experiments.
The platform
A shared intelligence guides what happens next.
The science
The platform builds a living understanding of each therapeutic program, including candidates, biological context, predictions, experimental outcomes and decisions. New results refine the next design cycle.
The compute
Across successive programs, that understanding can grow toward molecular superintelligence.
The architecture of operating molecular intelligence
Evidence, models and physics- Sovereign by design. Your science and intelligence stay yours.
- Auditable decision lineage. Evidence, methods and uncertainty remain exposed.
- Calibrated on each DMTA cycle. New assay results extend the program-specific evidence base.
Layer 04 · Governed execution
96 jobs governed
Program questions are translated into coordinated models and simulations.
Simulation on HPCModel inference on GPUOther work
- PlanNine simulations, within the cycle budget
- AllocateDynamics and free energy on HPC, models on GPU
- SuperviseTwo failed jobs restarted from checkpoint
- RecordInputs, method versions and outputs kept for review
Plans the work
The program decision determines the models and simulations required.
Governs execution
GPU and HPC follow program policy.
Keeps the result
Outputs return to the program with context and provenance.
Advances every computational outcome back into the world model
Keep the scientific reasoning available for review.
Rayca plans the analyses required by a program question, runs the relevant models and simulations and checks the results before they reach the project team.
Agent run 4-117
RunningGoal: nominate 8 analogues for cycle 5 that clear C-07 to C-12 with predicted ΔΔG at or below −1.0 kcal/mol. Read on a representative KRAS G12C series.
3:56
Elapsed7
Agents9 of 12
Steps done64 h
GPU, of 1600:001:002:003:004:005:00
Planner
Evidence
Memory
Design
Simulation
Compute
Critic
Events
- 3:56ComputeJob 71 preempted, resubmitted on gpu-h100-07
- 3:51SimulationRBFE edge RC-0504 to RC-0440 converged, −1.62 ± 0.21 kcal/mol
- 2:26PlannerPlan revised: R3 re-enumerated under C-09, RC-0523 replaces RC-0521
- 2:12CriticFlag: RC-0521 bound-state strain 2.4 kcal/mol, above C-09
- 0:44MemoryModels retrained on cycle 4; rule C-12 written
Adaptive plan
The plan changes when evidence changes.
Recorded execution
Every input, method and reason remains inspectable.
Critic in the loop
Contradictions are challenged before review.
Institutional memory
Every failure becomes a rule the next molecule must respect.
Ask why a candidate failed and the answer is already connected to the evidence.
Rayca keeps the supported reason a compound dropped out and turns it into a constraint that guides every future design, across programs and modalities.
Why did RC-0521 drop out?
It carries a basic amine on the piperazine. Three earlier compounds with that group showed a Caco-2 efflux ratio above 10.
Can we keep the scaffold?
Yes. I removed the amine from the next series and added the constraint to every future design.
Design principle saved
Institutional memory
A documented failure is saved as a design principle, so the next series never repeats it.
Saved design principle
No basic amine on the piperazine.
PARP2 windowcLogP < 3No basic amine
Applied automatically to every enumerated series from cycle 5 onward.
Our approach
A therapeutic candidate must work across many dimensions.
Rayca evaluates biological activity, developability and delivery together.
Discovery, development and translational teams each see a different part of the candidate. Rayca brings their evidence into a shared view before major design decisions are fixed.
Integrated candidate assessment
RC-0512- Medicinal chemistrySAR, synthetic tractability
- Structural biologyBinding mode, pocket states
- BiochemistryPARP1 and PARP2 potency
- DMPKSolubility, permeability, clearance
- SafetyhERG, blood counts
- TranslationalExposure at projected dose
3 vectors open
H-bond to Gly863
PARP1 1.2 nM
85 µM soluble
hERG > 30 µM
50 mg once daily
Why it matters
Project teams review an integrated candidate profile before committing the next experiment.
One DMTA cycle
A therapeutic design cycle that learns from its own results.
Evidence, design, simulation and experiment in continuous exchange.
A representative PARP1 program shows how evidence, learned design principles, new candidates, simulation and project review connect across one cycle.
1Observe · Molecular perception
Connect the program’s molecular evidence.
Assays, structures, chemistry, DMPK, safety and project decisions are linked with provenance. Conflicting measurements remain side by side.
1Molecular perceptionCycle 4
Evidence entering the molecular world model
- Compound registry3,120 compounds
- PARP1 and PARP2 assays9,804 values
- Crystal structures12 files
- ADME panel418 results
- Rat safety study1 report
- Meeting notes64 documents
4,612
Records resolved6
Sources2
Conflicts held2Learn · Principle formation
Learn from why earlier candidates failed.
