Container definition

The container a module uses is defined in the container section. This container can be provided for various engines and runtimes, like docker, apptainer and podman, from official and private repositories (see dockerhub and quay). A container can also be defined as a link to a valid OCI artifact (e.g. *.img, *.simg, *.tar.gz).

Note

For the Non-Local Means module, neither the depot.galaxyproject.org or biocontainers repositories defined in the auto-generated template host the dependencies we need. For most of the neuroimaging software, you’ll find them on dockerhub. A popular repository for versioned images for neuroimaging is maintained by neurodesk; their images are located here.

Note that Neurodesk are not image managers and will not version your software if they don’t have use for it.

Since our dependency is scilpy, we will use their images. Do the following replacements in the container section :

container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
    'https://scil.usherbrooke.ca/containers/scilus_2.1.1.sif' :
    'scilus/scilus:2.1.1' }"

What is happening here ?

On the first line, the module template has provided some logic to select between a docker or apptainer image, that have two distinct formats and require different configuration. It does it using the workflow.containerEngine variable. Apptainer images can be built from docker, and if the task.ext.singularity_pull_docker_container parameter is set, it will do so from 'scilus/scilus:2.1.1'. Else, it will download scilus_2.1.1.sif using the link.

Conda environment

nf-neuro does not use conda environments and does not run tests for it, but it is possible to define a conda environment following nf-core guidelines. If doing so, open an issue to discuss it with the maintainers.

For now, we disable conda support by filling the environment.yml file with the content below :

name: denoising_nlmeans
channels: [] # Empty, since we do not use conda.
dependencies: [] # Empty, since we do not use conda.

With this, the main portion of the module is complete. Next, you will add metadata and tests to the module. You will also learn how to include quality control that get automatically collected into a report for your users to review after execution.