NMD-VCell Research Workbench Module: Research Workbench · evidence-to-experiment workflow NMD = neuromuscular disorders
v1.0 candidate Frozen 25 Jul 2026 DOI pending
v1.0.0-database-resource Schema 1.1 Model ridge-safe-v2.3 Benchmark repeated-fold-v2.2 Build EA-20260729-15 v1.0 is a database and evidence-governance release; it is not a validated disease-prediction or clinical decision-support release.
Current evidence ceiling L2 observed HepG2 limited L3a context No independent DMD perturbation validation Open boundary

Evidence transition

Candidate evidence-state transition

Current ranking policy. Candidate records have no current numeric rank. They are displayed by required next action and then gene; historical scores and ranks remain frozen for reproducibility.
GFOD2 historical state
Rank 1

Historical computational result only.

GFOD2 current state

Provisional hold

External evidence update pending auditable quantitative import.

Claim ceiling

Status update only

Not a validated negative-effect or treatment claim.

Current status JSON · Status history JSON · External update registry JSON · Transition rules JSON

Does the resource still hold without GFOD2?

Yes, within the bounded benchmark claim. Removing GFOD2 from all 20 v2.3 realizations leaves 2,159 targets and preserves both RMSE conclusions. It does not establish that GFOD2 is biologically unimportant.
Corrected two-stage RMSE vs train mean

−0.000371 to −0.000224

The 95% interval remains below zero after exclusion.

Module-block RMSE vs train mean

−0.000550 to −0.000032

The dependence-aware interval remains below zero.

Module-level cosine endpoints

Still mixed

Both intervals continue to cross zero; no directional overclaim is added.

Remaining action queue

20 candidates

Four active action groups remain; no replacement top gene or numeric rank.

Machine-readable sensitivity object · Read the bounded report · Download all 20 gates


What clears a provenance-pending update?

Gate 1

Traceable and reproducible

Data, code, environment, source and independent units must resolve to frozen objects.

Gate 2

Context matched

Endpoint, biological setting and project independence determine whether evidence can speak to this claim.

Gate 3

Absence actually tested

For no material signal, the full confidence interval must fit inside a predeclared negligible-effect interval. P > 0.05 alone is insufficient.

Download the TSV template · Read the field dictionary · Open the protocol · Inspect the machine-readable contract

Governance boundary. The preflight is advisory. It cannot update GFOD2 or any other candidate automatically; an authorized manual transition remains required.

Bounded v2.4 matrix result. The complete deposited-matrix scan is structurally verified; exact gene identity and biological transfer remain blocked.

What was established

What remains unresolved

Machine-readable evidence

Summary JSON · Structure gates · Identity gates · Guide sensitivity gates