STRING Protein-Protein Interactions Scraper

Scrape protein-protein interaction (PPI) networks from STRING-DB — combined confidence scores plus all seven evidence channels, top interaction partners per protein, and name-to-STRING-id resolution. No API key required.

STRING is the reference database of known and predicted protein-protein interactions, integrating experimental data, curated pathway databases, genomic context and automated text-mining into a single confidence-scored network. This actor turns that into clean, tabular JSON you can drop into a spreadsheet, a graph tool, or a data pipeline.

Runs at roughly ~$0.50 per 1,000 records on the Apify platform.

---

What it does

Three modes:

Mode	What you get
network	Interactions among a given set of proteins (the edges inside your list).
partners	The top interaction partners of each input protein (network expansion).
resolve	Map protein names/gene symbols to STRING ids + annotation.

Evidence channels

Every interaction carries a combined_score and the seven STRING evidence channels, so you know why two proteins are linked:

Field	Channel	Meaning
neighborhood_score	n	Conserved genomic neighborhood
fusion_score	f	Gene fusion events
phylogenetic_score	p	Phylogenetic co-occurrence
coexpression_score	a	Co-expression
experimental_score	e	Experimental / biochemical data
database_score	d	Curated database (pathways, complexes)
textmining_score	t	Automated literature text-mining

The actor also computes strongest_channel (the channel with the highest sub-score) so you can filter for, say, experimentally-backed edges at a glance.

Score scale

STRING's JSON API returns confidence as floats in the range 0-1 (e.g. 0.999). Some STRING outputs multiply these by 1000. Each record includes a score_scale field reporting the observed range ("0-1" here), and the scraper detects and labels 0-1000 automatically if the API ever returns integer-scaled values.

---

Input

Field	Type	Default	Description
mode	select	network	network, partners, or resolve.
identifiers	array	["TP53","EGFR","MDM2","BRCA1"]	Protein names / gene symbols / STRING ids.
species	string	9606	NCBI taxonomy id (9606 = human, 10090 = mouse, 7227 = fly, 4932 = yeast).
partnerLimit	integer	20	Max partners per protein (partners mode only).
maxItems	integer	500	Max records to output (hard cap 5000).

Example

{
  "mode": "network",
  "identifiers": ["TP53", "EGFR", "MDM2", "BRCA1"],
  "species": "9606",
  "maxItems": 500
}

---

Output

Interaction record (network / partners):

{
  "type": "interaction",
  "protein_a": "MDM2",
  "protein_b": "TP53",
  "string_id_a": "9606.ENSP00000258149",
  "string_id_b": "9606.ENSP00000269305",
  "tax_id": "9606",
  "combined_score": 0.999,
  "neighborhood_score": 0.0,
  "fusion_score": 0.0,
  "phylogenetic_score": 0.0,
  "coexpression_score": 0.116,
  "experimental_score": 0.999,
  "database_score": 0.9,
  "textmining_score": 0.999,
  "score_scale": "0-1",
  "strongest_channel": "experimental",
  "source": "string-db",
  "scraped_at": "2026-08-11T00:00:00Z"
}

Resolved record (resolve):

{
  "type": "resolved",
  "query": "TP53",
  "string_id": "9606.ENSP00000269305",
  "preferred_name": "TP53",
  "tax_id": 9606,
  "taxon_name": "Homo sapiens",
  "annotation": "Cellular tumor antigen p53; Acts as a tumor suppressor ...",
  "source": "string-db",
  "scraped_at": "2026-08-11T00:00:00Z"
}

All fields are nullable.

---

Use cases

• Network biology — build and export PPI graphs for a gene set.
• Drug target discovery — find the partners of a target and rank edges by experimental evidence.
• Systems biology — assemble multi-protein interaction maps across species.
• Gene set analysis — resolve identifiers and pull interactions for a panel of genes from an experiment.

---

How this differs from our other biology scrapers

This account publishes several life-science data actors. Pick the one that matches the entity you care about:

• This actor (STRING-DB) → protein-protein interaction networks: which proteins bind/associate with which, with confidence scores and evidence channels. It is about edges between proteins.
• Reactome Pathways Scraper → curated biological pathways and reactions (ordered mechanistic steps), not scored interaction networks. Use Reactome when you want pathway membership and reaction hierarchy; use STRING when you want the interaction graph and its confidence.
• UniProt Scraper → deep annotation of single proteins (sequence, function, domains, GO, cross-refs). UniProt describes one protein; STRING describes relationships between proteins.
• AlphaFold Structures Scraper → predicted 3D structures of individual proteins.
• ChEMBL Scraper → bioactivity of small molecules against targets.

In short: UniProt = the protein, AlphaFold = its shape, ChEMBL = drugs hitting it, Reactome = the pathway it sits in, STRING = who it talks to.

---

Data source, licensing & etiquette

• Data © STRING consortium, distributed under Creative Commons Attribution 4.0 (CC BY 4.0). Please cite STRING (Szklarczyk et al., Nucleic Acids Research) when you use the data.
• The actor sends a caller_identity parameter with every request, as STRING requests. No API key is needed.
• Please keep request volumes reasonable and respect STRING's access guidelines.

This is an unofficial scraper and is not affiliated with or endorsed by the STRING consortium.