\n\n# ChatSpatial\n\n**MCP server for spatial transcriptomics analysis via natural language**\n\n[](https://doi.org/10.64898/2026.02.26.708361)\n[](https://openreview.net/forum?id=xZ814yNaUW)\n[](https://www.enar.org/meetings/spring2026/)\n[](https://www.ibc2026.org/home)\n[](https://github.com/cafferychen777/ChatSpatial/actions/workflows/ci.yml)\n[](https://pypi.org/project/chatspatial/)\n[](https://www.python.org/downloads/)\n[](https://opensource.org/licenses/MIT)\n[](https://docs.cafferyang.com/)\n[](https://github.com/users/cafferychen777/packages/container/package/chatspatial)\n\n
\n\n
\n \n
\n\nChatSpatial replaces ad-hoc LLM code generation with **schema-enforced orchestration**. Instead of generating arbitrary scripts, the LLM selects tools and parameters from a curated registry, making spatial transcriptomics workflows more reproducible across sessions and clients.\n\nChatSpatial exposes **20 schema-validated MCP tools** that orchestrate **66 spatial transcriptomics methods** across **15 analytical categories**. The tools are the stable natural-language interface; the methods are the analysis backends selected through tool parameters.\n\nThe server implements MCP `2026-07-28` through the official Python SDK v2 and\ncontinues to serve `2025-11-25` clients through SDK-managed protocol negotiation.\nSTDIO remains the secure local default; Streamable HTTP is available for\nexplicitly configured HTTP deployments.\n\n---\n\n## Start Here\n\nInstall [`uv`](https://docs.astral.sh/uv/getting-started/installation/) once, then\nregister ChatSpatial without creating or managing a Python environment:\n\n**Codex:**\n\n```bash\ncodex mcp add chatspatial -- uvx --from chatspatial chatspatial server\n```\n\n**Claude Code:**\n\n```bash\nclaude mcp add --scope user chatspatial -- \\\n uvx --from chatspatial chatspatial server\n```\n\n`uvx` creates an isolated environment on first launch and reuses its cache on\nlater launches. Restart the MCP client after adding the server.\n\nThe command above installs the standard runtime. To make all 15 composable\nPython method families available in the same isolated MCP environment, use:\n\n```bash\nuvx --from 'chatspatial[full]' chatspatial server\n```\n\n`full` includes CellRank, FastCCC, the maintained spatial-domain and\nregistration backends, annotation, enrichment, and the other portable Python\nfamilies. R bridges, AESTETIK, and rctd-py remain separate because they have\nsystem, platform, or large-runtime requirements. See the installation guide\nbefore enabling those families.\n\nThen:\n\n1. **Run your first analysis** — [Quick Start](docs/quickstart.md)\n2. **Choose optional method families or a persistent environment** — [Installation Guide](docs/installation.md)\n3. **Configure another MCP client** — [Configuration Guide](docs/advanced/configuration.md)\n4. **Inspect or reproduce the manuscript results** — [Reproducibility workspace](reproducibility/README.md)\n\n**Docker quick start:**\n\n```bash\ndocker pull ghcr.io/cafferychen777/chatspatial:v1.4.0\n```\n\n**Minimal example prompt:**\n\n```text\nLoad /absolute/path/to/spatial_data.h5ad and show me the tissue structure\n```\n\nIf you use Docker, mount host data to `/data` and prompt with the container path, for example `/data/spatial_data.h5ad`.\n\n> ChatSpatial works with **any MCP-compatible client** — Claude Code, Claude Desktop, Codex, OpenCode, and other MCP-capable tools.\n\n---\n\n## Capabilities\n\nCurrent coverage includes 66 methods across 15 analytical categories, exposed through 20 MCP tools. Supports 10x Visium, Xenium, Slide-seq v2, MERFISH, seqFISH.