Open perrymil opened 9 months ago
Did you set up RELION's Python environment by conda?
Hi,
No, the spack package should handle managing that environment without need for conda. It worked fine for relion@4.0.x and considering relion needs to run on an HPC environment, conda isn't a great solution to make this available for all of our users without having to redo this for each user. Or at least, I think thats how it should work...
RELION 5.0 assumes you set up a conda environment with the provided environment.yaml
.
If you don't want to use conda, you need to install required packages yourself and point the Python runtime to RELION by cmake -DPYTHON_EXE_PATH
.
Probably you need a custom Spark build script but I have never used Spark so I cannot help you.
conda isn't a great solution to make this available for all of our users without having to redo this for each user
No, you just install conda and set up the environment in a common application folder and let others use it. You don't have to make the common folder globally writable.
I have the same issue when running blush, problem not fix: ModuleNotFoundError: No module named 'relion_blush'
Something went wrong in the external Python call... Command: relion_python_blush Class3D/job032/run_it001_class001_external_reconstruct.star --gpu ,
Using python executable: /home/ljw/anaconda3/bin/python
mpirun detected that one or more processes exited with non-zero status, thus causing the job to be terminated. The first process to do so was:
Process name: [[5638,1],1] Exit code: 1
And I download a new package from GitHub and replace the old environment.yaml ljw@One:~/relion$ conda env update -f environment.yml Channels:
done #
#
#
#
@V3eneno
Your Python interpreter for the relion-5.0
conda environment is /home/ljw/anaconda3/envs/relion-5.0/bin/python
but apparently you have specified /home/ljw/anaconda3/bin/python
during installation. Fix this mismatch by rebuilding RELION with cmake -DPYTHON_EXE_PATH=/home/ljw/anaconda3/envs/relion-5.0/bin/python
. Note that you should delete your existing build directory before rerunning this command.
Thanks! it fixed!
I am attempting to install relion ver5.0 on our HPC cluster running rhel7.
We use spack to build and install software, and I attempted to update the existing spack package to use the new ver5.0 branch
After installing it, when running it errors out with the below:
I saw mention of a similar problem,
https://github.com/3dem/relion/issues/1009
But we aren't using the --pipeline-control option to my knowledge. I was thinking it might have something to do with the downloading weights via ftp issue, considering our network blocks insecure ftp connections... but I wasn't sure how to test that.
Additionally, other binaries display a similar error output: