Skip to content

Using Snk to manage workflows

The snk command line interface (CLI) provides several options and commands to manage Snakemake workflows. Here's a detailed breakdown of the usage.

Installing workflows

The snk install command is used to install a workflow. You can specify the path, URL, or GitHub name (user/repo) of the workflow to install. The simplest way to install a workflow is from Github. The Snakemake workflow catalog has an index of Snakemake workflows.

For example we can install snk-basic-pipeline (an example workflow from the Snakemake tutorial) with the following command:

snk install Wytamma/snk-basic-pipeline

By default Snk workflows will be installed (cloned) into a snk folder in the parent directory of the python bin directory (configurable with $SNK_HOME or --home). Snk will then create a executable for the workflow at $SNK_HOME/bin and symlink the executable to python bin directory (configurable with $SNK_BIN or --bin). 

/home/wwirth/.conda/envs/snk/
├── bin ($SNK_BIN)
│   ├── python
│   ├── snk-basic-pipeline -> .../snk/bin/snk-basic-pipeline
│   └── ...
├── lib
│   └── ...
└── snk ($SNK_HOME)
    ├── bin
    │   └── snk-basic-pipeline (symlinked to $SNK_BIN)
    └── workflows
        └── snk-basic-pipeline
            ├── config
            └── workflow
                ├── Snakefile
                ├── envs
                └── ...

Note

To globally install a workflow you could run SNK_BIN=~/.local/bin snk install Wytamma/snk-basic-pipeline assuming ~/.local/bin is in you $PATH (e.g. export PATH=$PATH:~/.local/bin in .bashrc)

Local install

To install a local workflow simply pass the path to snk install. 

snk install path/to/snakemake/workflow

Use the --editable flag to install local workflows in editable mode (useful for development). This will symlink the path to the SNK_HOME directory so all changes will be reflected in the CLI.

snk install -e path/to/snakemake/workflow

Isolated workflow environments

By default snk will use the snakemake and snk-cli version installed in the current environment.

Use the --isolate flag to create a new python virtual environment for the workflow. This will install a standalone version of snakemake and snk-cli in the environment. By isolating workflows you can use workflows that depend on different versions of snakemake at the same time.

snk install --isolate Wytamma/snk-basic-pipeline

To install a specific version of snakemake in the isolated environment, use the --snakemake flag. This will create a python virtual environment with the specified version of snakemake. 

snk install --snakemake 7.32.4 Wytamma/snk-basic-pipeline

You can add dependencies to the environment with the --dependency flag. This will install the specified python package in the environment. Useful for v8 workflows that use Snakemake plugins.

snk install --snakemake ">8.0" --dependency snakemake-executor-plugin-slurm Wytamma/snk-basic-pipeline

Other install options

Several options exist to modify the install process.

Use --name to rename the workflow. The new name will be used to call the CLI.

snk install --name my-workflow Wytamma/snk-basic-pipeline && my-workflow -h

Use --tag to specify the tag (version) of the workflow to install. You can specify a branch name, or tag. If not specified, the latest commit will be installed.

snk install --tag v1.0 Wytamma/snk-basic-pipeline && my-workflow -v

Use --commit to specify the commit (SHA) of the workflow to install. If not specified, the latest commit will be installed.

snk install --commit a725d3a4 Wytamma/snk-basic-pipeline && snk-basic-pipeline -v

Use --config to install a workflow with a non-standard config location. This is useful for installing a workflow that does not follow Snakemake best practices. The path should be relative to the workflow directory.

snk install --config path/to/config Wytamma/snk-basic-pipeline

Use --snakefile to install a workflow with a non-standard Snakefile name. Will rename the given file to Snakefile. This is useful for installing a workflow that does not follow Snakemake best practices. The path should be relative to the workflow directory.

snk install --snakefile path/to/Snakefile Wytamma/snk-basic-pipeline

Use --resource specify a resource that is required at run time. Resources will be copied to the run directory at runtime. The path should be relative to the workflow directory. 

snk install --resource path/to/resource Wytamma/snk-basic-pipeline

If the the workflow should not use conda environments, use --no-conda. This will prevent the workflow from creating a conda environment and will not use the --use-conda flag when running the workflow. 

snk install --no-conda Wytamma/snk-basic-pipeline

Installation can be forced with the --force flag. This can be used to overwrite existing workflows.

snk install --force Wytamma/snk-basic-pipeline

Listing workflows

The snk list command is used to view the installed workflows. 

snk list

┏━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Workflow              ┃ Version  ┃
┡━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ snk-basic-pipeline    │ 3445c7cd │
├───────────────────────┼──────────┤
│ variant-calling       │ v2.1.1   │
├───────────────────────┼──────────┤
│ workflow              │ editable │
└───────────────────────┴──────────┘

Note

Use --verbose (-v) to show workflow installation paths.

Uninstall workflows

The snk uninstall command is used to uninstall workflows. You must pass uninstall the name of the workflow (e.g. only the repo part of user/repo if installed from Github). 

snk uninstall workflow

Uninstalling workflow
  Would remove:
    <SNK_HOME>/workflows/workflow/*
    <SNK_BIN>/workflow
Proceed (Y/n)? y

Note

Use --force to force uninstall without asking.

Ejecting workflows

The cp -r $(workflow-name -p) workflow-name command is used to eject the workflow from the package. This will copy the workflow files to the current working directory. This will allow you to modify the workflow and run it with the standard snakemake command.

Following modification of the workflow you can run snk install ./workflow-name --force to install the updated workflow.