gen3-metadata-templates

Turn a Gen3 schema into friendly Excel submission templates, then validate the filled-in workbooks — with every error pinned to the exact sheet, row, and column, in plain English.

Documentation

• Quickstart — the shortest path from a schema to a validated workbook.
• Concepts — nodes, links, paths, and submitter_id, for readers new to Gen3.
• Generating templates — path selection, node/column filtering, and what every part of the workbook means.
• Filling in a template — how linking, one-to-many, and multi-value cells work.
• Validating — the error types, the annotated copy, and exit codes.
• CLI reference — every command and flag.
• Library API — using the core from Python.
• Troubleshooting — common problems and fixes.

Install

pipx install gen3-metadata-templates      # installs the `g3mt` command

# everything of one kind (e.g. all your clinical data) in one workbook
g3mt generate schema.json --category clinical -o clinical_template.xlsx

# ...or a single node
g3mt generate schema.json sample -o sample_template.xlsx
# ...fill sample_template.xlsx in Excel...
g3mt validate sample_template.xlsx --schema schema.json --annotate checked.xlsx

The schema can be a local file or an http(s):// URL — e.g. a bundle published on GitHub:

g3mt generate \
  https://raw.githubusercontent.com/AustralianBioCommons/acdc-schema-json/refs/tags/v1.2.0/dictionary/prod_dict/acdc_schema.json \
  lipidomics_file -o lipidomics_file_template.xlsx

Why this exists

Submitting metadata to a Gen3 data commons means producing records that conform to a graph-shaped data model: nodes (subject, sample, file, …) linked parent-to-child, each with typed, sometimes-controlled properties. That is hard for people who aren't fluent in linked data — and mistakes are usually only discovered late, as opaque errors.

g3mt closes that gap from both ends:

• Generation produces a workbook a non-specialist can fill in confidently — parent links and controlled values are dropdowns, so most errors can't be made in the first place.
• Validation catches whatever still slips through and reports it as "Sheet subject, cell C4: 'ten' isn't a whole number" rather than a stack trace.

Input and output

Key features

• Whole-category templates — --category clinical gives a researcher every clinical sheet in one workbook, in the right fill order, without having to know the node names.
• One sheet per node, in fill order — parents before children, so links always resolve.
• Cross-sheet link dropdowns — a subject.submitter_id column on the sample sheet is a dropdown of the IDs you entered on the subject sheet. Reusing a parent ID across child rows is all "one-to-many" requires — no theory needed.
• Controlled-value dropdowns — enum and boolean fields become dropdowns.
• Self-documenting workbooks — required vs optional headers, a type/required hint row, per-column description comments, plus an Instructions sheet and a full Dictionary sheet.
• Path selection — when a node is reachable by more than one chain of parents, you choose which one the template covers.
• Precise, plain-English validation — errors located to the cell, rephrased for non-developers, with an optional highlighted copy of your file.
• Use it as a CLI or a Python library — the CLI is a thin shell over an importable core.

Requirements

• Python ≥ 3.9.5
• gen3-validator ≥ 2.1 (installed automatically)

Development

git clone https://github.com/AustralianBioCommons/gen3-metadata-templates.git
cd gen3-metadata-templates
pip install poetry
poetry install
poetry run pytest -vv

Before pushing, run the same checks CI does:

poetry run ruff check .          # lint
poetry run ruff format --check . # formatting
poetry run pytest -vv            # tests

To preview the documentation site locally:

poetry install --with docs
poetry run mkdocs serve      # then open http://127.0.0.1:8000

Continuous integration & releases

• CI (.github/workflows/build.yml) runs on every push and pull request to main: Ruff lint + format check, the test suite on Python 3.9–3.12, a g3mt --help smoke test, and a strict docs build.
• Publishing (.github/workflows/publish_pypi.yml) runs when a GitHub Release is published: it verifies the release tag matches the package version, then builds and publishes to PyPI. It needs a PYPI_API_KEY repository secret.
• TestPyPI (.github/workflows/publish_testpypi.yml) can be triggered manually from the Actions tab to publish a pre-release; it needs a TESTPYPI_API_KEY secret.

To cut a release:

poetry version <major|minor|patch>       # bump the version in pyproject.toml
git commit -am "chore: release vX.Y.Z"
git tag vX.Y.Z && git push --tags
gh release create vX.Y.Z --generate-notes # publishing the release runs the workflow

A note on 2.0

Version 2.0 is a full rewrite. The old Python API (generate_xlsx_template, make_node_template_pd, PropExtractor) has been replaced by the library and g3mt CLI documented here.

License

Apache 2.0 — see LICENSE.