Ciffly

Featured solution · Life sciences

Causal single-cell AI for IBD drug response.

A production-grade web platform for inflammatory bowel disease drug-response analysis. It combines patient single-cell data with a causal foundation model to surface the cell populations, pathways, and gene drivers behind treatment response.

Developed by Ciffly in partnership with mvdresearch LLC.

Explore the platform Book a walkthrough
sidneystreetai.com / cohort / anti-TNF / donor-view COHORTS Anti-TNF · IBD cohort A 42 JAK-i · IBD cohort B 28 Anti-TNF · Cohort C 19 JAK-i · Cohort D 33 FILTERS Responders Non-responders Single-cell UMAP · Anti-TNF cohort A responders vs non-responders · 128,540 cells UMAP-1 UMAP-2 Neighborhood 1 Neighborhood 2 Neighborhood 3 Responders Non-resp. Gene drivers Neighborhood 1 · 14 candidates IL23R TNFRSF1B STAT3 JAK2 NOD2 CARD9 ATG16L1 AI SUMMARY IL-23 / TH17 axis is the strongest responder signal, consistent across cohorts A and B. Pathway enrichment (KEGG + Reactome) · p < 0.01 Pathway enrichment · top terms IL-17 signaling JAK-STAT signaling Cytokine-receptor binding T-cell differentiation NOD-like receptor Autophagy
Preview

Representative preview. For the live product, visit sidneystreetai.com.

What it does

From cohort to cell to gene — in a browser.

Cohort explorer

Browse patient donors across published IBD cohorts (anti-TNF and JAK-inhibitor studies) and pick the slice you want to interrogate.

Single-cell neighborhoods

See which cell neighborhoods diverge most between responders and non-responders, with interactive UMAP embeddings.

Gene & pathway drivers

Surface candidate gene drivers behind drug response for each neighborhood, with pathway enrichment layered on top.

AI-generated summaries

Plain-English summaries of each analysis step, ready to drop into a report or deck.

Why it's different

Causal, not correlational.

Built on a foundation model trained across pan-cohort single-cell atlases, so the signals generalise beyond any single trial.

End-to-end in the browser.

No notebooks, no local setup — pick a cohort, select donors, run.

Designed for teams.

Role-based access, admin approval flow for new users, audit trails, and private-dataset upload ship with the platform.

A partnership between

Ciffly mvdresearch LLC

Ready to see it in action?

Open the live platform at sidneystreetai.com and run a full analysis in under 10 minutes.

Open Sidney Street AI Talk to the Ciffly team