TomoBabel / CETS metadata examples

Companion files for the TomoBabel / CETS background page, inspected on 2026-09-17.

File	Origin and scope
minimal-cets.yaml	Authored synthetic CETS Dataset, validated against core commit 0c8c4e3c7a2885f29caba551ffcaacde45c00101. Includes an image, coordinate frames, a point annotation, and a particle-map reference. File paths are fictional. Uses a corner origin, not the centered rigid-converter profile.
empiar-12104-tomogram.excerpt.yaml	Selected fields from an upstream CETS JSON example. Source values are unchanged; the referenced tilt series is omitted. Includes a real public MRC URL. The full historical source lacks 50 tilt-image IDs required by current core.
portal-10445.excerpt.yaml	Selected fields from recorded portal converter input. This is not a CETS document. Includes real image URLs and a GO membrane annotation. The selected image and annotation metadata describe different sampling grids.
provenance.yaml	Exact repository revisions, source URLs, source and excerpt checksums, and transformation descriptions.
validation.txt	Results of local metadata checks against the pinned CETS model.

The excerpts are reading aids, not complete submission or conformance fixtures. No volume arrays were downloaded for this page. The upstream JSON files remain linked from the provenance manifest; authorship and broader experimental context should be obtained from those sources and their associated archives.

Reading the synthetic example

The image has 4 Å sampling and a corner-origin physical frame. Its point at [40, 80, 24] Å therefore corresponds to array coordinates [10, 20, 6]. The particle map's coord_index: 0 refers to that first point. Its name deliberately makes no biological identification.

The example was validated with the patched Pydantic Dataset model from cets_data_model.models.models, then serialized to JSON and loaded again. YAML parsing or Pydantic acceptance does not check all referenced files, graph relationships, or spatial conventions. The real-data excerpts must not be mistaken for independently validated, complete CETS documents.

Reproducing the local example checks

From the LAMBDA repository root, create an isolated environment with the model revision and validation packages used for the checks. The distribution version is cets_data_model==0.1.0; the Git commit identifies the precise model snapshot. These commands use the committed patched model, so model generation is unnecessary.

uv venv --python 3.12.9 /tmp/cets-example-check
uv pip install --python /tmp/cets-example-check/bin/python \
  'cets_data_model @ git+https://github.com/TomoBabel/cets-data-models.git@0c8c4e3c7a2885f29caba551ffcaacde45c00101' \
  'pydantic==2.13.5' 'PyYAML==6.0.3'

Run the following to check all local YAML examples and their recorded checksums, then validate the synthetic Dataset and its JSON round-trip:

/tmp/cets-example-check/bin/python - <<'PY'
from hashlib import sha256
from importlib.metadata import version
from pathlib import Path

import yaml
from cets_data_model.models.models import Dataset

root = Path('src/docs/background/tomobabel-cets-examples')
manifest = yaml.safe_load((root / 'provenance.yaml').read_text())
for name, expected in manifest['validation']['packages'].items():
    assert version(name) == expected, (name, version(name), expected)
for entry in manifest['examples']:
    payload = (root / entry['local']).read_bytes()
    assert len(payload) == entry['bytes'], entry['local']
    assert sha256(payload).hexdigest() == entry['sha256'], entry['local']
    yaml.safe_load(payload)
data = yaml.safe_load((root / 'minimal-cets.yaml').read_text())
model = Dataset.model_validate(data)
reloaded = Dataset.model_validate_json(model.model_dump_json())
assert model.model_dump() == reloaded.model_dump()
print('Local YAML/checksum checks and synthetic Dataset JSON round-trip: passed')
PY

This reproduces the local example checks in validation.txt. The historical source comparisons and structural probes in that file are separate investigation results; the command above does not fetch their upstream fixtures or run those probes. The pinned source URLs and checksums are recorded in provenance.yaml.