search

bokulich-lab / q2-fondue

Functions for reproducibly Obtaining and Normalizing Data re-Used from Elsewhere
BSD 3-Clause "New" or "Revised" License
20 stars 6 forks source link

SRA Toolkit still not properly configured after installation #130

Closed misialq closed 2 years ago

misialq commented 2 years ago

It turns out that when installing fondue (+SRA Toolkit) by following the installation instructions , the SRA Toolkit appears not to be properly configured if it had never been installed on the system before.

Steps to reproduce:

  1. To mimic a clean OS, start up a virtual machine using e.g. Vagrant.
  2. Follow the installation instructions from the README.
  3. After activating the fondue environment, try to run any of the SRA Toolkit's executables. For example, run prefetch.

Expected behaviour: Prefetch's help is displayed.

Actual behaviour: An error message is displayed:

This sra toolkit installation has not been configured.
Before continuing, please run: vdb-config --interactive

Notes: This can be solved by running vdb-config -i, even without changing any of the settings. My suggestion would be to just leave those configuration steps up to the user after all - we can just add a line to the README saying what they need to run. This would also become an issue when installing through conda, so it would be easier if the users can just run it themselves. Also, having thought about it, maybe we should leave it up to them to decide where the cache location is rather then put it in the home directory by default (as it is now).

adamovanja commented 2 years ago

Thanks for documenting this. If we include vdb-config -i in the ReadMe, I suggest we include our recommendation of settings with it.

Could we not just include vdb-config -i without any setting changes within and a save exit (x) in the install-sra-tools.sh script? Then we could always mention in the ReadMe that the (set-up affine) user can further motify this by running vdb-config -i.

misialq commented 2 years ago

Yes, that would be my preferred option too, I think (I mean, just run it and exit). Then whoever wants to, they can adjust the cache or whatever else.

misialq commented 2 years ago

Oh wait, sorry, no. That would be possible if you run the script yourself but if you install from conda then its more complicated. That's why the best would be to let the user just run it themselves (they could still do it without altering any settings). And maybe it's not so bad to make them aware there actually are some things to be set for the SRA Toolkit (just by stating those in the readme) - that could potentially save us some issues in the future...

adamovanja commented 2 years ago

agreed 👍🏼 I would also add this link as a resource for interested users.