Protian Entity — Human Protein & RNA Knowledge Graph

Protian Entity is an industrial-grade data product: curated L1 entity tables (Protein, RNA) with reproducible ETL, validation contracts, and QA artifacts.

Quick summary

• Primary entity: human Protein L1 table — data/processed/protein_master_v6_clean.tsv (v6 snapshot)
• RNA L1 artifacts are published as release assets (rna-l1-v1) with manifest.json and QA reports
• Validation tool: tools/kg_validate_table.py

Badges

• CI: .github/workflows/data-qa.yml runs data validation on PRs and pushes validation reports as artifacts

Contents

• data/processed/ — curated TSV L1 tables
• pipelines/ — ETL code and contracts
• docs/ — schema, data dictionary, quality gates, and admin guides
• tools/ — validation and QA helpers

Quickstart (validate current protein table)

# run the table validator and write a report
python3 tools/kg_validate_table.py \
  --contract pipelines/protein/contracts/protein_master_v6.json \
  --table data/processed/protein_master_v6_clean.tsv \
  --out build/validate/protein_master_v6_report.json

# open the JSON report
less build/validate/protein_master_v6_report.json

CI / Non-interactive usage

• CI runs must be non-interactive. Tools and scripts must support CI mode by either a --yes / --ci flag or CI=true environment variable.
• To run locally in non-interactive mode: CI=true ./scripts/run_full_build.sh --yes (if present).
• See docs/GITHUB_ACTIONS_SETTINGS.md for repository-level Actions configuration (Admins only).

Data release process

1. Run pipelines to produce L1 artifacts and manifest.json (row counts, checksums, git commit, build timestamp).
2. Run all validation contracts and attach reports.
3. Create a GitHub Release and upload artifacts + manifest.json.

Governance & contribution

• See CONTRIBUTING.md for branching, validation, and release checklists.
• Use PR templates and attach validation report artifacts for any change that modifies tables or contracts.

What stays out of git

• Large raw inputs and big output artifacts should be published as release assets or stored in object storage — do NOT commit >100MB files.

License

• MIT (see LICENSE)

Contact

• Repo owner: @hazelian0619 — open an issue or PR for changes

---

This README provides a concise starting point. For field-level schema and examples, see docs/DATA_DICTIONARY.md and data/processed/README.md.