{
  "registry_schema": "nmd-vcell-dataset-registry/1.0",
  "interface_build": "EA-20260817-57",
  "resource_release": "v1.2.0-measured-dmd-evidence",
  "evidence_freeze": "2026-08-03",
  "dataset_count": 5,
  "data_universe_pointer": "/resource/api/v1.1/data_universe.json",
  "data_universe_public_inventory_record_count": 126,
  "registry_scope": "task_audited_computational_role_subset_not_total_project_data",
  "required_metadata_axes": [
    "accession_and_permanent_url",
    "disease_species_tissue",
    "donor_age_sex_stage",
    "modality_and_observation_unit",
    "cell_count_or_sample_size",
    "perturbation_dose_time",
    "biological_and_technical_replicates",
    "batch_and_platform",
    "raw_processed_analysis_hash_chain",
    "leakage_grouping_and_benchmark_use"
  ],
  "direct_dmd_candidate_perturbation_dataset_count": 0,
  "direct_dmd_candidate_perturbation_dataset_state": "NOT_AVAILABLE_YET",
  "records": [
    {
      "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."
    },
    {
      "registry_object_schema": "nmd-vcell-dataset-registry-record/1.0",
      "dataset_id": "GSE293514",
      "accession": "GSE293514",
      "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE293514",
      "stable_url": "/resource/dataset/gse293514/",
      "json_url": "/resource/api/v1.1/datasets/gse293514.json",
      "title": "Human myoblast fusion CRISPR screen",
      "source_identity": {
        "accession": "GSE293514",
        "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE293514",
        "license_status": "Source repository and publication terms apply; verify at accession",
        "source_verification_required": true
      },
      "biological_context": {
        "disease_or_context": "Muscle context; not DMD rescue",
        "species": "Homo sapiens",
        "tissue_or_model": "human myoblast fusion screen",
        "cell_type": "Human myoblast",
        "donor_age_sex_stage_state": "source_publication_required"
      },
      "measurement_design": {
        "modality": "screen endpoint and validation workbooks",
        "observation_unit": "screen-level and selected validation objects",
        "cell_count_or_sample_size": "7,197 screen genes; 250 hits; 125 individually validated genes",
        "cell_count_state": "not_released_as_cell_population_truth",
        "endpoint": "Fusion screen, individual validation and grouped CROP-seq myogenesis state"
      },
      "perturbation_design": {
        "perturbation": "Knockout screen",
        "dose": "knockout screen; dose not harmonized in current object",
        "timepoints": "Publication endpoint",
        "direct_dmd_candidate_perturbation_outcome": false,
        "direct_dmd_candidate_perturbation_truth": false
      },
      "replication_and_batch": {
        "biological_replicate_state": "source_publication_required",
        "technical_replicate_state": "source_publication_required",
        "batch_platform_state": "source_publication_required"
      },
      "file_hash_chain": {
        "raw": "source_workbooks_checksummed_when_in_release",
        "processed": "derived_context_screen_objects_available",
        "analysis": "derived_tables_in_resource_manifest"
      },
      "benchmark_partitioning": {
        "leakage_grouping": "context_screen_only_not_training_truth",
        "permitted_benchmark_use": [
          "L3a_context_screen_interpretation"
        ],
        "donor_holdout_qualified": false,
        "disease_state_holdout_qualified": false,
        "prospective_holdout_qualified": false
      },
      "ai_readiness_state": "CONTEXT_SCREEN_ONLY",
      "current_ai_role": "Qualify muscle-context coverage and endpoint compatibility",
      "decision_blocker": "Not an independent candidate perturbation in a DMD model",
      "research_value": {
        "value_schema": "nmd-vcell-dataset-research-value/1.0",
        "what_it_proves": "Healthy-human-myoblast knockout perturbation effects can be read against a fusion endpoint and named validation context.",
        "what_it_cannot_prove": "It cannot establish that the same candidate rescues or worsens a DMD muscle phenotype.",
        "best_model_use": "Use for muscle-context coverage, endpoint compatibility and external-context stress tests.",
        "next_experiment_handoff": "Convert candidate coverage into a DMD/control myogenic Study Card with fusion and molecular endpoints.",
        "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."
