Measure first, then calibrate
These outcomes may create the disease-relevant truth needed for evidence update and model calibration. No Prediction object is required.
DMD Minimum Perturbome
Flagship 01 · Stage A Panel Draft
DMD Minimum Perturbome
Twenty-four perturbation and calibration slots are now assigned to start a prospective DMD myogenic validation loop. Assignment is complete; experimental registration is not.
Panel Draft, not experiment. The target and control identities, stratification, prospective holdout roles and planning thresholds are explicit. Reagent sequences, the final cell system, blinded calibration, approvals and an immutable Study remain required before truth generation. A Prediction is required only for a later prediction-validation track.
Selection frame
The panel samples utility, uncertainty, mechanism and assay failure
| Stratum | Slots | Purpose |
|---|---|---|
| Muscle-context candidates | 6 | Immediate disease-context assay utility |
| High uncertainty | 4 | Test expression eligibility and abstention |
| Mechanism diversity | 4 | Avoid a one-pathway panel |
| Known myogenesis controls | 4 | Calibrate differentiation and fusion response |
| Negative controls | 4 | Estimate process and guide baseline |
| Toxicity controls | 2 | Calibrate viability detection |
24-slot assignment ledger
Every slot has an identity, purpose and prospective role
| Slot | Stratum | Target or control | Mechanism | Prospective role | Current evidence or control source |
|---|---|---|---|---|---|
| A-C01 | muscle context candidates | ADAM10 candidate gene | cell-surface proteolysis and signalling | prospective development role sealed | computational_replication; muscle TPM 3.1162; dependency flag false |
| A-C02 | muscle context candidates | CALR candidate gene | ER proteostasis and calcium handling | prospective development role sealed | computational_replication; muscle TPM 88.6307; dependency flag false |
| A-C03 | muscle context candidates | DDX19B candidate gene | RNA export and stress adaptation | prospective development role sealed | computational_replication; muscle TPM 7.22865; dependency flag false |
| A-C04 | muscle context candidates | MON1A candidate gene | endolysosomal trafficking | prospective holdout role sealed | muscle_assay; muscle TPM 8.05698; dependency flag false |
| A-C05 | muscle context candidates | MPHOSPH6 candidate gene | RNA exosome and transcript turnover | prospective development role sealed | computational_replication; muscle TPM 4.94586; dependency flag false |
| A-C06 | muscle context candidates | ZNF133 candidate gene | transcriptional regulation | prospective holdout role sealed | muscle_assay; muscle TPM 4.55781; dependency flag false |
| A-U01 | high uncertainty candidates | CPEB1 candidate gene | cytoplasmic RNA regulation | prospective development role sealed | verify_expression; muscle TPM 0.731389; dependency flag false |
| A-U02 | high uncertainty candidates | DNAAF3 candidate gene | dynein assembly | prospective holdout role sealed | verify_expression; muscle TPM 0.0755557; dependency flag false |
| A-U03 | high uncertainty candidates | EPS8L1 candidate gene | actin remodelling | prospective holdout role sealed | verify_expression; muscle TPM 0.236384; dependency flag false |
| A-U04 | high uncertainty candidates | ZFP69B candidate gene | transcriptional regulation | prospective development role sealed | verify_expression; muscle TPM 0.155208; dependency flag false |
| A-M01 | mechanism diversity candidates | DNM1 candidate gene | membrane fission and vesicle dynamics | prospective development role sealed | computational_replication; muscle TPM 1.39988; dependency flag false |
| A-M02 | mechanism diversity candidates | EHMT2 candidate gene | chromatin repression | prospective holdout role sealed | computational_replication; muscle TPM 6.40587; dependency flag false |
| A-M03 | mechanism diversity candidates | LMO2 candidate gene | transcription-factor complex assembly | prospective holdout role sealed | computational_replication; muscle TPM 15.237; dependency flag false |
| A-M04 | mechanism diversity candidates | NAGLU candidate gene | lysosomal glycan turnover | prospective development role sealed | risk_review; muscle TPM 8.66362; dependency flag true |
