Schema to describe content of the BDF Knowledgebase (BDFKB), including tools and related entities as well as the relationships between them.

https://ARPA-H-BDF.github.io/bdfkb-schema

• src/ - source files (edit these)
  • bdfkb_schema
    • schema -- LinkML schema (edit this)
    • datamodel -- Python classes. Located here to conform with python conventions so users can, for example from bdfkb_schema.datamodel.bdfkb_schema import ToolInfo
    • docs - Files for customizing Markdown docs
    • data - Data files used for testing
• project/ Auto generated by LinkML tools, typically these are committed to source control, but they are ignored here until there is a specific use for them.
• docs/ - Auto generated Markdown using GenDoc and files in project/docs
• wip/ - Work in progress, including sub-schemas

bdfkb-schema provides a CI stage for you to intgrate into your existing CI process, ensuring your tool is always up to date with the latest schema requirements.

To consume, add the following job to your GitHub CI Actions:

validate-tool:
  uses: ARPA-H-BDF/bdfkb-schema/.github/workflows/ci-validate.yaml@main
  with:
    python-version: '<YOUR PYTHON VERSION HERE (as a string)>'

The LinkML cookiecutter uses make so this is the easiest way to get started. Here are some examples:

• Test all things make test
• Test schema make test-schema
• Generate pydantic classes make gen-pydantic
• And more!

Under the hood, the Makefile is basically running python and using the linkml generator tools. So you can run the same commands directly if you choose.

• Run tests poetry run python -m unittest discover
• Generate mermaid diagram of schema gen-erdiagram -f mermaid src/bdfkb_schema/schema/bdfkb_schema.yaml > mermaid.md
• And more!

uv is a modern tool for python that aims to replace pip, pip-tools, pipx, poetry, pyenv, twine, virtualenv, and more. Learn more here https://github.com/astral-sh/uv

Some of team uses uv, so you might find uv bits scattered around this project. However, since the LinkML cookiecutter project still uses poetry, this project is not yet fully embracing uv. Install uv and then uv tool install linkml and explore at your own risk.

In order to test a schema change on the existing dataset, one recommended approach is:

1. Push your branch of changes from bdfkb-schema to GitHub
2. Go online and copy the raw file url for bdfkb-schema.yaml on your feature branch
3. Move to the bdfkb-data repository. In main.py, update the SCHEMA_PATH url to the url you just copied.
4. Run the main.py file using whichever runtime you prefer (uv, poetry, conda)

This project was made with linkml-project-cookiecutter, however much of what the cookie cutter provides is not needed here and thus removed. We opted to [mostly] stick with cookiecutter template so this would be familiar to someone already familiar with the LinkML project (and template).