SingleR Remote Tool (MCP Server)#
TOU validation and deployment status (2026-08-16)#
R 4.4.3/SingleR 2.8.0, loopback discovery, and a deterministic query/reference call passed with 240 aligned labels. Public publication, cross-user isolation, representative reference/accuracy validation, broad concurrency, and recovery remain incomplete. Authenticated private Platform import and owner testing passed on 2026-08-16; public publication and independent-caller authorization/isolation remain untested.
Operation:
run_singler_annotateStart:
python -m tooluniverse.remote.singler.singler_toolEndpoint:
http://127.0.0.1:8029/mcpProvider configuration: set
TOOLUNIVERSE_REMOTE_DATA_ROOTand optionallyRSCRIPT_BIN; use approved relative query/reference inputs and reviewed labels.TOU check:
tu doctor --forward http://127.0.0.1:8029/mcp --jsonPrivate relay:
tu serve --share --forward http://127.0.0.1:8029/mcp --name validation-singler --workers 1
Non-loopback binding requires TOOLUNIVERSE_API_TOKEN; otherwise keep the server on loopback. The relay requires TOOLUNIVERSE_SERVICE_KEY. Public publication and independent-caller testing were not run.
New-user check: python scripts/remote_validation/setup_skill_preflight.py --implementation singler. Add --check-provider-env before launch, --live after launch, and --check-connect-prereqs before sharing. Live preflight checks exact MCP discovery only; it does not run or validate a model. The pinned relay SDK is not on PyPI and currently requires authorized GitHub repository access plus a configured SSH key, so a working local MCP server does not by itself prove that a new operator can share it.
The authenticated 2026-08-16 Platform matrix found all 30 private owner relays online and all 41 operations discoverable. This implementation was imported as unpublished owner draft(s), configured with a 120-second timeout, and invoked through /expert-sessions/{id}/test. Across the set, 38 unique operations passed return-schema and semantic validation; the three USPTO operations returned exact provider HTTP 403 and remain credential-blocked. Public publication, independent-caller authorization/isolation, broad saturation, and persistent supervision were not tested.
See the complete setup and verification guide.
Serves SingleR (Aran et al., Nature Immunology 2019) — reference-based single-cell annotation — as the ToolUniverse remote tool run_singler_annotate.
SingleR labels each cell (not each cluster) by correlating its expression against a labeled reference transcriptome and assigning the best-matching cell type. It is the standard automated alternative to manual marker-gene annotation.
Served remotely because the engine is R/Bioconductor (SingleR + celldex). The Python side reads the query .h5ad with scanpy and hands the matrix to the bundled singler_annotate.R over a MatrixMarket interchange — no zellkonverter/basilisk required (the heavy R↔Python bridge is avoided).
Two reference modes:
built-in — a
celldexpanel by name (HumanPrimaryCellAtlasData,BlueprintEncodeData,MonacoImmuneData,DatabaseImmuneCellExpressionData,NovershternHematopoieticData,ImmGenData,MouseRNAseqData), downloaded on the server.bring-your-own — a labeled reference
.h5ad(ref_adata_path+ref_labels_key).
Supply log-normalized expression for both query and reference (SingleR’s default assumption).
Deploy#
pip install -r requirements.txt # scanpy + scipy (Python side)
Rscript -e 'BiocManager::install(c("SingleR","celldex","jsonlite"))' # R side
python singler_tool.py # starts the MCP server on 127.0.0.1:8029
Rscript must be on PATH (or set RSCRIPT_BIN). Expose remotely only behind
TOOLUNIVERSE_API_TOKEN (SMCP bind guard).
Register in ToolUniverse#
Tool definition: src/tooluniverse/data/remote_tools/singler_tools.json
(type: RemoteTool).