Skip to workspace

Recent & saved

Recent runs

Persists in this browser (gene → top association).

API health

Not checked

Engine

Run a query to load version.

Literature watch Per-browser watch list (Firestore). Watch gene saves the symbol; Check watches diffs PubMed PMIDs vs last snapshot.

    Start here: try IL6 or BRCA1 — after Run query, expand Related evidence for Literature scope (human MeSH vs broad PubMed) and structure links. Graph: Knowledge graph. Deep-links: ?gene=IL6, ?gene=TLK2&literatureScope=broad. Hard refresh (⌘⇧R / Ctrl+Shift+R) if stale.

    Under ~10s: → evidence grid & pharma panel → open the Decision tab for ranked actions. Optional: second gene in Compare uses POST /api/clinical/compare (confidenceV2 ranking). Sales demo: ?demo=1 auto-runs IL6.
    Multi-target ranking (1–50 genes)

    Enter symbols separated by comma, space, or newline. Uses POST /api/clinical/portfolio — same scoring pipeline as single-gene decision.

    Evidence
    Emerging associations (predict)

    Heuristic trend view from live evidence — GET /api/clinical/predict. Loads after each query.

    Knowledge graph (Firestore)

    Graph vs evidence — This block is a cached knowledge subgraph from GET /api/clinical/graph/query?format=visualization (Firestore or demo). The Evidence column is a live engine run; node/edge counts can differ from the grid (e.g. IL6).

    Populated via SDP-CLIN-7001 backfill. API returns graphContext.retrievedAt when loaded.

    Enter a gene above and click Run query — the graph loads here (API + on-screen preview). Empty until a search completes.

    Decision recommendations

    Evidence-grounded actions appear after a successful query (requires decision API on your deployment).

    Confidence & source mix

    Populated from the decision API (includes BYOD when your uploads match the gene).

    My datasets

    Upload cohort or biomarker tables so evidence and decisions can incorporate your signals. This browser session sends X-SDP-Anon-Id automatically.

    1. Paste JSON array or CSV (header gene,disease,score) — symbols must match your query (e.g. IL6).
    2. Upload, then Run query on the same gene so engine + decision ingest BYOD.
    3. Use checkmarks below to include or exclude datasets from evidence, report preview, and analyze.