decoupler is an established tool with a Bioinformatics Advances paper behind it — what is new here is asking for the analysis instead of writing the Python. That makes it most useful to two groups the project names directly: bench scientists who want the inference without the scripting, and agent developers who want to call decoupler from something else. Remote mode over HTTP is what lets the compute sit on a machine with the data on it.
A natural-language front end for decoupler, the published ensemble method for inferring biological activities from omics data. It covers the analysis steps around the inference too — reading the data in, clustering, differential expression and plotting.
- Reading and writing single-cell RNA-seq datasets
- Pathway activity and transcription-factor inference, which is decoupler's core capability
- Analysis steps around it — clustering and differential expression
- Plotting: violin plots, UMAP and t-SNE embeddings
Install the PyPI package decoupler-mcp with pip and confirm it runs. The client either launches the installed binary with the `run` argument over stdio, or you start it on a server with `--transport shttp --port 8000` and point clients at the /mcp path. Works with clients, plugins and agent frameworks that speak MCP.
One command — pip install decoupler-mcp