Anndata

The anndata skill enables AI agents to load, inspect, and manipulate AnnData objects, a standard format for storing single-cell genomics data containing both observations (cells) and variables (genes). It allows agents to perform subsetting, filtering, and basic statistical operations directly within the Python environment.

When to use it

• Loading .h5ad or .h5 files created by tools like Scanpy or Seurat for immediate analysis.
• Inspecting metadata structures such as obs, var, and uns dictionaries before processing biological data.
• Subsetting datasets to isolate specific cell populations or gene clusters based on defined criteria.
• Converting raw AnnData objects into Pandas DataFrames for visualization or custom statistical modeling.

Key capabilities

• File Loading: Reads standard AnnData file formats (.h5ad, .h5) efficiently.
• Structure Inspection: Accesses and displays obs (observations), var (variables), and uns (unstructured data) attributes.
• Data Manipulation: Supports subsetting rows and columns to filter cells or genes dynamically.
• Format Conversion: Transforms AnnData objects into Pandas DataFrames for broader compatibility.

Example prompts

• "Load the file 'my_data.h5ad' and display the first 10 rows of the observation metadata."
• "Filter this AnnData object to keep only cells where the 'cell_type' is 'T_cell' and show the resulting shape."
• "Convert the current AnnData object into a Pandas DataFrame containing only the gene expression matrix."

Tips & gotchas

Ensure your AI agent has access to the anndata Python package installed in its environment, as file paths must be absolute or relative to the working directory. Large single-cell datasets may require significant memory; verify hardware constraints before loading massive .h5ad files.