This is the main remote registry for Singularity HPC.
ποΈ Read the documentation ποΈ βοΈ Browse the container module collection βοΈ
This registry showcases Singularity HPC (shpc), and provides the default set of containers for its default registry. Contributions are very much welcome, so please do submit a pull-request if you'd like more software to be added, or open an issue to request a new addition!
The documentation has an introduction about registries, and explains how to organize the required files for a container entry.
First of all, there is an automated way of getting most of the container.yaml
written via GitHub magic π© automation.
- Fork of this repository.
- Go to the Actions tab.
- Select "Generate New Container" in the left-hand side menu.
- Click the "Run workflow" dropdown on the right-hand side.
- Fill in the name of the container you want to build a
container.yaml
for, as well as a URL and a description (both are mandatory fields). Leave the branch asmain
. - Hit the green "Run workflow" button, wait and marvel at the magic β¨ happening. The GitHub Action will automatically open a pull request with most of the
container.yaml
filled in for you. You can check the branch out, edit it further β in particular refine the list aliases.
Note that all of these scripts registry singularity-hpc (shpc) to be installed:
$ pip install singularity-hpc
If you find a container missing aliases (or having extras) and you don't want to manually open files,
we provide a helper script to do so. After cloning the repository, you could check out a new branch and add an alias
(note the --registry
defaults to the present working directory where you are running the script, so this
should be run in the root of the cloned repository):
$ python .github/scripts/alias.py add quay.io/biocontainers/samtools test /opt/bin/test
If you ask to add an alias that already exists, this might mean changing the path, and you need to use force:
$ python .github/scripts/alias.py add quay.io/biocontainers/samtools test /opt/bin/test --force
And then to remove:
$ python .github/scripts/alias.py remove quay.io/biocontainers/samtools test
If an alias doesn't exist, you will get an error on remove. After these changes you would want to open a pull request to persist your changes to the registry.
The equivalent helper exists for features. As an example, here is how to set a boolean (true/false) or value of None):
$ python .github/scripts/feature.py add quay.io/biocontainers/samtools home true
$ python .github/scripts/feature.py add quay.io/biocontainers/samtools home false
$ python .github/scripts/feature.py add quay.io/biocontainers/samtools home none
And how to remove it:
$ python .github/scripts/feature.py remove quay.io/biocontainers/samtools home
And then to remove:
$ python .github/scripts/alias.py remove quay.io/biocontainers/samtools test /opt/bin/test
The same rule applies for using --force
.
And finally, we have the same for environment. Here are examples:
$ python .github/scripts/env.py add quay.io/biocontainers/samtools maintainer vsoch
$ python .github/scripts/env.py remove quay.io/biocontainers/samtools maintainer
We have a script that will generate (non existing) modules for BioContainers, and it is run once a week! It works by way of using an updated cache at [https://github.com/singularityhub/shpc-registry-cache] generated directly from Biocontainers, which not only captures aliases for a latest tag, but also derives the accumulated counts across all 8K+ containers. With these counts we can generate aliases as follows:
- Start with the loaded global counts, counts.json
- Subset to those in a container, the alias counts
- Rank ordering from least to greatest (lower frequency is a more unique commands we are interested in)
- Including any counts with a frequency <= 10 (this accounts for containers with many unique aliases)
--min-count-inclusion
- Above that threshold, including the next N
--additional-count-inclusion
(less unique but possibly important or interesting) - Use these to generate a new container.yaml for the file (if it does not exist yet!)
To run the above, you'll need the cache cloned locally, and singularity-hpc installed
$ pip install git+https://github.com/singularityhub/singularity-hpc@main
$ pip install requests pipelib beautifulsoup4
$ git clone https://github.com/singularityhub/shpc-registry-cache /tmp/cache
And then to run the script (this shows the defaults)
$ python .github/scripts/get_biocontainers.py --cache /tmp/cache --registry $(pwd) --min-count-inclusion 10 --additional-count-inclusion 25
from the root. Since this added over 8K containers to the registry, we needed a new strategy for running the updater TBA!
Refer to the documentation for a list of the container.yaml
fields, especially the required ones.
A pull-request that doesn't contain the required fields will have changes requested to add them.
This registry is only for container images freely available. We can't accept container.yaml
for private or access-restricted containers.
Furthermore, consider the following tips when making the pull-request:
- Ideally, the container tags listed in
container.yaml
should be actual versions, not genericlatest
,stable
, which tend to be moving targets. Remember that containers and Singularity HPC are there for reproducibility ! - The convention here is to put containers hosted on the Docker Hub at the top of the repository, e.g. ruby, rather than in a
docker.io
sub-directory. - By putting your name down as
author
, you accept you may be contacted to review further updates of thecontainer.yaml
We use the all-contributors tool to generate a contributors graphic below.
This code is licensed under the MPL 2.0 LICENSE.