Is not like an issue but more like an aha moment about picrust2 HOT 3 CLOSED

Actually, this made me thinking. I have of course an .Rprofile file in my HOME folder, that tells R where to locate the (locally) installed R packages. I just renamed this to Rprofile (no dot), and re-started picrust2. This seems to work like a charm! Now there is no confusion between my local R packages and the picrust2 conda environment when R is started from the latter.

This may actually be helpful to add to the installation procedure for picrust2.

from picrust2.

Interesting, since I having a related problem. Installed picrust2 as a conda environment on our computing cluster, but get the following error:

Error running this command:
Rscript /mnt/users/larssn/.conda/envs/picrust2/lib/python3.8/site-packages/picrust2/Rscripts/castor_nsti.R picrust_out/out.tre /home/work/tmp5a10ia2k/known_tips.txt /home/work/tmp5a10ia2k/nsti_out.txt

Standard error of the above failed command:
Error: package or namespace load failed for ‘castor’ in dyn.load(file, DLLpath = DLLpath, ...):
unable to load shared object '/net/fs-2/scale/OrionStore/Home/larssn/R/myLib/Matrix/libs/Matrix.so':
libRlapack.so: cannot open shared object file: No such file or directory
Execution halted

It looks to me like the R script castor_nsti.R has problems loading the R package castor, which in turn loads the Matrix package. But both are loaded without any problems from my local R. Is it not strange that this conda environment uses my locally installed R packages?

from picrust2.

I am just closing this issue as it seems to be fixed. Thanks everyone for your follow-up here! It seems like there are several issues that can be run into, particularly with the M1 Macs, so I think it might be too confusing to add anything to the install pages, but hopefully anyone that runs into this issue will be able to search here for the solutions!

from picrust2.