    },
    {
      "registry_object_schema": "nmd-vcell-dataset-registry-record/1.0",
      "dataset_id": "GSE272233",
      "accession": "GSE272233",
      "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE272233",
      "stable_url": "/resource/dataset/gse272233/",
      "json_url": "/resource/api/v1.1/datasets/gse272233.json",
      "title": "DMD-locus CRISPR-correction transcript reference",
      "source_identity": {
        "accession": "GSE272233",
        "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE272233",
        "license_status": "Source repository and publication terms apply; verify at accession",
        "source_verification_required": true
      },
      "biological_context": {
        "disease_or_context": "Duchenne muscular dystrophy",
        "species": "Homo sapiens",
        "tissue_or_model": "DMD-locus correction reference",
        "cell_type": "Patient-derived and corrected cellular models",
        "donor_age_sex_stage_state": "grouped_sample_identities_available"
      },
      "measurement_design": {
        "modality": "bulk RNA expression",
        "observation_unit": "bulk sample",
        "cell_count_or_sample_size": "21 samples; 3 per reported group",
        "cell_count_state": "not_single_cell",
        "endpoint": "Bulk RNA expression and correction contrasts"
      },
      "perturbation_design": {
        "perturbation": "DMD-locus correction",
        "dose": "DMD-locus correction; candidate perturbation absent",
        "timepoints": "Study endpoint",
        "direct_dmd_candidate_perturbation_outcome": false,
        "direct_dmd_candidate_perturbation_truth": false
      },
      "replication_and_batch": {
        "biological_replicate_state": "3_per_reported_group",
        "technical_replicate_state": "source_publication_required",
        "batch_platform_state": "source_publication_required"
      },
      "file_hash_chain": {
        "raw": "upstream_accession_required",
        "processed": "processed_matrix_checksummed",
        "analysis": "derived_tables_in_resource_manifest"
      },
      "benchmark_partitioning": {
        "leakage_grouping": "orthogonal_reference_no_candidate_training",
        "permitted_benchmark_use": [
          "disease_correction_context_only"
        ],
        "donor_holdout_qualified": false,
        "disease_state_holdout_qualified": false,
        "prospective_holdout_qualified": false
      },
      "ai_readiness_state": "ORTHOGONAL_REFERENCE_ONLY",
      "current_ai_role": "Check disease-correction direction and context",
      "decision_blocker": "Candidate transcripts are observed but candidates were not perturbed",
      "research_value": {
        "value_schema": "nmd-vcell-dataset-research-value/1.0",
        "what_it_proves": "DMD-locus correction changes disease-context transcript programs in patient-derived/corrected models.",
        "what_it_cannot_prove": "It cannot isolate a displayed candidate's perturbation effect because those candidates were not perturbed.",
        "best_model_use": "Use as orthogonal DMD directionality and disease-correction context, not as candidate-training truth.",
        "next_experiment_handoff": "Use the corrected-vs-DMD contrast to choose endpoints for a separate candidate perturbation study.",
        "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."
    },
    {
      "registry_object_schema": "nmd-vcell-dataset-registry-record/1.0",
      "dataset_id": "GSE52529",
      "accession": "GSE52529",
      "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE52529",
      "stable_url": "/resource/dataset/gse52529/",
      "json_url": "/resource/api/v1.1/datasets/gse52529.json",
      "title": "Human myogenic differentiation time course",
      "source_identity": {
        "accession": "GSE52529",
        "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE52529",
        "license_status": "Source repository and package terms apply; verify at accession",
        "source_verification_required": true
      },
      "biological_context": {
        "disease_or_context": "Non-disease trajectory reference",
        "species": "Homo sapiens",
        "tissue_or_model": "primary skeletal myoblast differentiation",
        "cell_type": "Primary human skeletal myoblast",
        "donor_age_sex_stage_state": "not_suitable_for_unseen_donor_perturbation"
      },
      "measurement_design": {
        "modality": "single-cell expression",
        "observation_unit": "single cell",
        "cell_count_or_sample_size": "271 cells",
        "cell_count_state": "271_cells",
        "endpoint": "Single-cell expression, state and pseudotime"
      },
      "perturbation_design": {
        "perturbation": "None",
        "dose": "unperturbed",
        "timepoints": "0, 24, 48 and 72 hours",
        "direct_dmd_candidate_perturbation_outcome": false,
        "direct_dmd_candidate_perturbation_truth": false
      },
      "replication_and_batch": {
        "biological_replicate_state": "source_package_required",
        "technical_replicate_state": "source_package_required",
