Onkydra
CHANGELOG

What we shipped this week.

A weekly cadence through August 17. One post per shipped surface: engineering, methodology, product, business. Honest about what worked and what didn't.

29 June 2026·3 min read

Research-use-only · the line goes on every report header, every footer, every JSON export.

We added a single-source-of-truth disclaimer module, wired it into the workspace report (top and bottom), and embedded the full text into every JSON export envelope. Customers, grant officers, and regulators get the same line.

Read post →

29 June 2026·4 min read

Leave-one-out cross-validation: where the joint model helps, and where it hurts.

We tested our own claim that a joint model beats a per-gene baseline. At n=49 it helps on Brier and hurts on log-loss. At n=222 it wins on both. We are reporting the mixed result, not the convenient half.

Read post →

29 June 2026·4 min read

Ewing sarcoma: the same engine on a second cancer, built and then shelved.

Cohort sampling run on a second cancer with sharply different molecular biology, anchored to two published Ewing cohorts fetched through the cBioPortal API. Ewing is built but shelved pending DMG proof-out and domain-expert review, and we report the validation result that came back honest rather than convenient.

Read post →