This is a first draft of a metadata proposal meant to track important information about the raw data and files. It is not currently written as a schema, but as an example document with comments and questions scattered throughout. See FAANG and HumanCellAtlas for more formalised schemas along similar lines that we could possibly aim for but would require time and dedicated resource.

Proposal would be to have/manage these YAML files in a github repo (e.g. vgp-metadata) with a species/<species>/metadata.yaml file structure mirroring the AWS bucket. Advantages would be:

• If we organise a well defined schema and validator the data could be validated via continuous integration with travis or the like.
• Gives a place where internal and external users can point out mistakes and suggest changes or additions.
• Gives an automatic history of changes.

Adding the files to AWS could probably be set up to happen via a git-hook? But, we could just leave the AWS bucket for the data and the github repo for the metadata. Do we want these files to be the source of populating the website/DB-backend? Strategy would probably be OK for the ~260 ordinal species. Beyond that we will hopefully have obtained some dedicated resource to develop a proper metadata solution.