Celltypist Catalog Tools#

Configuration File: celltypist_catalog_tools.json Tool Type: Local Tools Count: 2

This page contains all tools defined in the celltypist_catalog_tools.json configuration file.

Available Tools#

CellTypist_get_model (Type: CellTypistCatalogTool)#

Retrieve metadata for one CellTypist model by filename, including its training description, numbe…

CellTypist_get_model tool specification

Tool Information:

  • Name: CellTypist_get_model

  • Type: CellTypistCatalogTool

  • Description: Retrieve metadata for one CellTypist model by filename, including its training description, number of resolved cell types, version, source DOI, and download URL. Filenames look like ‘Immune_All_Low.pkl’ and are returned by CellTypist_search_models. This returns catalog metadata only; running annotation requires the celltypist Python package and an expression matrix. This tool returns catalog metadata only. To actually annotate cells, use the remote tool run_celltypist_annotate, which runs a chosen model against an expression matrix.

Parameters:

  • filename (string) (required) Model filename, e.g. ‘Immune_All_Low.pkl’. Case-insensitive.

Example Usage:

query = {
    "name": "CellTypist_get_model",
    "arguments": {
        "filename": "example_value"
    }
}
result = tu.run(query)

CellTypist_search_models (Type: CellTypistCatalogTool)#

Search the CellTypist catalog of pre-trained models for automated single-cell type annotation, co…

CellTypist_search_models tool specification

Tool Information:

  • Name: CellTypist_search_models

  • Type: CellTypistCatalogTool

  • Description: Search the CellTypist catalog of pre-trained models for automated single-cell type annotation, covering immune compartments, developmental atlases, and individual tissues. Filter by keyword against model descriptions, e.g. keyword=’lung’. Returns model filename, what it was trained on, how many cell types it resolves, and a download URL. Use this to choose the right classifier for a tissue; use PanglaoDB_* or CellMarker_* instead if you want literature-curated marker genes for a named cell type. This tool returns catalog metadata only. To actually annotate cells, use the remote tool run_celltypist_annotate, which runs a chosen model against an expression matrix.

Parameters:

  • keyword ([‘string’, ‘null’]) (optional) Case-insensitive filter on model description or filename, e.g. ‘lung’, ‘immune’, ‘fetal’, ‘brain’.

  • min_celltypes ([‘integer’, ‘null’]) (optional) Only return models resolving at least this many cell types, e.g. 50.

  • limit ([‘integer’, ‘null’]) (optional) Maximum models to return (default 25, max 100).

Example Usage:

query = {
    "name": "CellTypist_search_models",
    "arguments": {
    }
}
result = tu.run(query)