Suggestion: Package scripts in a Conda repository

[bebatut]

Hi there!

I’ve been using the scripts in this repository, and they’ve been very helpful. I wanted to suggest packaging the scripts into a Bioconda repository to make them easier to install and manage, especially for users who want a consistent and reproducible environment.

By providing a Conda package, it would be much easier for us to install the scripts and keep them up to date. I'd be happy to help if needed!

Thanks for considering this suggestion!

Bérénice

[bluenote-1577]

hi @bebatut

I have had this request before, but it's a bit weird of a packaging situation.

The main contribution in this repo is (1) scripts, which can be conda packaged easily, and (2) a set of metadata files. It's a bit awkward to package metadata files in a conda repo, because you'll have to access them e.g. python script.py METADATA.TSV, but conda obfuscates where your files are located.

To me, it's simpler to have it all in a repo, where it's obvious where your metadata files are located.

Let mek now what you think.

[bgruening]

I agree, the metadata files are large and should not be shipped with a package. There are a few possibilities other packages with a similar problem are doing:

• using a post-processing hock in conda to download the "data" during tool installation
• offering a download option as part of the tool, e.g. python script.py --remote-metadata URL
• using an ENV variable to point to the "data" and let the user deal with downloading the "data" on its own

Do you think anyone of those will work?

[bluenote-1577]

@bgruening

These sound reasonable, but I'm trying to understand the motivation for this. Why is not cloning the repo sufficient, or something like wget https://github.com/bluenote-1577/sylph-utils/raw/refs/heads/main/virus/IMGVR_4.1_metadata.tsv.gz for downloading single metadata files?