search

tbepler / topaz

Pipeline for particle picking in cryo-electron microscopy images using convolutional neural networks trained from positive and unlabeled examples. Also featuring micrograph and tomogram denoising with DNNs.
GNU General Public License v3.0
170 stars 62 forks source link

Singularity with Relion integration #80

Open Jason-vR opened 3 years ago

Jason-vR commented 3 years ago

Hi,

I suffer from a lack of permission on our GPU cluster so opted to go for the singularity Topaz container as a way to access the Relion integration. Thank you for this option, it is a fantastic way to quickly port complex package to our various compute resources.

I have it running now but had to add a new flag to make the container (singularity pull shub://nysbc/topaz) start in the current directory i.e, from where it is launched by run_topaz_train.py.

I am no containerisation expert but I think the runscript may need to be changed to default to CWD. Also, I didn't need to bind-in the system folders as they are accessible anyway.

Here is the successful command that I now reference as --topaz_path in Relion's External job:

"singularity run --nv --home $PWD topaz_latest.sif"

The quotes are needed to pass the full command.

I am not sure how much of this is down to our system configuration but thought it might help anyone else going the singularity route.

Jason

tbepler commented 3 years ago

@alexjnoble can you check on this to see if we should update the main singularity build here?