Reduce dependencies for anno_snakemake

[babayagaofficial]

• [ ] bakta is conda installable, so easy to set up conda environment for bakta rule. BUT it is dependent on a specific database, so we will have to either ask the user to download and setup the database, or come up with something that does it at installation of Pling

• [x] minimap2 is not installable through conda

[leoisl]

minimap2 is on bioconda: https://anaconda.org/bioconda/minimap2

[babayagaofficial]

aaah I didn't see that in the minimap2 documentation!! cheers!

[leoisl]

For bakta db, I've seen a tool (don't remember which one now) that would download a big DB (tens of GB) during installation. That was a bit of an issue as the installation would take hours and we didn't know what was happening. Also, in some clusters, software is hosted in a small but fast filesystem (like /hps/software in codon), in which large DBs should not be hosted. And the container would also be tens of GB, which is not ideal. I have a preference of adding a command to pling, something like pling --download-bakta-db or pling --prepare-annotation, whatever you want, which will download the bakta DB to a specified output dir. Especially because not everyone wants to use the annotation pipeline (e.g. in RH we need just the align pipeline). tbpore has a similar command: https://github.com/mbhall88/tbpore#download , but it downloads a minimap2 DB

[babayagaofficial]

Yes I think that would also be my preference -- this is actually what bakta itself does as well

[babayagaofficial]

all dependencies for anno_snakemake are removed, and I'll make a separate issue for the bakta-db

[iqbal-lab]

agree! or offer the user a chance to poiunt you to a predownloaded db?