| A-P01 | known myogenesis or dmd controls | MYOD1 myogenesis positive control | myogenic commitment and fusion initiation | assay calibration control | PMID:33355126 |
| A-P02 | known myogenesis or dmd controls | MYOG myogenesis positive control | terminal myogenic differentiation | assay calibration control | PMID:33355126 |
| A-P03 | known myogenesis or dmd controls | MYMK fusion positive control | myoblast membrane fusion | assay calibration control | PMID:33355126 · PMID:26858401 |
| A-P04 | known myogenesis or dmd controls | MYMX fusion positive control | myoblast membrane fusion | assay calibration control | PMID:35642635 · PMID:33355126 |
| A-N01 | random or negative controls | NTC-A non targeting control | non-targeting guide control | assay calibration control | Reagent identity must be frozen with the selected vector system |
| A-N02 | random or negative controls | NTC-B non targeting control | non-targeting guide control | assay calibration control | Reagent identity must be frozen with the selected vector system |
| A-N03 | random or negative controls | SAFE-INTERGENIC-A safe intergenic control | safe intergenic targeting control | assay calibration control | Reagent identity must be frozen with the selected vector system |
| A-N04 | random or negative controls | SAFE-INTERGENIC-B safe intergenic control | safe intergenic targeting control | assay calibration control | Reagent identity must be frozen with the selected vector system |
| A-T01 | toxicity controls | POLR2A viability positive control | core transcription and viability | assay calibration control | PMCID:PMC6926425 |
| A-T02 | toxicity controls | RPA3 viability positive control | DNA replication and viability | assay calibration control | DOI:10.1038/s41419-025-07587-z |
Within each stratum, display order is the frozen slot order, not a target rank. Panel checksum: e18d58935bf32500c833f1e0b530c73e90ee64ea0da04b0e346010f0410d0aec
Two lifecycle roles
Truth generation and prediction validation remain separate.
Freeze a prediction, then unseal
These prospective roles become prediction-validation conditions only after an eligible model exists. Their outcomes cannot tune features, thresholds or models before custodian-authorized unsealing.
Candidate exclusion audit
Not assigned is a recorded decision, not disappearance
| Gene | Current action | Stage A state | Reason | Replacement eligibility |
|---|---|---|---|---|
| GFOD2 | hold pending external audit | Not assigned | HOLD PENDING EXTERNAL AUDIT | INELIGIBLE UNTIL HOLD RESOLVED |
| INTS13 | risk review | Not assigned | RISK OR CONTEXT REVIEW REQUIRED BEFORE REPLACEMENT | ELIGIBLE ONLY AFTER PREDECLARED SLOT FAILURE |
| RAC3 | risk review | Not assigned | RISK OR CONTEXT REVIEW REQUIRED BEFORE REPLACEMENT | ELIGIBLE ONLY AFTER PREDECLARED SLOT FAILURE |
| RNASEH2C | risk review | Not assigned | RISK OR CONTEXT REVIEW REQUIRED BEFORE REPLACEMENT | ELIGIBLE ONLY AFTER PREDECLARED SLOT FAILURE |
| RNF8 | risk review | Not assigned | RISK OR CONTEXT REVIEW REQUIRED BEFORE REPLACEMENT | ELIGIBLE ONLY AFTER PREDECLARED SLOT FAILURE |
| WDR4 | risk review | Not assigned | RISK OR CONTEXT REVIEW REQUIRED BEFORE REPLACEMENT | ELIGIBLE ONLY AFTER PREDECLARED SLOT FAILURE |
| ZNF236 | risk review | Not assigned | RISK OR CONTEXT REVIEW REQUIRED BEFORE REPLACEMENT | ELIGIBLE ONLY AFTER PREDECLARED SLOT FAILURE |
Factorial axes
Every outcome retains context, state, time and replication
DMD + matched control
DMD-derived human myogenic system plus matched healthy or isogenic-corrected system.
CRISPRi continuity
Two independent reagents per gene preserve continuity with the observed CRISPRi substrate. Down-regulation is a causal test, not a predicted therapeutic direction.
Myoblast to myotube
Proliferation, early differentiation, fusion and maturing myotube are retained as separate biological states.
Reagent · pair · batch
At least four donor or isogenic pairs are planned. Cells, fields and wells are measurements, not independent biological replicates.
Mechanism timeline
Early mechanism precedes late phenotype
T0Baseline
Control state and assay eligibility.
