Source: https://github.com/aipoch/medical-research-skills

When to Use

• You need to search genes/proteins by keyword and species across common databases (e.g., Ensembl/UniProt/NCBI).
• You want to fetch detailed metadata for one or many Ensembl/UniProt/NCBI identifiers.
• You need to retrieve nucleotide/protein sequences for downstream analysis or pipelines.
• You want to obtain or predict protein structures (PDB download or AlphaFold prediction) from a sequence.
• You need to query expression resources (e.g., ARCHS4, CELLxGENE, Bgee) or run enrichment analysis (Enrichr).

Key Features

• Unified wrapper (scripts/wrapper.py) exposing multiple gget subcommands through a consistent interface.
• Gene/protein search and identifier resolution across multiple databases.
• Rich gene/protein information retrieval (annotations and metadata).
• Sequence retrieval for provided IDs.
• Structure workflows: PDB retrieval and AlphaFold-based prediction (optional plotting).
• Expression querying across popular expression atlases.
• Enrichment analysis via Enrichr.
• Backed by the upstream gget Python library.

Dependencies

• Python 3.9+ (recommended)
• gget (latest compatible version)
• pandas (latest compatible version)

Install:

uv pip install gget pandas

Example Usage

The skill is accessed via the unified wrapper script:

1) Search for genes by keyword

python scripts/wrapper.py search --keywords "insulin" --species "human"

2) Retrieve gene information by Ensembl ID

python scripts/wrapper.py info --ids "ENSG00000034713"

3) Fetch sequences by Ensembl ID

python scripts/wrapper.py seq --ids "ENSG00000034713"

4) Predict protein structure with AlphaFold (optional plotting)

python scripts/wrapper.py alphafold --sequence "MKWMFK..." --plot

Implementation Details

• Wrapper entrypoint: scripts/wrapper.py acts as a dispatcher that maps subcommands (e.g., search, info, seq, alphafold) to the corresponding gget library functions, normalizing CLI arguments and output behavior.
• Supported modules/functions:
  • ref: Download reference genomes/annotations.
  • search: Keyword-based gene/protein lookup (Ensembl/UniProt/NCBI).
  • info: Detailed gene/protein metadata retrieval for one or multiple IDs.
  • seq: Nucleotide/protein sequence retrieval for provided IDs.
  • structure: Structure retrieval (PDB) and AlphaFold prediction.
  • expression: Expression queries (ARCHS4, CELLxGENE, Bgee).
  • enrichment: Enrichr-based enrichment analysis.
• Notes on AlphaFold: The alphafold subcommand requires additional setup depending on the environment (e.g., model/data availability). Use --plot to request visualization output when supported.
• Further reference: See references/module_reference.md for detailed module-level documentation and parameters.