Upgrade execution · 2026-09-09
Every planned upgrade has an implementation state and a next accountable event.
Repository, data, research and workflow work is recorded separately from events that require an independent custodian, wet-lab measurement, external user or publication authority.
External validation
P48/P49 remains fail-closed until five people and six separations are real.
Conditional Monge passed 29/29 runtime tests. Appointments remain 0/5, separation 0/6, target access false and newly scored groups zero.
ADAM10 pilot
Design sensitivity is complete; Stage A measurement is the next scientific event.
All 480 scenarios use 20,000 simulations. The current two-pair design remains a calibration pilot, with 39 formal-lock inputs and zero observed outcomes.
Independent evaluation
A five-task formative protocol is ready for 5–8 external researchers.
Completion, time, navigation errors, claim-boundary errors and independent-unit errors are captured without treating project-team browser checks as external participants.
Research scope
Disease and spatial tasks now have separate biological axes and statistical units.
DMD, FSHD, DM1 and SMA retain disease-specific endpoints. GSE297388 representation stability is separated from GSE277637 nonspatial context and from DMD-versus-control claims.
Open task registry