AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

database-lookup

Queries documented public database APIs with explicit endpoints, filters, pagination, and provenance. Used when a scientific, regulatory, financial, or other database-backed fact must be retrieved reproducibly from a named source rather than inferred from general knowledge.

Original source
K-Dense-AI
License
MIT
Pinned revision
5f2bc0341354fd5590aef76a533ada1e1e5b4e62

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

Ask your assistant

Use Agentic Memory skills to read k-dense-ai--database-lookup at revision 5f2bc0341354fd5590aef76a533ada1e1e5b4e62, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 83

references/string.md

STRING REST API

Base URL

https://string-db.org/api

URL Pattern

/api/{output_format}/{method}
  • output_format: json, tsv, tsv-no-header, image, svg (not all formats for all endpoints)
  • method: endpoint name (see below)

Authentication

No API key required. All endpoints are public.

Key Endpoints

1. Resolve protein identifiers

Map protein names/identifiers to STRING internal IDs. Always do this first to get canonical STRING IDs.

GET /api/json/get_string_ids?identifiers={query}&species={taxid}

| Parameter | Type | Description | |-------------|--------|-------------| | identifier | string | Required. Protein name, gene symbol, or external ID. | | species | int | NCBI taxonomy ID (9606 = human, 10090 = mouse). Recommended to avoid ambiguity. |

Example:

https://string-db.org/api/json/get_string_ids?identifiers=TP53&species=9606

Response:

[
  {
    "stringId": "9606.ENSP00000269305",
    "preferredName": "TP53",
    "ncbiTaxonId": 9606,
    "taxonName": "Homo sapiens",
    "annotation": "Cellular tumor antigen p53; ..."
  }
]

2. Get interaction partners (network)

GET /api/json/interaction_partners?identifiers={proteins}&species={taxid}

| Parameter | Type | Description | |--------------------|--------|-------------| | identifiers | string | Required. Protein name(s). Use %0d (newline) to separate multiple. | | species | int | NCBI taxonomy ID. | | limit | int | Max number of interaction partners to return (per input protein). | | required_score | int | Minimum combined score (0-1000). Default: 400. Common thresholds: 400 (medium), 700 (high), 900 (highest). | | network_type | string | functional (default, all associations) or physical (predicted physical associations, which need not prove direct binding). |

Example:

https://string-db.org/api/json/interaction_partners?identifiers=TP53&species=9606&limit=10&required_score=900

Response:

[
  {
    "stringId_A": "9606.ENSP00000269305",
    "stringId_B": "9606.ENSP00000261842",
    "preferredName_A": "TP53",
    "preferredName_B": "MDM2",
    "ncbiTaxonId": 9606,
    "score": 0.999,
    "nscore": 0,
    "fscore": 0,
    "pscore": 0,
    "ascore": 0.93,
    "escore": 0.994,
    "dscore": 0.9,
    "tscore": 0.981
  }
]

Score channels: nscore (neighborhood), fscore (fusion), pscore (phylogenetic co-occurrence), ascore (co-expression), escore (experimental), dscore (database/curated), tscore (text mining).


3. Get network interactions between a set of proteins

GET /api/json/network?identifiers={proteins}&species={taxid}

| Parameter | Type | Description | |------------------|--------|-------------| | identifiers | string | Required. Protein names separated by %0d (newline-encoded). | | species | int | NCBI taxonomy ID. | | required_score | int | Minimum combined score (0-1000). | | network_type | string | functional or physical. | | add_nodes | int | Number of additional interactors to add (expands the network). |

Example — network among a set of proteins:

https://string-db.org/api/json/network?identifiers=TP53%0dBRCA1%0dATM%0dCHEK2%0dMDM2&species=9606&required_score=700

Returns all pairwise interactions among the input set.


4. Network image

GET /api/image/network?identifiers={proteins}&species={taxid}
GET /api/svg/network?identifiers={proteins}&species={taxid}

Returns a PNG image or SVG of the interaction network.

Example:

https://string-db.org/api/image/network?identifiers=TP53%0dBRCA1%0dMDM2&species=9606

5. Functional enrichment analysis

Perform Gene Ontology, KEGG pathway, and other enrichment analysis on a set of proteins.

