Setting environnement variables about scib-pipeline HOT 9 CLOSED

Hi. I've found the origin of my problem. I had an old .Rprofile file in my home directory which directed the $R_LIBS to the wrong library. I don't know why i didn't try this first but i realised it after using .libPaths() in R in the scib-R4.0 environment and seeing that the conda R library i wanted was in 2nd position. After removing this file from my home directory i was able to set the environments up with out any troubles.
Sorry I haven't thought of checking this before. Thank you so much for taking the time to help me!

from scib-pipeline.

Hi @t-nol ,

That's correct, the .libPaths() needs to be the conda path in order to work with the conda environment. Just to make sure, are you replacing <conda-prefix> with your actual environment path? And do you get any errors when you set the environment variables or activate/deactivate the environments?

from scib-pipeline.

@mumichae
Yes I did replace <conda-prefix> by the environment path. Both activate.d/env_vars.sh and deactivate.d/env_vars.sh files exist and contain the same code as env_vars_*.sh files provided in the envs/ directory.
No i haven't seen any errors while activating/deactivating the environments.

from scib-pipeline.

Sorry for the late reply. I updated the repository to include a single-command installation script and renamed the environment names. If you're still setting up the environments, I switch to the latest update.

To your issue: could you also double-check which R instance you're using (I assume you also have a system-wide R installation that might be competing with the conda installation)

e.g.

conda activate scib-R4
which R

or, if you use the updated version:

conda activate scib-R4.0
which R

from scib-pipeline.

No problem, thank you for taking the time.

I've been busy myself and haven't managed to set up the environments. There is a system wide version of R installed so this might be it.
I will try using the updated version, in the meantime here is what i get for the scib-R4 env :

conda activate scib-R4
which R
/CONDAS/users/tnoel/scib-R4/bin/R

from scib-pipeline.

So.. I've tried installing the updated version with the single-command installation script. Both environnement were created but I still get this error after :

Install R packages through R
scib pipeline environment...
Erreur : Failed to install 'kBET' from GitHub:
  package ‘rprojroot’ was installed before R 4.0.0: please re-install it
Exécution arrêtée
ERROR conda.cli.main_run:execute(41): `conda run Rscript -e remotes::install_github('theislab/kBET', quiet=TRUE)` failed. (See above for error)

I'm guessing it's still not the right R library that is being used in the conda environments. The cluster I'm working on has the version 3.6.0 of R installed but i tried switching to R version 4.1.1 and still got the same message. Any idea what could cause this ?

from scib-pipeline.

Yes, I think it's still the same issue. Could you check any variables that R uses, such as R_HOME etc.?

from scib-pipeline.

Here's what i tried :

• in the scib-R4.0 environment :

conda activate scib-R4.0
echo $R_HOME
/CONDAS/users/tnoel/scib-R4.0/lib/R
echo $R_LIBS

echo $R_LIBS_SITE

echo $R_LIBS_USER

• in the scib-pipeline-R4.0 env :

conda activate scib-pipeline-R4.0
echo $R_HOME
/CONDAS/users/tnoel/scib-pipeline-R4.0/lib/R
echo $R_LIBS

echo $R_LIBS_SITE

echo $R_LIBS_USER

R_LIBS and a few others do not seem to be defined. I am not very familiar with the way R works so if there are any others you'd want me to check please let me know.

from scib-pipeline.

Great that you figured out the issue! I also wasn't quite aware of the impact of .Rprofile. I'll be including that in a trouble-shooting section of the README.

from scib-pipeline.