Contributing to sdrf-annotated-datasets

April 18, 2026 · View on GitHub

Thank you for helping improve public proteomics metadata. This repository holds SDRF annotations for ProteomeXchange (and related) datasets. The SDRF specification itself lives in bigbio/proteomics-sample-metadata.

Who this is for

Contributions are welcome from:

  • Community annotators and tool authors
  • Members of the SDRF / HUPO-PSI ecosystem
  • ProteomeXchange partners and curators interested in re-annotation of public experiments

The goal is to advance consistent, validated sample-to-data metadata so datasets can be reused, integrated, and reprocessed with less ambiguity.

Repository layout

Add or update files under:

datasets/{ACCESSION}/{ACCESSION}.sdrf.tsv

Use the public accession as the folder name (PXD…, MSV…, PMID…, etc.). If a project needs more than one SDRF file, keep them under the same accession folder and use descriptive suffixes (see existing examples).

Sandbox (sandbox/)

If an accession is not yet valid (for example it fails parse_sdrf validate-sdrf, or the file is an empty placeholder), keep it under sandbox/ instead of datasets/ until it is fixed. See sandbox/README.md. CI only blocks merges on validation failures under datasets/.

Pull request workflow

  1. Fork this repository and create a branch from dev (or main if that is the integration branch you are targeting).

  2. Add or edit SDRF files only under datasets/.

  3. Validate locally before opening a PR (same command CI uses):

    pip install --upgrade "git+https://github.com/bigbio/sdrf-pipelines.git@main"
parse_sdrf validate-sdrf --sdrf_file "datasets/YOUR_ACCESSION/YOUR_FILE.sdrf.tsv" --use_ols_cache_only

  4. Open a pull request with a short summary of the dataset, what changed, and how you validated it (manual review, tool output, etc.).

  5. CI will re-run validation on changed .sdrf.tsv files using sdrf-pipelines from GitHub main.

Human and agentic contributions

Manual curation and automated or agent-assisted annotation are both welcome. For agentic annotation, prefer sdrf-skills as the main environment: it bundles the recommended agent skills, editor/coding assistant rules, and repeatable workflows on top of sdrf-pipelines validation.

If a contribution was produced or assisted by an LLM or other agent:

  • Prefer small, reviewable pull requests (one accession or a coherent batch).
  • Cite sources in the PR description (repository record, publication, PX landing page) so reviewers can verify organism, condition, and file mapping.
  • Do not invent sample identifiers or file names: align rows with the actual record in the public archive.
  • Run parse_sdrf validate-sdrf locally and fix errors before opening the PR.
  • If ontology or term choices are uncertain, say so explicitly and prefer terms that match the published experiment description.

See README.md for a concise “agentic annotation” checklist.

Code of conduct

Be respectful and constructive. For substantive specification debates, use the proteomics-sample-metadata issue tracker.

License

By contributing, you agree your contributions are licensed under the same terms as this repository (LICENSE).