Spatial Transcriptomics Agent
Omi's specialised co-pilot for spatial work
I bring your spatial transcriptomics data to life — Visium, Xenium, MERFISH, Stereo-seq, Slide-seq — and integrate it with your dissociated scRNA so you can see not just what your cells are, but where they are and who they're sitting next to.
What I can do for you
I load and QC spatial datasets across all major platforms, run deconvolution (Cell2location, RCTD, Tangram) to map your scRNA cell types onto Visium spots, or directly cluster Xenium/MERFISH at single-cell resolution.
I detect spatially variable genes, identify tissue niches and microenvironments using BANKSY or GraphST, and quantify cell-type colocalization — telling you which cell types form neighbourhoods together.
I run spatial cell-cell communication (COMMOT, MISTy) that respects physical proximity — so a predicted ligand-receptor interaction only counts when the cells are actually within signalling distance.
I overlay clinical or pathology annotations (tumor edge, stroma, immune infiltrate) and run differential analyses across spatial regions — perfect for digital pathology and immuno-oncology studies.
Examples of what you can ask me
Copy any of these straight into the demo, or adapt them to your data.
- 1"Deconvolute my Visium slide using my matched scRNA reference."
- 2"Find spatially variable genes in my Xenium tissue."
- 3"Identify tumor-immune neighbourhoods in my MERFISH sample."
- 4"Which cell types co-localize at the tumor-stroma boundary?"
- 5"Run spatial cell-cell communication and show interactions within 50 µm."
- 6"Compare gene expression in tumor core vs invasive front."
How I work
I run real Scanpy (Python) or Seurat (R) code on the secure MCP server — no hallucinations, no made-up gene lists. Every result comes with the exact code I executed and the parameters I used, so your analysis is fully reproducible and ready for the Methods section.
Best for
Tissue biologists, neuroanatomy folks, immuno-oncology researchers, developmental atlas builders, and anyone who needs to know where the biology is happening, not just what's happening.
References
- Cell2location (Kleshchevnikov et al., 2022) – Nature Biotechnology
- RCTD (Cable et al., 2022) – Nature Biotechnology
- Tangram (Biancalani et al., 2021) – Nature Methods
- BANKSY (Singhal et al., 2024) – Nature Genetics
- GraphST (Long et al., 2023) – Nature Communications
- COMMOT (Cang et al., 2023) – Nature Methods
- Squidpy (Palla et al., 2022) – Nature Methods
Try Spatial now
Jump into the demo with a starter prompt already loaded. Upload your data, or play with our example dataset first.