I think the move to MongoDB makes things much simpler and is a big improvement
Removing the local web page also makes things much simpler - I think this is fine as long as we can demonstrate that people can effectively reuse the "hub" web pages to browse their own models.
We probably do still need to include a quality metric, but I'm happy to postpone this work.
This PR is meant as suggestions for discussion - don't feel like you have to accept everything.
Major:
move docker setup/config away from code
include cli container in docker-compose
install GEMMI in cli container
include a Snakemake file that can run all steps within cli container
add generator to create metadata for Baker/Pfam models
add SPARQL client to get uniprot accession from gene name
add 'quick start' section to README
Minor:
tidy up logging
minor changes to cli command names (personal preference - feel free to ignore)
add example .env.example as template
TODO:
get cli container to run with user (non-root) permissions
Thanks for your work on this - general thoughts:
This PR is meant as suggestions for discussion - don't feel like you have to accept everything.
Major:
cli
container in docker-composecli
containercli
containerMinor:
.env.example
as templateTODO:
cli
container to run with user (non-root) permissions