NMD-VCell Evidence & Experiment Workbench Audit disease-specific evidence, compare gaps, and freeze the missing test Module: Research Workbench · evidence-to-experiment workflow NMD = neuromuscular disorders
Search Data & API Current release

Current limit: disease modules compare measured states and intervention evidence, but none emits a calibrated patient-level prediction. DMD candidate validation remains prospective.

Research evidence only 21 prioritized candidate perturbation records 2,160 benchmark perturbation targets 250 observed human-myoblast fusion hits Prospective DMD candidate validation pending Boundary & release
21 Workbench candidates are surfaced for consideration, not ranked as validated targets 2,160 target-level perturbation responses form the frozen HepG2 benchmark substrate GSE293514 Stage B0 contributes 250 healthy-human-myoblast fusion hits and 125 individually validated genes No validated DMD perturbation prediction v1.2.0-measured-dmd-evidence Frozen 3 August 2026 Schema 1.1 Model ridge-safe-v2.3 Benchmark repeated-fold-v2.2 Build EA-20260804-49 DOI pending Open evidence boundary →

Disease evidence module · multidisease-evidence-v0.1.0

FSHD · Facioscapulohumeral muscular dystrophy

Sparse patient-state activation connected to a causal DUX4 time course.

Open module APICompare disease axes
Patient nuclei32,273
Donor-level inference1 FSHD2 + 1 control · descriptive
DUX4 time course24 samples · 4 wells/condition
14 h response6,986 genes · q<0.05, |LFC|≥1

Primary reanalysis interpretation

Positive, null and negative axes stay separate.

DUX4-target-high nuclei rise from 12–24% at day 3 to 51–56% at day 5 in this one-donor FSHD2 model, while direct DUX4 detection remains about 0.03–0.13%. The module therefore uses a target program, not DUX4 alone.

Measured evidence

Patient state, causal/intervention axis and ceiling in one figure.

Download PDF
FSHD disease evidence module showing measured data, intervention or correction analysis, null results and the claim boundary.

Datasets

Each object has a declared biological and statistical role.

MEASURED_DESCRIPTIVEGSE143452 ↗

Patient-derived FSHD2 and control myotube nuclei

8 technical samples; 32,273 nuclei; 1 FSHD2 and 1 control donor

Sparse DUX4-target state and heterogeneity
INDUCEDGSE205421 ↗

Inducible DUX4 human myoblasts

24 RNA-seq samples; 4 culture wells per condition; 2, 6 and 14 h

Causal timing and stress-intervention design
REGISTEREDGSE303359 ↗

FSHD myoblast membrane-injury response

8 single-cell libraries; large raw matrix registered, not reanalyzed in v01

Current membrane-repair / ferroptotic-stress extension

Candidate establishment

Anchors, markers, endpoints and intervention nodes are typed—not pooled into one score.

ObjectRoleEvidence statePermitted interpretation
DUX4etiologic_anchorINDUCEDDisease-initiating transcription factor; not a ranked output
DUXAstate_markerMEASURED+INDUCEDDownstream state marker; not therapeutic validation
ZSCAN4state_markerMEASURED+INDUCEDDownstream state marker; not therapeutic validation
LEUTXstate_markerMEASURED+INDUCEDDownstream state marker; not therapeutic validation
MAPK8intervention_nodeSOURCE_INTERVENTIONExperiment-design node; no patient efficacy claim
MAPK14intervention_nodeSOURCE_INTERVENTIONExperiment-design node; no patient efficacy claim
Supports
  • DUX4-target state stratification
  • induction kinetics
  • stress-intervention experiment design
Does not support
  • patient-calibrated efficacy
  • therapeutic ranking
  • donor-general disease prevalence from one donor per state

Machine-readable module

Summary, datasets, candidate logic and claim boundary are separate objects.

summary.json · datasets.json · candidate-logic.json · claim-boundary.json