Prepare containerized BIDS App as a DataLad dataset

Understand the concept "container DataLad dataset"

BIDS Apps, by definition, are container images that process BIDS-formatted datasets [Gorgolewski2017]. Besides input BIDS datasets, BABS also requires BIDS App container to be in a DataLad dataset. It's a bit hard at first to understand the concept of "container DataLad dataset" - why container is a dataset now? Here, "DataLad dataset" is a DataLad concept: "a dataset" means "a collection of files in folders", and this "DataLad dataset" is version tracked by DataLad. So "container DataLad dataset" means "a collection of container image(s) in a folder tracked by DataLad".

Toy BIDS App

We prepared a toy BIDS App that can be used for quick testing. It counts non-hidden files in a subject's (or a session's) folder. The detailed descriptions can be found here. Its docker image is publicly available on Docker Hub.

How to prepare a container DataLad dataset of BIDS App?

Step 1. Get BIDS App container image

As the data processing will be performed on a cluster, and usually clusters only accept Singularity image (but not Docker image), you probably need to build the BIDS App as a Singularity image. Below is an example of building a Singularity image of toy BIDS App from Docker Hub:

toybidsapp_version="0.0.7"   # or the latest tagged version
toybidsapp_version_dash="0-0-7"   # same version, just in different format: connected by dashes
singularity build \
    toybidsapp-${toybidsapp_version}.sif \
    docker://pennlinc/toy_bids_app:${toybidsapp_version}

Step 2. Create a container DataLad dataset

You may use DataLad command datalad containers-add to add the built Singularity image (sif file) of BIDS App to a DataLad dataset:

datalad create -D "Note about the container" toybidsapp-container
cd toybidsapp-container
datalad containers-add \
    --url /full/path/to/toybidsapp-${toybidsapp_version}.sif \
    toybidsapp-${toybidsapp_version_dash}

Note the last argument is the image NAME in the container DataLad dataset. This string can only have characters and dashes in it. Remember what you assign as the image NAME because you will copy it for argument --container_ds when babs init.

(Optional) Step 3. Delete the original sif file

Lastly, as you have copied the sif file into this container DataLad dataset, you can delete the original sif file:

rm /full/path/to/toybidsapp-${toybidsapp_version}.sif

References

For more details, please refer to:

datalad containers-add's command-line interface: DataLad documentation
DataLad Handbook: containers.

Gorgolewski2017: Gorgolewski, K. J., Alfaro-Almagro, F., Auer, T., Bellec, P., Capotă, M., Chakravarty, M. M., Churchill, N. W., Cohen, A. L., Craddock, R. C., Devenyi, G. A., Eklund, A., Esteban, O., Flandin, G., Ghosh, S. S., Guntupalli, J. S., Jenkinson, M., Keshavan, A., Kiar, G., Liem, F., … Poldrack, R. A. (2017). BIDS apps: Improving ease of use, accessibility, and reproducibility of neuroimaging data analysis methods. PLoS Computational Biology, 13(3), e1005209. https://doi.org/10.1371/journal.pcbi.1005209