6–12 hMechanism
Target engagement, early transcription and signalling.
24–48 hTransition
Regulatory program, pathway activity and state proportion.
Day 5Function
Fusion index, morphology, DMD function and toxicity.
Endpoint and estimand
One primary interaction, explicit effect and equivalence margins
Primary endpoint
Percentage of analyzable nuclei contained within myosin-heavy-chain-positive syncytia with at least two nuclei.
Planning day 5; final timing requires blinded calibration.
DMD-specific estimand
For each candidate, (DMD perturbation − DMD non-targeting control) − (matched-control perturbation − matched-control non-targeting control) on fusion index.
Planning effect threshold
10 absolute percentage points for a minimally important fusion-index effect.
Status: provisional, not registered.
Planning equivalence margin
±5 absolute percentage points, with the 90% interval wholly inside the margin and engagement plus QC passing.
Replication and QC
A pathway delta cannot outrun engagement, viability or biological replication
At least 70% CRISPRi mRNA reduction
Below-threshold reagents produce an engagement failure, not a biological null.
At least 80% of non-targeting control
Below-floor functional changes are toxic responders unless a prespecified orthogonal assay separates toxicity from function.
500 analyzable nuclei per well
At least two qualified wells, NTC CV no more than 20%, and positive-control fusion shift of at least 15 percentage points.
4 minimum, 6 target pairs
Two differentiation batches and two same-direction qualified reagents are retained; blinded variance can increase precision but cannot lower the effect threshold.
Outcome ontology
Failure, null and heterogeneity are first-class results
robust_responderdmd_specific_responderdonor_variablestate_specifictoxic_respondernull_with_equivalence_margindiscordantqc_failureinconclusive
Primary-source anchors
Controls and endpoint definitions remain traceable
Human myotube formation is determined by the MyoD-Myomixer/Myomaker axis
Human myogenesis and fusion-control rationale
Structure-function analysis of myomaker domains required for myoblast fusion
CRISPR MYMK loss-of-function fusion control
Impaired activity of the fusogenic micropeptide Myomixer causes myopathy
Human MYMX fusion-defect evidence
CRISPR deletion of CTG expansions in patient-derived myogenic cells
Fusion-index measurement definition in human myogenic cells
A solid-phase transfection platform for arrayed CRISPR screens
POLR2A viability-control evidence
Genome-wide CRISPR screen using RPA3 as a cell-essential control
RPA3 viability-control evidence
Hard gates
Assignment is complete; registration is still blocked
| Gate | Requirement | Current state |
|---|---|---|
| GATE-A1 | freeze the 24 slot assignments and selection rules | Designed, not yet completedPANEL_DRAFT_COMPLETE_SIGNOFF_PENDING |
| GATE-A2 | freeze one primary endpoint and numeric minimally important effect | Defined in the research contractPROVISIONAL_DEFAULT_DEFINED_BLINDED_CALIBRATION_PENDING |
| GATE-A3 | freeze negligible-effect margin | Defined in the research contractPROVISIONAL_DEFAULT_DEFINED_BLINDED_CALIBRATION_PENDING |
| GATE-A4 | freeze sample size or pilot variance rule | Designed, not yet completedBLINDED_VARIANCE_RULE_DRAFTED_SIGNOFF_PENDING |
| GATE-A5 | freeze target-engagement, viability and imaging QC thresholds | Defined in the research contractPLANNING_THRESHOLDS_DEFINED_SYSTEM_CALIBRATION_PENDING |
| GATE-A6 | confirm biospecimen, institutional and laboratory approvals | No result yetBLOCKED_EXTERNAL_APPROVAL_RECORDS_MISSING |
| GATE-A7 | register an immutable Study before execution; freeze a Prediction only for conditions assigned to prediction validation | Designed, not yet completedAWAITING_REGISTERED_TRUTH_GENERATION_STUDY |
Boundary. This is a provisional 24-slot Panel Draft, not a registered experiment. Fourteen candidate genes and ten calibration controls are assigned, but reagent sequences, the final cell system, blinded calibration, approvals, an immutable truth-generation Study and measured outcomes remain absent. Prediction objects belong only to a later validation track. No assignment is a treatment, efficacy or validated DMD-response claim.