bedpostx with Docker and Singularity!

In the previous article, I discussed the preprocessing of diffusion data. In this article, I’ll demonstrate how to use bedpostx with Docker and Singularity!

Tools and files used in this article:

• PREPROCESSED.zip
• docker
• nvidia-docker
• singularity

The version of FSL used in the docker is 5.0.10 and it has the cuda 8 version of bedpostx_gpu from here. PREPROCESSED.zip is preprocessed diffusion data using my dtiQA Singularity/Docker image.

The input to the docker/singularity container is a folder containing the diffusion data, a binary mask, and a corresponding config file named bedpostx.conf. The options to the config file are explained below:

• bedpostx_name – must be either “bedpostx” or “bedpostx_gpu”
• bedpostx_params – input parameters to bedpost; typically the defaults are used and this is left blank

Getting bedpostx to run is relatively simple, but having a prepackaged container in Docker and Singularity (plus a bonus QA PDF) just makes life so much easier. To run the example, do the following:

wget http://justinblaber.org/downloads/articles/bedpostx/PREPROCESSED.zip
unzip PREPROCESSED.zip
rm PREPROCESSED.zip
mkdir OUTPUTS
mv PREPROCESSED INPUTS
vim INPUTS/bedpostx.conf
 

Put this inside bedpostx.conf:

bedpostx_name = bedpostx_gpu
bedpostx_params = 
 

and then save it. Next, run the docker with:

sudo docker run --rm \
--runtime=nvidia \
-v $(pwd)/INPUTS/:/INPUTS/ \
-v $(pwd)/OUTPUTS:/OUTPUTS/ \
--user $(id -u):$(id -g) \
justinblaber/bedpostx:1.0.0
 

Or, if you don’t have sudo, run the singularity container with:

singularity run -e \
--nv \
-B INPUTS/:/INPUTS \
-B OUTPUTS/:/OUTPUTS \
shub://justinblaber/bedpostx_app:1.0.0

If you don’t have a gpu (highly, highly recommended with bedpostx), you can set:

bedpostx_name = bedpostx
 

And then remove the --runtime=nvidia or --nv flag. If all goes well, the output PDF should look like:

There you go! Sooooo simple.