Experimental Metadata Example

A worked example accompanying the Experimental Metadata Specification.

zarr.json dca block

The following shows the dca block in the zarr.json of one deconvolved store from an example live 3D time-lapse dataset (cultured human iPSCs, live confocal fluorescence). It carries the required experimental floor mirrored alongside the channel and normalization-statistics metadata the block already holds. Because this is a time-lapse dataset, the conditionally-required acquisition.time_increment_s is included. The complete curated record, including the optional and recommended fields not shown here, lives in the sibling Parquet and is described below.

The experimental blocks sit directly in dca (no wrapper key), next to version / url / channels / normalization_statistics. Per-value provenance is the provenance map keyed by each value’s dotted field_path (so the values stay clean and provenance travels as a parallel layer). Every value records at least its source_kind; here all human, since the dataset was hand-curated, whereas an agent-extracted value would additionally carry a grounding citation. The channels and normalization_statistics shown are defined by the Channel Metadata Specification and Normalization Statistics Specification specs and their values here are illustrative.

{
  "attributes": {
    "dca": {
      "version": "0.2",
      "url": "s3://my-bucket/datasets/example/002_deconv.ome.zarr",
      "channels": [
        {"name": "H2B-mCherry", "index": 0, "description": "nuclei (H2B-mCherry reporter)"},
        {"name": "membrane-GFP", "index": 1, "description": "cell membrane (membrane-GFP reporter)"}
      ],
      "normalization_statistics": {
        "0": {"dataset_statistics": {"p1": 104, "p5": 112, "p95": 932, "p99": 1187}},
        "1": {"dataset_statistics": {"p1": 99, "p5": 108, "p95": 845, "p99": 1062}}
      },
      "study": {
        "source": "Example Imaging Lab",
        "study_type": ["FourDPlus", "TimeLapse", "FP"],
        "description": "Long-term 3D+t live fluorescence imaging of cultured human iPSCs expressing nuclear and membrane reporters, following cell growth and division.",
        "year_imaged": "2023",
        "dataset_name": "example",
        "release_date": "2023-05-01",
        "license": "CC-BY-4.0",
        "experimenter": [
          {"first_name": "Jane", "last_name": "Doe", "institution": "Example Imaging Lab"},
          {"first_name": "John", "last_name": "Roe", "institution": "Example Imaging Lab"}
        ]
      },
      "biosample": {
        "group_id": "example__002_deconv",
        "tissue_type": "cell culture",
        "organism": [{"ontology_term_id": "NCBITaxon:9606", "label": "Homo sapiens"}],
        "tissue": [{"ontology_term_id": "CL:0002620", "label": "induced pluripotent stem cell"}],
        "disease": [{"ontology_term_id": "PATO:0000461", "label": "normal"}],
        "development_stage": [{"ontology_term_id": "na", "label": "na"}]
      },
      "specimen": {
        "specimen_state": "live",
        "preparation_method": [{"ontology_term_id": "FBbi:00100027", "label": "adherent live cells"}]
      },
      "study_component": {
        "imaging_method": [{"ontology_term_id": "FBbi:00000251", "label": "confocal microscopy"}],
        "assay": [{"ontology_term_id": "EFO:0002909", "label": "microscopy assay"}]
      },
      "acquisition": {"time_increment_s": 600.0},
      "provenance": {
        "study.source": {"source_kind": "human"},
        "study.study_type": {"source_kind": "human"},
        "study.description": {"source_kind": "human"},
        "study.year_imaged": {"source_kind": "human"},
        "study.dataset_name": {"source_kind": "human"},
        "study.release_date": {"source_kind": "human"},
        "study.license": {"source_kind": "human"},
        "study.experimenter": {"source_kind": "human"},
        "biosample.group_id": {"source_kind": "human"},
        "biosample.tissue_type": {"source_kind": "human"},
        "biosample.organism": {"source_kind": "human"},
        "biosample.tissue": {"source_kind": "human"},
        "biosample.disease": {"source_kind": "human"},
        "biosample.development_stage": {"source_kind": "human"},
        "specimen.specimen_state": {"source_kind": "human"},
        "specimen.preparation_method": {"source_kind": "human"},
        "study_component.imaging_method": {"source_kind": "human"},
        "study_component.assay": {"source_kind": "human"},
        "acquisition.time_increment_s": {"source_kind": "human"}
      }
    }
  }
}

Additional metadata in the Parquet

The dca block above carries only the required floor. The sibling Parquet is the source of truth: it holds the complete record, including the recommended and optional fields below. Each also carries its provenance, keyed the same way by field_path; free-text additional_metadata values use <block>.additional_metadata.<name>, so all provenance has one home.

Recommended study fields. accession_id ("S-BIAD0000", an example BioImage Archive data deposit) and the related_publication entry (Doe et al. 2023, doi "10.0000/example.0000", the paper). The data deposit and the paper are distinct: the deposit identifier goes in accession_id, the paper DOI in related_publication.

Instrument and acquisition details (optional). The OME instrument set beyond the conditionally-required time_increment_s: microscope (an inverted spinning-disk confocal microscope), objective (63x, oil immersion), and imaging_environment (37 °C, 5% CO₂).

Processing variant. Imaging experiments often contain multiple data pre-processing steps to optimize segmentation, tracking, or other downstream outputs. This dataset contains raw, denoised, and deconvolved versions of the same Zarr stores; the store here is the deconvolved one, so its Parquet records ImageData.processing_variant = "deconvolved". The full derivation relationship is tracked in the Scientific Dataset Catalog as a Lineage Edge (transformed_from); processing_variant is the zarr-local label only.

Specimen growth conditions. The culture medium and substrate go in the existing optional Specimen.growth_conditions field, not additional_metadata: "mTeSR Plus medium on Matrigel-coated plates".

Free-text gap-fill (``additional_metadata``). Cell-line details with no first-class field go here, named to mirror the cross-modality cell-line guidance so they map cleanly if promoted later:

• Passage number (passage_number = "p32") — the culture passage at imaging time.

• Reprogramming method (reprogramming_method = "episomal") — how the iPSC line was derived. Donor age, reprogramming method, and passage number have no first-class field for cultured lines and live here per the cell-line guidance above.