This connects Claude to the STRING database's 59 million proteins and 20 billion interaction records across 5000+ species. You get six core functions: map gene names to STRING IDs, pull interaction networks with confidence scores, generate network visualizations, discover interaction partners, run GO/KEGG enrichment analysis, and test if protein sets are significantly connected. The confidence thresholds matter here: 400 is standard, 700 is high confidence, 900 is very stringent. It's built for systems biology work where you need to move quickly from a gene list to pathway context or from one protein to its functional neighborhood. The enrichment uses Fisher's exact test with FDR correction, so the statistics are proper.
npx skills add https://github.com/davila7/claude-code-templates --skill string-database