FREE COHORT · RESEARCH USE ONLY

A free 200-profile simulated
H3 K27M DMG cohort, on us.

Name a target. We email you back 200 simulated co-mutation profiles sampled from a Gaussian-copula joint fitted to 60 real H3 K27M cases pooled from DKFZ and CPTAC, plus a one-page memo on the target you named. No call, no upsell pop-up, just the cohort.

Anchor n

60 real H3 K27M cases

DKFZ + CPTAC pediatric

Joint distribution

Gaussian copula · Ledoit-Wolf

Σ shrunk toward a sparse literature prior, λ = 0.577

Posterior draws

200 sampled profiles

Monte Carlo resolution, not sample size

Bootstrap CIs

Over the real n=60

Not over the 200 draws, the honest framing

Research-use only. Onkydra is a hypothesis-generation and prioritisation engine. It is not a diagnostic, not a medical device, and does not replace wet-lab validation. Same line that sits in every paying-customer report. Read it in full on the methodology page or in the FAQ.

AVAILABLE AT LAUNCH

The free DMG cohort ships when Onkydra opens to the public. Join the waitlist and we will send you your first cohort on launch day, no target-selection form needed.

Join the waitlist
WHAT THE EMAIL CONTAINS

One JSON manifest, one memo, no fluff.

DELIVERABLE 1

Cohort JSON manifest

200 Monte Carlo draws over molecular phenotype space, each with subgroup assignment (H3.1 vs H3.3), age, anatomic location, and a simulated co-mutation set. Sampled deterministically from a Gaussian-copula joint fitted to the 60 real K27M anchor cases.

DELIVERABLE 2

One-page memo on your target

A short memo on how the target distributes across the cohort's subgroups, with its carrier count and a 95% percentile bootstrap CI over the real n=60 anchor. Co-mutation rates are now conditioned on the subgroup, so ACVR1 tracks H3.1 as it does in the real cohort. Templated, deterministic, no LLM prose.

WHAT IT IS NOT

Not real patients · not a diagnosis

Every row is sampled from the joint, never copied from a real record. The memo is a research-use-only hypothesis. It does not recommend a treatment, identify a candidate patient, or replace wet-lab work.

WHY GIVE IT AWAY

You see the engine before you pay

The same sampler, the same critic, the same anchor cohort runs in the paid workspace. The free memo is one target. The workspace lets you iterate across targets. Pinned targets and re-runs on every ClinVar / DepMap delta are in build, not yet live.

Want the full workspace?

Iterate across targets and read the Critic's flags as they fire. Pinned targets and re-runs on every delta are in build, not yet live.

See pricing →