Skills string-database
🧬

string-database

Safe 🌐 Network access

Query protein interaction networks from STRING database

Also available from: davila7

Researchers need to understand protein-protein interactions to study biological systems and disease mechanisms. This skill provides direct access to STRING's comprehensive database of 59M proteins and 20B+ interactions across 5000+ species.

Supports: Claude Codex Code(CC)
🥉 73 Bronze
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "string-database". Get interaction partners for BRCA1 with high confidence

Expected outcome:

  • Top 10 BRCA1 interaction partners (confidence > 700):
  • BRCA2 - DNA repair protein, 990 confidence
  • RAD51 - DNA recombinase, 985 confidence
  • PALB2 - BRCA2 binding protein, 980 confidence
  • TP53 - Tumor suppressor, 750 confidence
  • CHEK2 - Checkpoint kinase, 720 confidence
  • Network contains 5 high-confidence interactions supporting BRCA1's role in homologous recombination repair.

Using "string-database". Perform functional enrichment on DNA repair genes

Expected outcome:

  • Significant GO Biological Process terms (FDR < 0.05):
  • DNA repair (GO:0006281) - 12 genes, FDR 1.2e-15
  • Double-strand break repair (GO:0006302) - 8 genes, FDR 3.4e-10
  • Cell cycle arrest (GO:0007050) - 6 genes, FDR 8.1e-8
  • KEGG Pathways: DNA replication (mmu03030) - 5 genes, FDR 0.0012
  • Top hub proteins: TP53, BRCA1, ATM form a highly connected module

Security Audit

Safe
v4 • 1/17/2026

The string-database skill is a legitimate bioinformatics tool for accessing protein-protein interaction data from the STRING database (string-db.org), a trusted ELIXIR resource. All 291 static findings are false positives: backticks in documentation are code formatting, HTTP requests target the official STRING API, file writes are for saving network images, and 'cryptographic' and 'reconnaissance' patterns are misinterpreted scientific terminology.

4
Files scanned
1,586
Lines analyzed
1
findings
4
Total audits

Risk Factors

🌐 Network access (3)
scripts/string_api.py:21-22 scripts/string_api.py:62-67 scripts/string_api.py:116-121
Audited by: claude View Audit History →

Quality Score

64
Architecture
90
Maintainability
87
Content
20
Community
100
Security
78
Spec Compliance

What You Can Build

Analyze differentially expressed genes

Upload a list of proteins from RNA-seq or proteomics experiments to identify enriched pathways and interaction networks.

Study protein function and interactions

Investigate specific proteins to discover interaction partners, visualize networks, and understand biological roles.

Build and analyze biological networks

Construct comprehensive interaction networks and test if proteins form significant functional modules.

Try These Prompts

Basic protein network 
Get the protein interaction network for TP53 in humans with medium confidence (400), including 5 additional nodes, and save as PNG image.
Functional enrichment 
Perform functional enrichment analysis on these proteins: TP53, BRCA1, ATM, CHEK2, MDM2. Show GO biological processes with FDR < 0.05.
Cross-species comparison 
Get interaction networks for p53 protein in human (9606) and mouse (10090) with high confidence (700), then compare the top 10 interactors.
Pathway analysis 
Analyze this DNA repair protein list: map IDs, get interaction network with 700 confidence, test PPI enrichment, perform GO/KEGG enrichment, and generate evidence-colored network image.

Best Practices

  • Always map protein identifiers first using string_map_ids for faster and more accurate queries
  • Use appropriate confidence thresholds: 400 for standard analysis, 700 for high-confidence interactions
  • Include species parameter (NCBI taxon ID) for networks with more than 10 proteins

Avoid

  • Do not query with more than 100 proteins in a single call - split large lists into batches
  • Avoid using very low confidence thresholds (< 150) without biological justification
  • Do not ignore species specification for multi-protein networks

Frequently Asked Questions

What is the STRING database?
STRING is a comprehensive database of protein-protein interactions covering 59M proteins and 20B+ interactions across 5000+ organisms, integrating data from experiments, databases, and text-mining.
Which species are supported?
STRING supports 5000+ species. Common ones include human (9606), mouse (10090), rat (10116), fly (7227), yeast (4932), and zebrafish (7955).
What confidence threshold should I use?
Use 400 for standard analysis, 700 for high-confidence interactions, and 150 for exploratory analysis. Higher thresholds give fewer but more reliable interactions.
How do I cite STRING?
Cite the latest STRING publication from https://string-db.org/cgi/about. The data is freely available under Creative Commons BY 4.0 license.
Can I analyze more than 100 proteins?
Yes, but split large lists into batches of 100 or fewer proteins to avoid timeouts and ensure optimal performance.
What is the difference between functional and physical networks?
Functional networks include all evidence types (recommended for most analyses). Physical networks only show direct binding evidence (for structural studies).

Developer Details

Author

K-Dense-AI

License

CC-BY-4.0

Repository

https://github.com/K-Dense-AI/claude-scientific-skills/tree/main/scientific-skills/string-database

Ref

main

File structure

📁 references/

📄 string_reference.md

📁 scripts/

📄 string_api.py

📄 SKILL.md