Quickstart

Get access

Graphomics is provisioned per organization. Request access and you'll receive:

• your API base URL and a MCP endpoint,
• an API key (sent to you securely), and
• your organization's initial roles and members.

There is no self-serve signup — access is granted and scoped to your organization. Keep your API key secret; treat it like a password.

Everything you connect to is issued by us when your organization is provisioned — there is no shared public endpoint. The examples in these docs use $GRAPHOMICS_API_URL (your REST base URL, ending in /api/v1), $GRAPHOMICS_MCP_URL (your MCP endpoint), and two credentials: $GRAPHOMICS_TOKEN for MCP, sent as Authorization: Bearer, and $GRAPHOMICS_API_KEY for REST, sent in the X-API-Key header. Substitute the values you were given.

Connect over MCP

Every core Graphomics capability is exposed as Model Context Protocol tools over streamable HTTP. Point an MCP-capable client — Claude Code, an agent framework, or your own MCP client — at your Graphomics MCP endpoint and pass your credential.

Example MCP client configuration (Claude Code / Claude Desktop style):

{
  "mcpServers": {
    "graphomics": {
      "type": "http",
      "url": "${GRAPHOMICS_MCP_URL}",
      "headers": {
        "Authorization": "Bearer ${GRAPHOMICS_TOKEN}"
      }
    }
  }
}

Once connected, start with whoami to confirm your organization and what your token can do. Then ground a question against the graph — for example, kg_search(q="colitis") to find a disease, then kg_disease_taxa to see which microbes shift in it, each with its studies. See the MCP tool reference for the full catalog.

Query the knowledge graph over REST

Prefer plain HTTP? The knowledge graph is a REST API. Authenticate with the X-API-Key header:

curl -s $GRAPHOMICS_API_URL/diseases/search/parkinson \
  -H "X-API-Key: $GRAPHOMICS_API_KEY"

Find the microbes associated with a disease:

curl -s "$GRAPHOMICS_API_URL/diseases/{disease_id}/taxa?limit=25" \
  -H "X-API-Key: $GRAPHOMICS_API_KEY"

See the API reference for authentication, conventions, and the full endpoint catalog.

A typical grounded workflow

1. Ground — search the graph for the entities relevant to your question (disease, taxa, metabolites).
2. Execute — run the analysis in Operon on your data (e.g. diversity + differential abundance), or in your own pipeline.
3. Contribute — write findings back into the graph as Assertions, linked to the run that produced them.
4. Trace — query the lineage of any entity to see the experiments, analyses, and decisions behind it.