# SpatialScout

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Chandra Rup Daka](https://cerebralvalley.ai/u/Chandra_r_d)
- **GitHub:** https://github.com/chandrarup/spatialscout
- **Demo video:** https://youtu.be/KDtEAjG8lKo
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4805

SpatialScout is a spatial-omics co-scientist powered by Meta models, built for kidney researchers investigating unexpected molecular signals in tissue.
A researcher must connect tissue images, gene and protein measurements, analysis controls and papers before choosing a follow-up experiment. SpatialScout brings that work into one delegated investigation: it plans supported analyses, runs scientific tools, requests controls, checks claims and returns inspectable evidence, figures and suggested next experiments.
Our first working module tests whether RNA signals persist when the rules assigning molecules to cells become stricter. It combines an interactive tissue viewer, random-loss and cell-label sensitivity controls, independent numerical recomputation, model-based figure review and a follow-up lab for cited literature and separate protein-image evidence. Researchers can independently inspect biopsies, draw regions for fresh analysis, and explore measured RNA molecule depths in 3D with depth slicing and quantitative profiles. Compatible Xenium sections can also be uploaded for a fresh deterministic audit.
We ingested 22 deposited kidney RNA sections (21 distinct) and provide four real KPMP protein-imaging biopsies. The RNA cohort includes diabetic kidney disease and other glomerular diseases; the protein specimens are separate healthy/CKD/AKI participants. Results are exploratory research evidence, not clinical conclusions.
The featured live-model investigation made 18 model calls, took 179 seconds and cost an estimated $0.29; its follow-up lab has separate timing and cost. The public site transparently replays that recording, while sample-data audits compute fresh. Human time savings have not yet been measured.
Try it: https://chandrarupdaka--spatialscout.us-east.modal.run
The report links claims to computed artifacts and exports figures, a hash-linked audit log and a verification notebook.

## More from Meta Global AI Developer Hackathon

- [Caseload](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4802)
- [Team United](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4803)
- [I don't know what to say, i just want to say... E](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4804)
- [OrderGuard](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4806)
- [MI Group](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4807)
- [Oddlings](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4808)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/4805. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt