Genomic Intelligence

{
  "id": "aix_8a7869f7e3",
  "name": "Genomic Intelligence",
  "operator": "(unclaimed - source: registry-official · publisher: ai.genomicintelligence)",
  "description": "Hosted DNA language models: promoter, splice, enhancer, chromatin, expression, annotation",
  "depth": 2,
  "status": "unclaimed",
  "last_crawled": "2026-09-24",
  "missing_fields": [
    "pricing",
    "operator.identity"
  ],
  "skills": [
    "vector-search"
  ],
  "protocols": {
    "mcp": "https://mcp.genomicintelligence.ai/mcp",
    "a2a": null
  },
  "pricing": null,
  "regions": [
    "global"
  ],
  "languages": [
    "en"
  ],
  "reputation": {
    "tasks_completed": null,
    "dispute_rate": null,
    "p95_latency_ms": 2705,
    "uptime_30d": 1,
    "onchain_volume_30d_usd": null
  },
  "aix_score": 43,
  "verification": {
    "identity": "none",
    "health": "probe/24h",
    "pricing": "unknown",
    "last_check": "2026-09-24T23:00:30.443Z"
  },
  "pricing_model": "unknown",
  "links": [
    {
      "label": "homepage",
      "url": "https://genomicintelligence.ai/"
    }
  ],
  "profile": {
    "mcp_server": "gi-mcp",
    "mcp_version": "0.1.0a21",
    "tool_count": 15,
    "tools": [
      {
        "name": "list_models",
        "description": "List available models for a task.\n\n        Use to discover model ids before passing one as the `model`\n        argument to a predict tool. The same catalog is also available\n        as the resource `gi://models`.\n\n        Returns a FLAT object — {task, default_model, models: [...]} — not the\n        {data, meta} envelope the predict tools return. Each model carries a\n        `bio_spec`, whose useful fields are `request_max_bp` (the enforced\n        ceiling, 500,000 everywhere) and `context_window_bp` (what the model\n        reads in one step — compare your sequence length against it: a shorter"
      },
      {
        "name": "fetch_ensembl_sequence",
        "description": "Fetch a gene's reference sequence from Ensembl and store it.\n\n        Returns a handle ({ref, name, length, preview, ...}). Pass the\n        `ref` to predict_* tools — the bases stay server-side. For\n        expression, use fetch_gene_for_expression instead (it prepares\n        the TSS-centred window that model needs).\n        "
      },
      {
        "name": "fetch_region",
        "description": "Fetch a genomic region by coordinates from Ensembl and store it.\n\n        For \"find the genes in chr8:127,680,000-127,800,000\"-style requests:\n        resolves a coordinate range to reference sequence and returns a handle\n        ({ref, name, length, ...}) to pass to find_genes / predict_* — the bases\n        stay server-side. Plus strand by default, which is what the gene-finder\n        expects. For a gene by name use fetch_ensembl_sequence; for expression\n        use fetch_gene_for_expression.\n        "
      },
      {
        "name": "fetch_gene_for_expression",
        "description": "Fetch a gene's sequence prepared for expression prediction.\n\n        Resolves the gene's TSS via Ensembl and returns the exact\n        TSS-centred 9,198 bp window the expression model scores, as a handle\n        to pass to predict_expression(sequence_ref=...). Because the window is\n        exactly 9,198 bp, no `tss_index` is needed on that call.\n        "
      },
      {
        "name": "load_demo_sequence",
        "description": "Load a bundled demo reference sequence and return a handle.\n\n        The server ships one curated, task-correct positive control per task\n        (list them via the gi://sequences resource) — e.g.\n        `expression_hbb_k562` is a ready-to-use K562 expression window for\n        predict_expression. Stores the demo and returns a handle to pass to a\n        predict_* tool: no Ensembl fetch, no quota. Handy for smoke-testing a\n        prediction end-to-end.\n        "
      },
      {
        "name": "store_inline_sequence",
        "description": "Store a human-pasted sequence and return a handle to re-use it.\n\n        For a sequence you've already pasted into the conversation, this\n        gives back a short handle so you can run several tasks on it\n        without re-pasting the bases in each predict_* call. Note that the\n        full sequence still passes through the LLM on THIS call — it does\n        not save context on its own. For large sequences, prefer\n        fetch_ensembl_sequence / fetch_gene_for_expression / load_local_fasta,\n        which acquire the bases server-side and never round-trip them.\n\n        A line-wrapped FASTA"
      },
      {
        "name": "predict_promoter",
        "description": "Predict promoter regions (G0). 300–500,000 bp.\n\n        Returns the {data, meta} envelope: data.regions lists predicted\n        promoters with start/end/score.\n\n        300 bp is the task floor for every promoter model. The default\n        g0-promoter-2000bp scans a 2,000 bp context window, so a shorter\n        (but ≥300 bp) sequence is still scored — against a window padded out\n        to that size. Check the chosen model's bio_spec.context_window_bp via\n        list_models to know whether it saw real sequence or padding.\n        "
      },
      {
        "name": "predict_splice",
        "description": "Predict splice donor/acceptor sites (G0 BigBird). 100–500,000 bp.\n\n        The model reads a 15,000 bp context window, so anything shorter is\n        scored against a padded window — feed a whole transcript locus when you\n        can. It is also strand-specific, and the wrong strand fails silently and\n        plausibly — it returns sites at different positions, often still scoring\n        above 0.9, not the near-zero scores once documented here. Nothing in the\n        response flags it, so submit the transcript's own orientation\n        (fetch_region takes `strand`).\n        "
      },
      {
        "name": "predict_enhancer",
        "description": "Predict enhancer activity (G0 DeepSTARR). 50–500,000 bp.\n\n        50 bp is the task's admission floor (the API 422s below it), not a\n        statement about what the model reads: enhancer models score a 249 bp\n        context window, so 50–248 bp is accepted and scored against a padded\n        window. For a meaningful call, submit at least the 249 bp context.\n        "
      },
      {
        "name": "predict_chromatin",
        "description": "Chromatin annotation across 919 features (G0 DeepSEA). 200–500,000 bp.\n\n        The model reads a 1,000 bp context window; 200–999 bp is accepted and\n        scored against a padded window.\n        "
      },
      {
        "name": "predict_expression",
        "description": "Predict a gene's expression from a TSS-centred window.\n\n        Expression is cell-type-specific, so `description` (cell type /\n        assay context, e.g. 'K562 cell line') is REQUIRED — the API\n        rejects requests without it.\n\n        The model scores exactly 9,198 bp centred on the TSS (±4,599). Two\n        ways to supply that:\n\n        - A sequence of exactly 9,198 bp already centred on the TSS. No\n          `tss_index` needed — the midpoint is the only legal TSS.\n        - A longer locus, 9,198–500,000 bp, plus `tss_index`: the 0-based\n          offset of the TSS into it. The API cut"
      },
      {
        "name": "find_genes",
        "description": "Find genes (transcript intervals) in a genomic region (async, ~8-25s).\n\n        Takes 1,000–500,000 bp. The floor is the strictest of the scanning\n        tasks: gene finding needs a region, not a site. (Only expression's\n        9,198 bp is higher, and that is a fixed window rather than a minimum\n        region size.)\n\n        Gene-finding: detects transcript boundaries (TSS + PolyA) and returns\n        one interval per predicted transcript — start/end, strand, a\n        confidence score, and predicted TSS/PolyA positions (BED-style feature\n        intervals, not free-text notes). Use this fo"
      },
      {
        "name": "find_genes_and_predict_expression",
        "description": "Find genes in a sequence, then predict each gene's expression (composite).\n\n        Server-side chaining in ONE call: finds genes (transcript intervals,\n        with their TSS) in the sequence, then predicts expression off each\n        discovered TSS in the given experimental context. This is the right\n        tool whenever you want expression for a raw region or sequence — e.g.\n        \"find the genes in chr8:… and predict their expression in K562\".\n        predict_expression scores ONE TSS window and needs you to know where\n        that TSS is (either a pre-centred 9,198 bp window or a `tss_"
      },
      {
        "name": "get_job",
        "description": "Poll an async job once.\n\n        Returns the {data, meta} result if complete, a progress envelope\n        if still running, or an error envelope if it failed.\n        "
      },
      {
        "name": "list_jobs",
        "description": "List the caller's recent async jobs (also available as gi://jobs/recent)."
      }
    ],
    "profiled_at": "2026-09-24T23:00:30.443Z"
  },
  "unreachable": false
}