Claude Code skill for PlantGraph — Neo4j knowledge graph + REST API for plant gene annotations, interactions, co-expression, orthologues, GO terms, pathways, protein structures, and enrichment.
14.5M+ nodes, 47.9M+ relationships, 127 species, 132 node types, 149 relationship types. 140+ data sources, 42 real-time enrichment services. Built at UPSC (Umeå Plant Science Centre).
- Gene annotations (GO, PO, protein domains, MapMan bins, pathways)
- Protein-protein interactions, co-expression networks
- Cross-species orthologues and paralogues
- TF binding (DAP-seq validated)
- GWAS trait associations
- Histone marks, epigenetic data
- Single-cell expression profiles
- Nutrient/pathogen response data
- 290K+ publications
Arabidopsis thaliana, Brassica napus, Triticum aestivum, Oryza sativa (japonica + indica), Brachypodium distachyon, Zea mays, Populus trichocarpa, and 120+ more.
Not in graph: Hordeum vulgare, Brachypodium sylvaticum — orthologue chain strategies documented.
mkdir -p ~/.claude/skills/plantgraph-database
cp SKILL.md ~/.claude/skills/plantgraph-database/SKILL.mdInvoke /plantgraph-database in Claude Code. Provides:
- JWT auth setup and rate limits
- REST API reference: Graph, Enrichment, AI, Paper Evidence endpoints
- 42 enrichment sources (UniProt, STRING, AlphaFold, KEGG, InterPro, Expression Atlas, ...)
- AI: question generation, hypothesis generation, GNN predictions, chat
- Paper evidence: abstract/PDF scoring, claim verification (SSE)
- ID format rules per species (Arabidopsis
l.name, othersl.id) - Phytozome-to-PlantGraph ID conversion
- Cypher query patterns for common tasks
- Full schema: 132 node labels, 149 relationship types with counts
- R and Python integration with auth
MIT