Genomic Data Skills for Claude

This repository contains two complementary Claude skills for working with genomic data:

Skills

🧬 taxonomy-resolver

Purpose: Find and identify genomic data

Resolves organism names to NCBI taxonomy IDs, searches ENA for genomic data (FASTQ, assemblies), and retrieves BioProject details.

Key features:

Convert common names to scientific names with disambiguation
Search NCBI Taxonomy API for validation
Find FASTQ files, assemblies, and other data in ENA
Group results by BioProject with technical details
Intent-based filtering (RNA-Seq, WGS, ChIP-Seq, etc.)

APIs used: NCBI Taxonomy, ENA (European Nucleotide Archive)

🔬 iwc-workflow-recommender

Purpose: Recommend Galaxy workflows for analysis

Searches the IWC (Intergalactic Workflow Commission) workflow catalog and recommends appropriate workflows for genomic analysis.

Key features:

Search IWC workflow catalog by category
Match workflows to organisms and data types
Check data compatibility (single/paired-end, platform, etc.)
Provide TRS IDs for importing into Galaxy
ALWAYS recommends existing workflows (never writes custom code)

APIs used: IWC Workflow Manifest

Repository Structure

taxonomy-resolver/          (the repository)
├── taxonomy-resolver/      (skill: data discovery)
│   ├── resolve_taxonomy.py
│   ├── search_ena.py
│   ├── get_bioproject_details.py
│   ├── SKILL.md
│   ├── README.md
│   └── ...
├── iwc-workflow-recommender/  (skill: workflow recommendation)
│   ├── search_iwc_workflows.py
│   ├── SKILL.md
│   ├── README.md
│   └── ...
├── CLAUDE.md               (project-level guidance for Claude Code)
└── README.md               (this file)

Installation

For Claude Code

Copy both skill directories to your Claude skills folder:

cp -r taxonomy-resolver ~/.claude/skills/
cp -r iwc-workflow-recommender ~/.claude/skills/

For Claude.ai

Build each skill separately:

cd taxonomy-resolver && ./build.sh
cd ../iwc-workflow-recommender && ./build.sh

Then upload the generated ZIP files to Claude.ai (Settings → Features → Skills).

Testing

Each skill has its own test suite:

# Test taxonomy-resolver
cd taxonomy-resolver && bash test_skill.sh

# Test iwc-workflow-recommender
cd iwc-workflow-recommender && bash test_skill.sh

Network Requirements

Both skills require network access. Add these domains to your Claude environment's allowlist:

taxonomy-resolver:

api.ncbi.nlm.nih.gov
www.ebi.ac.uk

iwc-workflow-recommender:

iwc.galaxyproject.org

Philosophy

Both skills follow the principle: "Let the APIs do the work, Claude just orchestrates."

No data invention: Always defer to external APIs for authoritative information
No code generation: The IWC workflow recommender strictly prohibits writing custom analysis scripts
Disambiguation first: Never pass ambiguous inputs to APIs without clarification
Focused responsibilities: Each skill has one clear purpose

Workflow Example

A typical user interaction might use both skills:

taxonomy-resolver: "Find RNA-seq data for Plasmodium falciparum"
- Resolves organism name to taxonomy ID
- Searches ENA with RNA-Seq filter
- Returns BioProjects grouped by study with technical details
iwc-workflow-recommender: "What workflows can I use for this data?"
- Searches IWC catalog for Transcriptomics workflows
- Matches workflows to eukaryotic parasites
- Checks compatibility with paired-end Illumina data
- Recommends appropriate workflows with TRS IDs

License

Apache 2.0

taxonomy-resolver