ppi-string-query - SKILL.md Agent Skill

name: ppi-string-query description: > Query STRING database for protein-protein interactions with confidence scores. Use this skill when: (1) Finding interaction partners for a protein of interest, (2) Retrieving confidence scores for protein-protein interactions, (3) Building protein interaction networks for pathway analysis. license: MIT category: knowledge-retrieval tags: [string, protein-protein-interaction, ppi, network, interactions]

STRING Protein-Protein Interaction Query

Query the STRING database to retrieve protein-protein interactions with comprehensive confidence scores.

When to Use

Find interaction partners for a protein (by UniProt ID)
Retrieve confidence scores for PPIs (experimental, text mining, database)
Build protein interaction networks for pathway analysis
Identify potential protein complexes or functional modules

Workflow

Basic Query

from open_biomed.tools.tool_registry import TOOLS

# Query STRING for interaction partners
tool = TOOLS["ppi_string_request"]
results, _ = tool.run(uniprot_id="P04637")  # TP53

# Access results
for interaction in results:
    print(f"{interaction['partner_gene']}: {interaction['combined_score']}")

Custom Parameters

# High confidence interactions only, limit to 20
results, _ = tool.run(
    uniprot_id="P04637",
    species=9606,           # Human (default)
    required_score=700,     # High confidence (default)
    limit=20                # Max interactors
)

Parameters

Parameter	Type	Default	Description
`uniprot_id`	str	required	UniProt accession (e.g., P04637)
`species`	int	9606	NCBI taxonomy ID (9606=human)
`required_score`	int	700	Min confidence (150/400/700/900)
`limit`	int	50	Max interactors to return

Confidence Score Thresholds

Score	Level	Use Case
150	Low	Exploratory analysis
400	Medium	Balanced retrieval
700	High	Reliable interactions (default)
900	Highest	Very confident only

Expected Output

[
  {
    "query_protein": "TP53",
    "partner_string_id": "9606.ENSP00000340989",
    "partner_gene": "SFN",
    "combined_score": 0.999,
    "scores": {
      "experimental": 0.981,
      "text_mining": 0.859,
      "database": 0.75,
      "coexpression": 0.0,
      "phylogenetic": 0.0,
      "gene_fusion": 0.0,
      "neighborhood": 0.0
    },
    "ncbi_taxon_id": 9606
  }
]

Score Interpretation

Score Type	Source	Range
`combined_score`	Weighted combination	0-1
`experimental`	Wet-lab experiments	0-1
`text_mining`	Literature co-occurrence	0-1
`database`	Curated databases (BioGRID, etc.)	0-1
`coexpression`	Expression correlation	0-1
`phylogenetic`	Phylogenetic profiles	0-1
`gene_fusion`	Fusion events	0-1
`neighborhood`	Genomic proximity	0-1

Error Handling

Error	Solution
No interactions found	Lower `required_score` threshold
UniProt ID not recognized	Verify ID format (e.g., P04637)
Rate limited	Wait and retry; STRING allows ~5 req/sec
Wrong species	Check NCBI taxonomy ID

Common Organism IDs

Organism	Taxonomy ID
Human	9606
Mouse	10090
Rat	10116
E. coli	83333
S. cerevisiae	4932

References

examples/basic_query.py - Complete example script
references/score_details.md - Detailed score methodology
STRING API Docs: https://string-db.org/help/api/