        "batch_platform_state": "source_package_required"
      },
      "file_hash_chain": {
        "raw": "bioconductor_source_package_checksummed",
        "processed": "derived_trajectory_objects_available",
        "analysis": "derived_tables_in_resource_manifest"
      },
      "benchmark_partitioning": {
        "leakage_grouping": "trajectory_reference_no_perturbation_holdout",
        "permitted_benchmark_use": [
          "G4_unperturbed_state_reference_only"
        ],
        "donor_holdout_qualified": false,
        "disease_state_holdout_qualified": false,
        "prospective_holdout_qualified": false
      },
      "ai_readiness_state": "TRAJECTORY_REFERENCE_ONLY",
      "current_ai_role": "Build an unperturbed human-myogenic state reference",
      "decision_blocker": "No perturbation-conditioned trajectory and no longitudinal DMD state",
      "research_value": {
        "value_schema": "nmd-vcell-dataset-research-value/1.0",
        "what_it_proves": "A physical-time human myogenic differentiation reference exists across 0, 24, 48 and 72 hours.",
        "what_it_cannot_prove": "It cannot simulate disease progression, aging, treatment response or perturbation-conditioned fate.",
        "best_model_use": "Use as an unperturbed trajectory/state reference and timing scaffold.",
        "next_experiment_handoff": "Add candidate perturbation at defined lineage times with matched DMD/control cells and frozen fate endpoints.",
        "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."
    },
    {
      "registry_object_schema": "nmd-vcell-dataset-registry-record/1.0",
      "dataset_id": "GSE133344",
      "accession": "GSE133344",
      "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE133344",
      "stable_url": "/resource/dataset/gse133344/",
      "json_url": "/resource/api/v1.1/datasets/gse133344.json",
      "title": "CRISPRa single and combinatorial calibration reference",
      "source_identity": {
        "accession": "GSE133344",
        "permanent_url": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE133344",
        "license_status": "Source repository and publication terms apply; verify at accession",
        "source_verification_required": true
      },
      "biological_context": {
        "disease_or_context": "Non-muscle method calibration",
        "species": "Homo sapiens",
        "tissue_or_model": "K562 method-calibration reference",
        "cell_type": "K562",
        "donor_age_sex_stage_state": "not_a_donor_generalization_set"
      },
      "measurement_design": {
        "modality": "single-cell perturbation expression",
        "observation_unit": "single cell",
        "cell_count_or_sample_size": "287 perturbations; 112 singles; 132 pairs",
        "cell_count_state": "public_method_reference",
        "endpoint": "Single-cell gene expression"
      },
      "perturbation_design": {
        "perturbation": "CRISPRa singles and pairs",
        "dose": "CRISPRa singles and pairs; dose not harmonized in current object",
        "timepoints": "Study endpoint",
        "direct_dmd_candidate_perturbation_outcome": false,
        "direct_dmd_candidate_perturbation_truth": false
      },
      "replication_and_batch": {
        "biological_replicate_state": "source_publication_required",
        "technical_replicate_state": "source_publication_required",
        "batch_platform_state": "source_publication_required"
      },
      "file_hash_chain": {
        "raw": "upstream_accession_required",
        "processed": "processed_feature_table_checksummed",
        "analysis": "derived_calibration_objects_available"
      },
      "benchmark_partitioning": {
        "leakage_grouping": "method_calibration_only_not_dmd_training",
        "permitted_benchmark_use": [
          "G7_method_calibration_reference_only"
        ],
        "donor_holdout_qualified": false,
        "disease_state_holdout_qualified": false,
        "prospective_holdout_qualified": false
      },
      "ai_readiness_state": "METHOD_CALIBRATION_ONLY",
      "current_ai_role": "Calibrate activation and combination task design",
      "decision_blocker": "No matched muscle or DMD activation response for current candidates",
      "research_value": {
        "value_schema": "nmd-vcell-dataset-research-value/1.0",
        "what_it_proves": "CRISPRa single and pair perturbation design can be calibrated in a public single-cell method reference.",
        "what_it_cannot_prove": "It cannot support muscle/DMD overexpression claims or infer combination benefit for current candidates.",
        "best_model_use": "Use for activation-design and pair-task calibration outside the DMD biological claim space.",
        "next_experiment_handoff": "Move activation or pair hypotheses into a muscle/DMD-specific assay before interpreting biological effect.",
        "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."
    }
  ],
  "boundary": "The registry distinguishes disease-context evidence, perturbation evidence and prediction training outcome."
}