WHAT OUR CAUSAL INSIGHTS SERVICE ANSWERS
Causal insights from Same-Cell Datasets.
Causal Insights sits at target identification and validation, before a target is locked and before a program commits to it. What comes back is an answer to the question you brought.
WHY CURRENT METHODS FALL SHORT
Readouts you trust today may be pointing the wrong way
Most technologies have to kill the cell to read it. So when teams ask "what changed after treatment," they compare one group of cells sampled before intervention against a different group sampled after. The comparison is real, but the causality is inferred.
A 2026 Nature Reviews Genetics review notes that because sequencing destroys the cell it reads, computational methods must approximate a same-cell before/after comparison rather than measure one directly.
Causal Insights closes that gap by reading the same living cell before and after an intervention, so the data behind a go/no-go decision is measured, not inferred.
Mechanism of action
A mechanism can look confirmed in one cell state and silent or reversed in another. Same-cell measurement shows whether the response is truly sustained, lost, or state-dependent.
Resistance biology
Pre-existing resistance and acquired adaptation are hard to separate when different populations are compared before and after treatment. Same-cell tracking reveals which is which.
Biomarker validation
A biomarker may correlate with response in discovery, yet fail to report the correct biological mechanism. Same-cell data validates baseline markers against observed future response.
WHAT THE FIELD IS SAYING
The field is converging on the same gap
More data has not closed the direction problem. Independent benchmarks, industry commentary and the founding paper all land in the same place: endpoint measurement can tell you what a gene is associated with and often still get the sign of the effect backwards.
INDUSTRY VIEW
The bottleneck is the mechanism, not the molecule
Most clinical failures come from targeting the wrong biology, not from poorly engineered molecules. Even atlases of hundreds of millions of cells are described as orders of magnitude too small to close that gap. Accelerating a pipeline aimed at the wrong mechanism produces faster failures. — a16z, 03.08.2026
PREPRINT
Direction predicted no better than a coin flip
Across eight in-silico perturbation methods, the best performer matched experimental knockdown direction 40.9% of the time against CRISPR ground truth, where guessing would give 50%. Two methods on the same task produced anti-correlated gene rankings.
— Wu et al., bioRxiv, 19.08.2026
PEER-REVIEWED
Same-cell data recovered the correct direction
Reading the same macrophage before and after LPS identified Nfkbia as a negative-feedback inhibitor of the inflammatory response. An endpoint design on the same system read the sign the other way. Proof of principle, on 17 cells.
— Chen et al., Nature 608 (2022)
HOW IT WORKS
From uncertainty to decision
Causal Insights helps validate mechanisms and biomarkers, assess the directionality of a mode of action, distinguish responder cells and the genes that drove the response, and separate pre-existing resistance from adaptation, cell by cell. Cytosurge runs the workflow from experimental design through same-cell analysis and interpretation, with projects scoped in stages around clear milestones. From
Scope
We co-design the experiment; Scope, timeline, and deliverables are agreed before the experiment starts.
Execute
We run the Live-seq workflow on your cell model, in our lab.
Causal Analysis
We perform the causal analysis on the paired same-cell sequencing data.
Decide
You receive a written interpretation tied to your research question.