GET /api/json/enrichment?identifiers={proteins}&species={taxid}

| Parameter | Type | Description | |--------------|--------|-------------| | identifiers | string | Required. Newline-separated (%0d) protein names. | | species | int | NCBI taxonomy ID. |

Example:

https://string-db.org/api/json/enrichment?identifiers=TP53%0dBRCA1%0dATM%0dCHEK2%0dCDK2%0dCDKN1A&species=9606

Response:

[
  {
    "category": "Process",
    "term": "GO:0006974",
    "description": "cellular response to DNA damage stimulus",
    "number_of_genes": 6,
    "number_of_genes_in_background": 781,
    "ncbiTaxonId": 9606,
    "inputGenes": "TP53,BRCA1,ATM,CHEK2,CDK2,CDKN1A",
    "preferredNames": "TP53,BRCA1,ATM,CHEK2,CDK2,CDKN1A",
    "p_value": 1.2e-12,
    "fdr": 5.6e-10
  }
]

Categories include: Process (GO Biological Process), Function (GO Molecular Function), Component (GO Cellular Component), KEGG, Pfam, InterPro, SMART, Keyword (UniProt), Reactome, WikiPathways, HPO (Human Phenotype Ontology).


6. Get protein annotations/info

GET /api/json/get_string_ids?identifiers={proteins}&species={taxid}

Maps arbitrary names to STRING IDs with annotation text.

Example:

https://string-db.org/api/json/get_string_ids?identifiers=CDK2%0dp53&species=9606

Response:

[
  {
    "queryIndex": 0,
    "queryItem": "CDK2",
    "stringId": "9606.ENSP00000266970",
    "ncbiTaxonId": 9606,
    "taxonName": "Homo sapiens",
    "preferredName": "CDK2",
    "annotation": "Cyclin-dependent kinase 2; ..."
  }
]

7. Get homology / best-hit in another species

GET /api/json/homology_best?identifiers={proteins}&species={taxid}&species_b={taxid_b}

| Parameter | Type | Description | |------------|------|-------------| | identifiers | string | Source protein(s). | | species | int | Source species. | | species_b | int | Target species for homolog lookup. |

Example:

https://string-db.org/api/json/homology_best?identifiers=TP53&species=9606&species_b=10090

8. PPI enrichment (is my set more connected than expected?)

GET /api/json/ppi_enrichment?identifiers={proteins}&species={taxid}

Example:

https://string-db.org/api/json/ppi_enrichment?identifiers=TP53%0dBRCA1%0dATM%0dCHEK2&species=9606

Response:

[
  {
    "number_of_nodes": 4,
    "number_of_edges": 6,
    "average_node_degree": 3.0,
    "local_clustering_coefficient": 1.0,
    "expected_number_of_edges": 1,
    "p_value": 0.000123
  }
]

Common Species Taxonomy IDs

| Species | Taxon ID | |---------|----------| | Homo sapiens (human) | 9606 | | Mus musculus (mouse) | 10090 | | Rattus norvegicus (rat) | 10116 | | Drosophila melanogaster (fruit fly) | 7227 | | Saccharomyces cerevisiae (yeast) | 4932 | | Caenorhabditis elegans (worm) | 6239 | | Danio rerio (zebrafish) | 7955 | | Escherichia coli K12 | 511145 | | Arabidopsis thaliana | 3702 |

Rate Limits

  • No published hard rate limit, but the API is intended for programmatic access at moderate rates.
  • Recommended: max 1 request per second.
  • For large-scale data downloads, use the flat-file downloads on the STRING website instead.
  • If you send too many requests, you may receive HTTP 429 or temporary blocking.
  • Multiple identifiers per request is strongly preferred over multiple single-identifier requests.

Error Handling

  • Returns HTTP 400 for malformed requests.
  • Returns HTTP 404 if no matching protein is found.
  • Empty JSON array [] if the query is valid but returns no results (e.g., no interactions above the threshold).
  • Include species parameter whenever possible to avoid ambiguous identifier resolution.

Current STRING API contract also supports regulatory networks. Pin the returned STRING version and network type; best sequence similarity does not by itself establish orthology.