Question about --local behavior

[rferragud]

I would like to clarify how --local resolves the CARD database path.

When I run:
rgi load --card_json card.json --local
the database is created in the current directory. However, rgi main --local only works if I execute it from that same directory. If I run it from another location, RGI falls back to the Conda environment database path.

For HPC environments, what is the recommended way to use a database stored in a shared directory outside the Conda environment?

Thank you.

[raphenya]

@rferragud for HPC environments, create a virtual environment for RGI using tools like virtualenv, and then the rgi auto_load --clean --debug > load.log 2>&1 & will save the dbs in that environment. Example: the dbs will be in a directory like so:

../slurm/envs/rgi606/lib/python3.12/site-packages/app/_db/

[rferragud]

Thanks for the explanation.

In our case, we were hoping to keep the CARD database in a shared location outside the Python/virtual environment (for example a central directory on the HPC filesystem), so that multiple environments or workflows could use the same database without duplicating it inside each environment.

Is there currently a supported way to specify a custom database path for RGI (e.g. via a configuration option or environment variable), or does RGI always expect the database to live inside the environment where it was installed?

Thanks again for the clarification.

[raphenya]

@rferragud the latter. RGI always expect the database to live inside the environment where it was installed

[github-actions]

This issue is being automatically closed due to lack of activity for 30 days. While the CARD team does it's best to keep up, sometimes we are unable to do so. Please re-open this issue if you still need assistance.