§01 — Strata v2 · cross-indication opportunity engine
Start from the molecular state, not the disease.
Rank methylation-defined drug-response states that hold across the indication taxonomy — mechanism-anchored, tissue-confound-controlled, evidence-graded.
Research-use, GDSC cell-line proof-of-method. The positive control is confirmation of known biology (CDKN2A→palbociclib, published); novel leads are screen-derived hypotheses, evidence-graded — not clinical claims.
§02 — Agent transcript
Watch the agent reason across indications.
The agent scans every drug's own pathway genes for a methylation state whose response association is tissue-agnostic — surviving both tissue-adjustment and within-tissue replication (≥3 independent cancer types). It recovers the known 9p21 / MTAP → palbociclib positive control cold, then surfaces novel leads in the same unbiased pass. Below is a recorded run, played back step by step.
§03 — Featured opportunities
Positive control + two novel leads.
The same unbiased screen recovers known biology (positive control) and surfaces mechanism-anchored cross-indication hypotheses. Evidence licensing works by induction: when the known positive control is recovered cold, the novel leads in the same output inherit a license proportional to their evidence tier.
§04 — Negative control
The engine isn't forcing results.
§05 — Opportunity map · the portfolio
All L2–L3 candidates ranked.
Every entry has survived tissue-adjustment and/or within-tissue replication. These are not confirmations of prior literature — they are ranked hypotheses for follow-up, licensed by the screen's own positive control.
| Drug | Pathway | Marker | Level | Tissue-adj r | Pooled ρ | Within-sig tissues |
|---|
§06 — Evidence ladder · honest framing
How strong is the evidence?
Each opportunity is graded against the ladder below. v1 operated at L1 (interpretable marker, known biology). v2 targets L2–L3: tissue-agnostic associations surviving within-tissue replication.
v1 ≈ L1 · v2 targets L2–L3 — the positive control confirms L3 recovery; novel leads inherit that license.
Scope limit: tumor cell-line compartment only (GDSC). Not blood, liver, or patient data. All leads require orthogonal experimental validation before any clinical interpretation.