generated: '2026-08-01'
method: searched
source: https://missionbio.github.io/mosaic/manual/data_structure.html
format: HDF5 (.h5)
summary: >-
  Mission Bio has no OpenAPI to derive an entity graph from, but it does publish a
  real, documented data model: the Tapestri Pipeline emits a single .h5 (HDF5) file
  per sample, and the Mosaic Python library exposes it as a `Sample` (or `SampleGroup`
  for merged multi-sample files) containing one or more assay objects. Every assay
  inherits from `_Assay` and stores the same four-part structure: metadata, row
  attributes (per barcode/cell), column attributes (per feature), and layers
  (barcodes x features numeric matrices), plus an `info` store for arbitrary objects.
  The relationships below were read from the published data-structure and API
  reference pages — no field was invented.
entry_points:
  load: ms.load(h5path, raw=False, filter_variants=True, filter_cells=False, single=True)
  save: ms.save(sample, '/path/to/save/h5')
  import: import missionbio.mosaic as ms
  docs: https://missionbio.github.io/mosaic/manual/getting_started.html
entities:
- name: SampleGroup
  class: missionbio.mosaic.samplegroup.SampleGroup
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.samplegroup.SampleGroup.html
  description: Collection of Sample objects, returned when a merged multi-sample .h5 is loaded with single=False
- name: Sample
  class: missionbio.mosaic.sample.Sample
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.sample.Sample.html
  description: One Tapestri run/sample; container for the assay objects present in the .h5
  attributes: [file]
- name: _Assay
  class: missionbio.mosaic.assay._Assay
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.assay._Assay.html
  description: Abstract base for every assay; defines metadata / row_attrs / col_attrs / layers / info
- name: Dna
  class: missionbio.mosaic.dna.Dna
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.dna.Dna.html
  description: Single-cell DNA variant and genotype assay
  col_attrs: variants
  layers: [NGT, AF, DP, GQ, FILTER_MASK]
  layer_semantics:
    NGT: genotype call (0/1/2/3)
    AF: allelic fraction
    DP: read depth
    GQ: genotype quality
    FILTER_MASK: per-cell/variant filter mask
- name: Cnv
  class: missionbio.mosaic.cnv.Cnv
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.cnv.Cnv.html
  description: Copy-number assay, amplicon-resolved
  layers: [read_counts]
- name: Protein
  class: missionbio.mosaic.protein.Protein
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.protein.Protein.html
  description: Antibody-derived-tag immunophenotype assay
  col_attrs: antibodies
  layers: [read_counts, normalized_counts, scaled_counts]
- name: Rna
  class: missionbio.mosaic.rna.Rna
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.rna.Rna.html
  description: RNA assay (added in Mosaic 3.17.0 alongside the Tapestri GE pipeline)
- name: Ge
  class: missionbio.mosaic.ge.Ge
  docs: https://missionbio.github.io/mosaic/pages/missionbio.mosaic.ge.Ge.html
  description: Gene-expression assay produced by the Tapestri GE pipeline
relationships:
- {from: SampleGroup, to: Sample, kind: has_many, via: samples}
- {from: Sample, to: Dna, kind: has_one, via: dna}
- {from: Sample, to: Cnv, kind: has_one, via: cnv}
- {from: Sample, to: Protein, kind: has_one, via: protein}
- {from: Sample, to: Rna, kind: has_one, via: rna}
- {from: Sample, to: Ge, kind: has_one, via: ge}
- {from: Dna, to: _Assay, kind: inherits}
- {from: Cnv, to: _Assay, kind: inherits}
- {from: Protein, to: _Assay, kind: inherits}
- {from: Rna, to: _Assay, kind: inherits}
- {from: Ge, to: _Assay, kind: inherits}
- {from: _Assay, to: barcode, kind: has_many, via: row_attrs.barcode, note: rows are cells/droplet barcodes; the join key shared across all assays in a Sample}
- {from: _Assay, to: feature, kind: has_many, via: col_attrs.id, note: columns are variants (Dna), amplicons (Cnv) or antibodies (Protein)}
- {from: _Assay, to: layer, kind: has_many, via: layers, note: each layer is a barcodes x features matrix}
join_key:
  field: row_attrs.barcode
  note: >-
    The cell barcode is what makes Tapestri multi-omic — the same barcode indexes the
    Dna, Cnv and Protein layers of one Sample, so genotype and immunophenotype are
    joinable per cell.
algorithms_over_the_model:
- {name: COMPASS, class: missionbio.mosaic.algorithms.compass.COMPASS, purpose: copy number and mutation phylogeny}
- {name: GroupByGenotype, class: missionbio.mosaic.algorithms.group_by_genotype.GroupByGenotype, purpose: subclone grouping}
- {name: DnaAssignment, class: missionbio.mosaic.algorithms.dna_assignment.DnaAssignment, purpose: sample demultiplexing by DNA}
- {name: NSP, class: missionbio.demultiplex.protein.nsp.NSP, purpose: protein normalization}
- {name: SPARC, class: missionbio.demultiplex.phylogeny.sparc.SPARC, purpose: phylogeny reconstruction}
subway: null
x-evidence:
  fetched: '2026-08-01'
  urls:
  - https://missionbio.github.io/mosaic/manual/data_structure.html
  - https://missionbio.github.io/mosaic/manual/getting_started.html
  - https://github.com/MissionBio/mosaic/tree/master/docs/pages
  http_status: 200