Rayca links attrition events to molecular causes and supporting assays. Recurring liabilities constrain the next enumerated series.
2Principle formationCycle 4
Attrition matrix, compounds by assay
| PARP1 IC50 | PARP2 IC50 | Kinetic solubility | Caco-2 efflux | hERG | Rat blood counts | |
|---|---|---|---|---|---|---|
| RC-0412 | ||||||
| RC-0433 | ||||||
| RC-0451 | ||||||
| RC-0467 | ||||||
| RC-0480 | ||||||
| RC-0486 | ||||||
| RC-0490 | ||||||
| RC-0494 | ||||||
| RC-0497 | ||||||
| RC-0503 |
Each cell is one measured outcome. Red cells failed their target product profile threshold.
Failure mode
P-gp efflux
Caco-2 efflux ratio above 10, limiting cellular target engagement.
Design principle, now executable
No basic amine on the piperazine
Seen in 3 compounds, backed by 9 measurements
3Design · Intelligence-guided design
Generate new candidates within the program’s design principles.
Generative models explore new chemical matter inside the boundaries established by prior success and failure. Designs that violate known principles are removed before they consume experimental capacity.
3Intelligence-guided designCycle 4
Reference
Olaparib, the reference chemotype
61 proposed, 9 pass every rule
Each cell is one generated molecule. Ink passed every constraint, red broke a learned constraint, grey scored below the potency model.
RC-0512
Cyanopyridine replaces the amide cap
- PARP2 windowNext step
- cLogP below 3Pass
- No basic aminePass
RC-0515
A nitrogen added to the core ring
- PARP2 windowNext step
- cLogP below 3Pass
- No basic aminePass
RC-0518
A spiro ring replaces the piperazine
- PARP2 windowNext step
- cLogP below 3Pass
- No basic aminePass
RC-0521
A basic amine on the piperazine
- PARP2 windowNext step
- cLogP below 3Pass
- No basic amineBreaks rule
4Reason · Physical reasoning
Test promising candidates with physical simulation.
Docking, molecular dynamics and relative binding free energy test surviving designs at increasing resolution. Rayca selects, runs and records the computation required for the decision.
4Physical reasoningCycle 4
Runs on the GPU fleet
- Ensemble docking9 compounds · CPU
- Molecular dynamics, 100 ns6 compounds · GPU
- RBFE, PARP13 compounds · GPU
- RBFE, PARP23 compounds · GPU
Blue cells run this campaign.
Predicted PARP2 / PARP1 window
- RC-0512180x
- RC-0518140x
- RC-051540x
Line marks the 100-fold constraint.
5Act · Molecular action
Return a prioritized design and experiment plan.
The cycle returns a prioritized synthesis and experiment plan with its reasoning exposed. Scientists make the decision, and every observed result expands the intelligence behind the next cycle.
5Molecular actionCycle 4
Cycle 5 synthesis plan
- 01SynthesiseRC-0512 and RC-0518Pass every rule and are predicted to favour PARP1 by 140 to 180 times.
- 02Run firstPARP2 counter-screen and 7-day rat CBCConfirms the window in vitro and in vivo before scale-up.
- 03DeprioritiseRC-0515 and RC-0521RC-0515 misses the selectivity threshold; RC-0521 violates the efflux constraint.
4,612
records
3
constraints
9
simulations
1
partner assay
RC-0512 against the target product profile
- PARP1 IC50, predicted1.8 nM
- PARP2 / PARP1 window180x
- cLogP2.88
- Basic amineNone
- Caco-2 efflux ratio, predicted1.9
Molecular intelligence, by cycle
C1
C2
C3
C4
C5
Lab results return to the program record and cycle 5 starts from there.
PARP1-selective inhibitor program
Representative program. Values are illustrative.
Lab in the loop
Choose the experiment that resolves the most important uncertainty.
Return the result to the next design cycle.
When evidence is thin, Rayca can identify the experiment that will resolve the most valuable uncertainty, match it to a partner laboratory and prepare it for your approval. The observed result returns directly to the molecular world model.
01 / 04Design
Which experiment resolves the most?
- PARP2 cellular counter-screen0.92
- Kinetic solubility, three pH0.48
- Caco-2 permeability0.31
Selected by expected information gain
Why it matters
The next design cycle incorporates the new assay result and its provenance.
Lab in the loop
The agent picks the experiment that resolves the most.
