Decision-as-a-Service
The Experimental Intelligence Layer for Resistant Cancer.
Cytogent has built a closed-loop service so drug teams can prioritise combination experiments, test them with laboratory partners, and update plans from biological outcomes.
- Patient-derived organoids
- Multiplex imaging
- Combination ranking
- Visible uncertainty
- Independent confirmation
- EGFR-mutant lung cancer
- Laboratory partners
- Decision briefs
Sequencing reads what changed in a tumour. Only living cells show what two drugs do together. Cytogent connects both.
The bottleneck
Resistance writes its own rules.
After targeted therapy, tumours with the same starting mutation can escape in different ways. The hard part is choosing which experiments to run.
- Tumour
- Resistant clone
- Stroma
- T cells
- DNA
Too many drug pairs
Partners, doses and schedules multiply faster than any lab can test them.
Too little lab time
Every functional assay costs weeks and budget.
Too few living samples
Patient-derived models are scarce and precious.
Combination
The right pair takes resistant cells out.
What we have built
From hypothesis noise to a ranked batch.
Each batch is chosen for what it can teach, from the mutation to the living response.
01 · Genome
What changed
A mutation in the drug’s target, read from the tumour’s DNA.
02 · Cell
Where it acts
Inside the cancer cell, where the drug meets its target and signals reroute.
03 · Living model
What cells do
A patient-derived organoid under two drugs at once: the answer only an experiment gives.
The verdict
Living cells decide.
One drug leaves resistant cells behind. The right partner takes them out.
Untreatedlive cells
Drug Alive cells
Drug Blive cells
A + Blive cells
The loop
One question, through six decisions.
Step 1 of 6: Define. Frame the resistance question, the constraints and the experimental budget.
Standard of proof
Predictions are hypotheses. Experiments are the verdict.
A ranking is a reason to run an experiment, not a result. Key findings are re-tested before they count.
Ranked batch · wave 2
expected information
- Drug A + BConfirmed
- Drug A + CIn test
- Drug A + DQueued
- Drug B + CQueued
- Drug A + EAbstain
Bounded experimental batches
Each round is a small, feasible batch chosen for what it can teach, not an open-ended screen.
Visible uncertainty and abstention
Every ranking carries an uncertainty estimate. Where the evidence is too thin, the system should say so rather than guess.
Planned reproducibility
A signal counts once it has been seen again: out of batch, or with an orthogonal readout.
Project-specific data governance
Each sponsor's data stays in its own project, under the terms agreed for that project.
Where to start
Three ways in.
Drug teams
Bring a resistance question and scope a pilot.
Labs and biobanks
Help run and confirm the experiments that matter.
Everyone else
Why a cancer drug can stop working, in plain words.
Investing in the space? Read the investor view
Starting focus
One candidate. One resistance context. One measurable decision.
Our inaugural programme blueprint: combination selection in EGFR-mutant lung adenocarcinoma after osimertinib resistance.
