STRING — Protein Interactions

Query the STRING database for protein–protein interaction networks. Use the string_db_search tool for structured PPI data across 5,000+ species.

TOOLstring_db_search

Retrieve interaction partners, scores, and network data for a protein of interest.

Parameter	Type	Description
protein *	string	Gene symbol or STRING ID (e.g. `TP53`).
species	integer	NCBI taxonomy ID (default `9606` = Homo sapiens).
limit	integer	Max interaction partners (default 10, max 200).
min_score	integer	Combined score threshold 0–1000 (default 400). Use 700+ for high-confidence.
workspace_id	string	Set to save network edge list to your workspace.

from smartsbio import SmartsBio

client = SmartsBio(api_key="sk_live_...")

result = client.tools.run(
    tool_id="string_db_search",
    input={
        "protein": "TP53",
        "species": 9606,
        "limit": 15,
        "min_score": 700,
    },
)
print(f"Found {len(result['interactions'])} high-confidence interactions")
for edge in result["interactions"][:5]:
    print(f"  TP53 ↔ {edge['partner']:<10}  score={edge['score']}  type={edge['type']}")

Download PPI network as edge list

from smartsbio import SmartsBio

client = SmartsBio(api_key="sk_live_...")
ws_id = client.workspaces.list()[0].id

result = client.tools.run(
    tool_id="string_db_search",
    input={
        "protein": "EGFR",
        "species": 9606,
        "limit": 50,
        "min_score": 500,
        "workspace_id": ws_id,
        "output_path": "networks/",
    },
)
# Download TSV edge list
tsv = client.files.download(result["output_path"], workspace_id=ws_id, dest="./networks/")
print(f"Network TSV saved to {tsv}")

Score channels

STRING returns per-channel evidence scores alongside the combined score.

Channel	Field	Evidence type
coexpression	coexpression_score	Co-expression from RNA-seq / microarray datasets.
experimental	experimental_score	Experimental protein-interaction assays (Y2H, AP-MS).
database	database_score	Curated pathway databases (KEGG, Reactome, etc.).
textmining	textmining_score	Co-mention in PubMed abstracts.
combined	score	Bayesian integration of all channels (use this for filtering).