When evidence is thin, Rayca proposes the one experiment worth running next.
It matches the experiment to a partner laboratory and prepares it for your approval. The observed result returns directly to the molecular world model and the next design cycle.
Which experiment separates RC-0512 and RC-0518?
A PARP2 cellular counter-screen. It resolves the selectivity gap the models are least certain about.
Run it.
Matched to a partner lab and prepared for your approval. The result returns to the model.
Sent for approval
Lab in the loop
Rayca chooses the decisive experiment, waits for your approval, and learns from the result.
PARP2 cellular counter-screen
Result returned to the model
Selectivity rule updated
Run in a partner lab
The plate the agent designed reads out live, and every well returns to the molecular world model.
Selection still depends on biological evidence and experimental context.
Rayca evaluates proposed candidates against what the program has learned, using biological evidence, predictive models, physical simulation and developability. Project teams can inspect the basis for each priority.
Failure patterns guide future designs
Supported reasons for failure remain available as principles for future candidates.
- PARP2 window
- cLogP below 3
- No basic amine
The scientific question shapes the work
Rayca selects and supervises the models and simulations needed for the program decision.
RBFE campaign · 48 of 48 GPUs allocated
Recommendations retain scientific lineage
Recommendations trace to the molecules, evidence, methods and uncertainty that produced them.
Synthesise RC-0512›RBFE run 0418›parp2_assay.csv›Row 212
Scientists remain in command
Rayca expands the team’s capacity to reason and execute. Scientists retain every go and no-go decision.
Synthesise RC-0512 and RC-0518Approved by project team
Why it matters
Experimental capacity is focused on candidates supported by the program’s evidence.
Partnering
Apply Rayca to a difficult therapeutic program.
Start with a consequential molecular decision.
A first engagement focuses Rayca on one molecular decision that matters. Inside your environment, the system connects the evidence, forms design principles, runs targeted computation and returns prioritized designs and experiments with their reasoning exposed.
Program intelligence engagement
Defined therapeutic programDefine the decision
A defined program and consequential question
Establish the world model
Evidence connected inside your environment
Run the intelligence loop
Principles, design, physics and compute
Return molecular action
Prioritized designs and experiments
Program intelligence
The hidden constraints, unresolved questions and next actions inside a running program.
Molecular second opinion
Independent reasoning before the next synthesis or IND-enabling commitment.
Failure-to-principle analysis
Past attrition converted into constraints the next design must respect.
Intelligence-guided design cycle
New matter designed, physically tested and prioritized with complete lineage.
What the intelligence loop returns
Program measurements integrated4,612
Recurring failure modes3
Constrained designs scored61
Simulated in physics9
Prioritised for synthesis2
Molecular world model
Program evidence connected with provenance.
Executable design principles
Failure modes converted into future constraints.
Prioritized molecular actions
What to design, simulate and test next.
Reasoning for each recommendation
Evidence, methods and uncertainty exposed.
Small moleculesPeptidesAntibodiesNanobodiesDegraders
Across modalities
Small molecules, Peptides, Antibodies, Nanobodies, Degraders.
Vision
Rayca’s current platform learns within therapeutic programs.
Our research direction is molecular superintelligence.
Each program expands what the institution knows about molecular design, biological response and therapeutic performance. Over time, Rayca can connect biological hypotheses with development-candidate design across modalities.
The path to molecular superintelligence
Our directionCapabilityIntelligence compounding inside each institution
One hard programSubsequent programs and experiments
- L1ConnectedCompounds, assays and decisions share traceable molecular context.Today
- L2Principle-formingObserved outcomes define evidence-linked molecular constraints.Today
- L3OperationalThe system reasons, designs and runs computation.Today
- L4ExperimentalIt proposes and coordinates experiments with scientific approval, then learns from reality.In development
- L5Molecular superintelligenceIt connects biological hypotheses with development-candidate design across disciplines and modalities.Our direction
The destination is an intelligence capable of understanding and engineering molecular systems beyond the reach of any individual model, discipline or program.
Meet the teamToday
- Connected
- Principle-forming
- Operational
Running inside therapeutic programs now.
Next
- Experimental
- Molecular superintelligence
Experimental: in development. Molecular superintelligence: our direction.
Latest updates
All posts
Rayca Precision pitches as a Nordic Star finalist at Nordic Life Science Days 2026
Event Sep 9, 2026
Rayca Precision selected for the 2026 DayOne Accelerator in Basel
Programme Aug 19, 2026
Bring us the molecular problem your current tools cannot resolve.