{
  "registry_object_schema": "nmd-vcell-dataset-registry-record/1.0",
  "dataset_id": "NMDVCELL-HEPG2-CRISPRI",
  "accession": "internal-release-substrate",
  "permanent_url": "/resource/candidates",
  "stable_url": "/resource/dataset/nmdvcell-hepg2-crispri/",
  "json_url": "/resource/api/v1.1/datasets/nmdvcell-hepg2-crispri.json",
  "title": "Observed HepG2 CRISPRi candidate responses",
  "source_identity": {
    "accession": "internal-release-substrate",
    "permanent_url": "/resource/candidates",
    "license_status": "Internal release licence pending institutional approval",
    "source_verification_required": false
  },
  "biological_context": {
    "disease_or_context": "Context-limited; not DMD",
    "species": "Homo sapiens",
    "tissue_or_model": "HepG2 cell line",
    "cell_type": "HepG2",
    "donor_age_sex_stage_state": "not_applicable_cell_line"
  },
  "measurement_design": {
    "modality": "pooled perturbation response aggregate",
    "observation_unit": "governed candidate-level aggregate",
    "cell_count_or_sample_size": "21 governed candidate aggregates",
    "cell_count_state": "component counts retained in governed candidate objects",
    "endpoint": "2,000-dimensional response vector"
  },
  "perturbation_design": {
    "perturbation": "CRISPRi single-gene",
    "dose": "CRISPRi single-gene; dose not modeled in released object",
    "timepoints": "Frozen endpoint aggregate",
    "direct_dmd_candidate_perturbation_outcome": false,
    "direct_dmd_candidate_perturbation_truth": false
  },
  "replication_and_batch": {
    "biological_replicate_state": "limited_release_diagnostics_only",
    "technical_replicate_state": "not_exposed_as_public_raw_delta_store",
    "batch_platform_state": "processed release substrate; exact raw batch log not public"
  },
  "file_hash_chain": {
    "raw": "raw_delta_store_not_public",
    "processed": "checksummed_release_objects_available",
    "analysis": "derived_tables_in_resource_manifest"
  },
  "benchmark_partitioning": {
    "leakage_grouping": "target_holdout_only_no_donor_holdout",
    "permitted_benchmark_use": [
      "G0_same_context_target_holdout"
    ],
    "donor_holdout_qualified": false,
    "disease_state_holdout_qualified": false,
    "prospective_holdout_qualified": false
  },
  "ai_readiness_state": "BASELINE_READY_SAME_CONTEXT_ONLY",
  "current_ai_role": "Train and audit the same-context baseline",
  "decision_blocker": "No matched muscle or DMD perturbation outcome",
  "research_value": {
    "value_schema": "nmd-vcell-dataset-research-value/1.0",
    "what_it_proves": "A released same-context perturbation substrate exists for governed candidate-level CRISPRi response auditing.",
    "what_it_cannot_prove": "It cannot prove DMD rescue, muscle-cell function, patient response or disease-context transfer.",
    "best_model_use": "Use as a permanent G0 same-assay baseline and leakage-control benchmark.",
    "next_experiment_handoff": "Pair the selected candidate with a matched DMD/control myogenic perturbation assay before any disease-response claim.",
    "claim_boundary": "Dataset value is task-specific; no Dataset Card alone is evidence of candidate efficacy, patient response or clinical utility."
  },
  "boundary": "Dataset registration does not convert context evidence into DMD perturbation-response outcome."
}