To keep all the nf-core pipelines up-to-date with the latest version of the community standards, we have implemented a synchronisation tool. This ensures that updates to the community standards are propagated to all the nf-core pipelines.
If a pipeline is created using the
nf-core create command, everything is automagically set up for the synchronisation to work.
However, if a pipeline was not created using this command, this documentation will tell you how to set it up properly.
In brief, when the synchronisation tool is triggered it compares every pipeline against the new standards and opens pull requests when mismatches are found. These pull requests needs to be manually dealt with by the pipeline maintainers.
Behind the scenes, the synchronisation tool fetches the variables needed for a pipeline and uses this to trigger a
nf-core create --no-git command.
The result from this command is then compared against what is stored in a special branch of the pipeline called the
If any changes are detected, a pull request is created.
For this to work in practice, the
TEMPLATE branch needs to have a shared git history with the
master branch of the pipeline.
nf-core create command arranges this by enforcing a first commit to the
master branch before any development has taken place.
The next section deals with the case when the pipeline was not created by the
nf-core create command.
This section describes how to set up a correct TEMPLATE branch in the case your pipeline was not created with a TEMPLATE branch from the beginning. If you created a pipeline with the
nf-core create command, you should be all ready to go and can skip this step. Otherwise proceed with caution. It is probably a good idea to make sure you have all your local changes pushed to github and you could even make a local backup clone of your repository before proceeding.
You should also consider the option to restart your pipeline project by running the
nf-core create command and simply copy in the modifications you need into the newly created pipeline.
This walkthrough assumes that you are working directly with the head
nf-core fork of the pipeline.
It is possible (and potentially safer) to do this on your own fork instead, it's up to you.
First clone your pipeline into a new directory (in case we mess things up):
mkdir TMPDIR cd TMPDIR git clone https://github.com/nf-core/YOURPIPELINENAME.git
Then create the new TEMPLATE branch and delete all your files in order to have a completely empty branch:
cd pipeline_root_dir git checkout --orphan TEMPLATE && git rm -rf '*'
Make sure your branch is completely empty by checking the status of
$ git status On branch TEMPLATE No commits yet nothing to commit (create/copy files and use "git add" to track)
Regenerate your pipeline from scratch using the most recent template:
Make sure you are within your pipeline root directory before running these commands.
nf-core create --no-git
If your pipeline already has versioned releases (eg. you are not currently on
then specify the version number that you are currently on:
nf-core create --no-git --new-version 1.3dev
The version you choose should match the branch that you intend to merge with. If you already have a release, you should probably be merging in to
deveventually, so use the version number specified there.
Follow the prompts to fill in the pipeline name, description and author(s).
Make sure that you take the exact text that you already have already used in your pipeline's
nextflow.config file (
manifest.name etc.), if these have already been written.
This creates a new directory
YOURPIPELINENAME with the template pipeline files in it.
Now move these files into your root git directory:
mv nf-core-YOURPIPELINENAME/* . mv nf-core-YOURPIPELINENAME/.[!.]* . rmdir nf-core-YOURPIPELINENAME
Now make sure the newly created files are in the correct place. It should look similar to this:
$ git status On branch TEMPLATE No commits yet Untracked files: (use "git add <file>..." to include in what will be committed) .gitattributes .github/ .gitignore .travis.yml CHANGELOG.md CODE_OF_CONDUCT.md Dockerfile LICENSE README.md assets/ bin/ conf/ docs/ environment.yml main.nf nextflow.config nothing added to commit but untracked files present (use "git add" to track)
If it all looks good, then commit these files:
git add . git commit -m "Initial template commit"
For the nf-core bot to be able to access your
TEMPLATE branch, you need to push it to the upstream repository (https://github.com/nf-core).
git push origin TEMPLATE
The only remaining step is unfortunately a rather tedious one.
You have to merge the
TEMPLATE branch into your main pipeline branches, manually resolving all merge conflicts.
If your pipeline is in early development, you can do this with
master branch directly. If not, it's better to do this
in a branch and then you can make a pull-request to
master when ready.
git checkout dev git checkout -b template_merge git merge TEMPLATE --allow-unrelated-histories
You can try extra flags such as
-Xignore-space-at-eol if you find that the merge command shows entire files as being new.
You'll probably see a lot of merge conflicts:
Auto-merging nextflow.config CONFLICT (add/add): Merge conflict in nextflow.config Auto-merging main.nf CONFLICT (add/add): Merge conflict in main.nf Auto-merging environment.yml CONFLICT (add/add): Merge conflict in environment.yml Auto-merging docs/usage.md CONFLICT (add/add): Merge conflict in docs/usage.md
Go through each file resolving the merge conflicts carefully. Many text editors have plugins to help with this task. The Atom GitHub package is one example of an excellent interface to manage merge conflicts (see the docs).
It's highly recommended to use a visual tool to help you with this, as it's easy to make mistakes if handling the merge markers manually when there are so many to deal with.
Once you have resolved all merge conflicts, you can commit the changes and push to the GitHub repo:
git commit -m "Merged vanilla TEMPLATE branch into main pipeline" git push origin template_merge
The final task is to create a pull request with your changes so that they are included in the upstream repository.
Once your commits are finally merged into the
master branch, all future automatic template syncing should work.
When new releases of
nf-core/tools and it's associated template are released, pull-requests will automatically
be created to merge updates in to your pipeline for you.
There are rare cases, when the synchronisation needs to be triggered manually, i.e. it was not executed during an
nf-core/tools release on Github, or when you want to perform a targeted sync.
You can do so by running the
sync script, which is located in the top-level
./bin/ folder in the nf-core tools repository.
You need to provide two environment variables,
TRAVIS_TAG specifies the Github release tag you want to sync,
NF_CORE_BOT the nf-core bot user password.
To manuall trigger the syncronisation for ALL pipelines, just run:
To trigger synchronisation for a particular pipeline, run:
sync <pipeline name>