\n\n| Category | Example methods |\n|----------|---------|\n| **Data Loading & Preprocessing** | Scanpy I/O, QC, Normalization, HVG, PCA, Neighbors |\n| **Visualization** | Spatial plots, Embedding plots, Gene expression overlays |\n| **Spatial Domain Identification** | SpaGCN, STAGATE, GraphST, BANKSY, AESTETIK, Leiden, Louvain |\n| **Deconvolution** | FlashDeconv, Cell2location, RCTD (spacexr or rctd-py), DestVI, Stereoscope, SPOTlight, Tangram, CARD |\n| **Cell-Cell Communication** | LIANA+, CellPhoneDB, CellChat (`cellchat_r`), FastCCC |\n| **Cell Type Annotation** | Tangram, scANVI, CellAssign, mLLMCelltype, scType, SingleR |\n| **Differential Expression** | Wilcoxon, t-test, Logistic Regression, pyDESeq2 |\n| **Trajectory Inference** | CellRank, Palantir, DPT |\n| **RNA Velocity** | scVelo, VeloVI |\n| **Spatial Statistics** | Moran's I, Local Moran, Geary's C, Getis-Ord Gi*, Ripley's K, Co-occurrence, Neighborhood Enrichment, Centrality Scores, Local Join Count, Network Properties |\n| **Enrichment Analysis** | GSEA, ORA, Enrichr, ssGSEA, Spatial EnrichMap |\n| **Spatially Variable Genes** | SpatialDE, SPARK-X, FlashS |\n| **Multi-sample Integration** | Harmony, BBKNN, Scanorama, scVI |\n| **CNV Analysis** | InferCNVPy, Numbat |\n| **Spatial Registration** | PASTE, STalign |\n\n---\n\n## Documentation\n\n| Guide | Use this when... |\n|-------|------------------|\n| [Installation](docs/installation.md) | You need optional methods or a persistent Python environment |\n| [Docker](docs/docker.md) | You want a reproducible container runtime or local dependency resolution fails |\n| [Configuration](docs/advanced/configuration.md) | You need exact MCP client syntax or the runtime path model |\n| [Quick Start](docs/quickstart.md) | ChatSpatial is installed and you want the first successful analysis |\n| [Concepts](docs/concepts.md) | You need to choose an analysis strategy from a biological question |\n| [Examples](docs/examples.md) | You want copy-pasteable natural-language workflow prompts |\n| [Methods Reference](docs/advanced/methods-reference.md) | You need canonical tool names, method names, parameters, and defaults |\n| [Troubleshooting](docs/advanced/troubleshooting.md) | Setup, data loading, or analysis behavior is not working |\n| [Full Docs](https://docs.cafferyang.com/) | You want the complete documentation site |\n\n---\n\n## Reproducibility\n\nThe manuscript experiment scripts, small aggregate result tables, and\nsupplementary tables are versioned in [`reproducibility/`](reproducibility/README.md).\nLarge datasets, raw provider checkpoints, generated analysis directories, and\nmanuscript source files are intentionally kept outside Git. The reproducibility\nworkspace documents both the manuscript-era package baseline and the\ncurrent-checkout development workflow so historical evidence is not silently\nregenerated with a different ChatSpatial release.\n\n---\n\n## Citation\n\nIf you use ChatSpatial in your research, please cite:\n\n```bibtex\n@article{Yang2026.02.26.708361,\n author = {Yang, Chen and Zhang, Xianyang and Chen, Jun},\n title = {ChatSpatial: Schema-Enforced Agentic Orchestration for Reproducible and Cross-Platform Spatial Transcriptomics},\n elocation-id = {2026.02.26.708361},\n year = {2026},\n doi = {10.64898/2026.02.26.708361},\n publisher = {Cold Spring Harbor Laboratory},\n URL = {https://www.biorxiv.org/content/early/2026/03/01/2026.02.26.708361},\n journal = {bioRxiv}\n}\n```\n\nChatSpatial orchestrates many excellent third-party methods. **Please also cite the original tools your analysis used.**\n\n---\n\n## Contributing\n\nDocumentation improvements, bug reports, and new analysis methods are all welcome. See [CONTRIBUTING.md](CONTRIBUTING.md).\n\n
![\"ChatSpatial](\"assets/images/overview.jpg\")
\n