mirror of
https://github.com/MillironX/nfdocs-parser.git
synced 2024-10-08 22:14:43 +00:00
A Sphinx plugin for converting Nextflow docstrings into documentation
Thomas A. Christensen II
d86d85a182
Signed-off-by: Thomas A. Christensen II <25492070+MillironX@users.noreply.github.com> |
||
---|---|---|
.gitignore | ||
nfdocs-parser.py | ||
README.rst |
nfdocs-parser ============= A `Sphinx <https://www.sphinx-doc.org>`_ Extension for automatically generating documentation from `Nextflow <https://nextflow.io>`_ workflows, processes, and functions with YAML docstrings. Usage ----- I'm not putting this on `PyPi <https://pypi.org>`_ just yet, so you'll have to use this as a Git submodule. In your documentation project, run .. code-block:: bash mkdir _ext git submodule https://github.com/MillironX/nfdocs-parser.git _ext/nfdocs-parser.git Then update your ``conf.py`` to include the following lines: .. code-block:: python import os import sys sys.path.append(os.path.abspath('./_ext')) extensions = [ # Keep any other extensions, like InterSphinx and AutoSectionLabel here 'nfdocs-parser.nfdocs-parser', ] Inside your documentation file, include the following directive: .. code-block:: rst .. nfdocs:: ../path/to/nextflow/project/directory The extension will look for properly-formatted docstrings in all of your Nextflow files (extension ``.nf``) within that directory and its subdirectories and output the documentation in that file. Example ------- Take a simple Nextflow process, this is a pared-down example from `nf-core/modules <https://github.com/nf-core/modules>`_. .. code-block:: groovy process KRAKEN2 { input: tuple val(prefix), path(reads) path(db) output: tuple val(prefix), path("*classified*"), emit: classified tuple val(prefix), path("*unclassified*"), emit: unclassified tuple val(prefix), path("*report.txt"), emit: report script: """ kraken2 \\ --db ${db} \\ --classified-out ${prefix}.classified.fastq \\ --unclassified-out ${prefix}.unclassified.fastq \\ --report ${prefix}.kraken2.report.txt \\ ${reads} """ } This process still has a lot of unique inputs and outputs, so we annotate it using triple slashes (``///``) and YAML notation .. code-block:: groovy /// summary: Classifies metagenomic sequence data /// input: /// - tuple: /// - name: prefix /// type: val(String) /// description: Sample identifier /// - name: reads /// type: path /// description: List of input FastQ files /// - name: db /// type: path /// description: Kraken2 database directory /// output: /// - name: classified /// tuple: /// - type: val(String) /// description: Sample identifier /// - type: path /// description: | /// Reads classified to belong to any of the taxa in the Kraken2 /// database /// - name: unclassified /// tuple: /// - type: val(String) /// description: Sample identifier /// - type: path /// description: | /// Reads not classified to belong to any of the taxa in the /// Kraken2 database /// - name: txt /// tuple: /// - type: val(String) /// description: Sample identifier /// - type: path /// description: | /// Kraken2 report containing stats about classified and not /// classified reads process KRAKEN2 { ... } You will get output that looks something like this: Input ''''' +-------------------------+-------------------------------------------------------------------+ | **Tuple** | | | | +--------------------------+----------------------------------+ | | | | **prefix** (val(String)) | Sample identifier | | | | +--------------------------+----------------------------------+ | | | | **reads** (path) | List of input FastQ files | | | | +--------------------------+----------------------------------+ | | | | +-------------------------+-------------------------------------------------------------------+ | **db** (path) | Kraken2 database directory | +-------------------------+-------------------------------------------------------------------+ Output '''''' +--------------------------+-------------------------------------------------------------------+ | **classified** (Tuple) | | | | +--------------------------+----------------------------------+ | | | | val(String) | Sample identifier | | | | +--------------------------+----------------------------------+ | | | | path | Reads classified to belong to | | | | | | any of the taxa in the Kraken2 | | | | | | database | | | | +--------------------------+----------------------------------+ | | | | +--------------------------+-------------------------------------------------------------------+ | **unclassified** (Tuple) | | | | +--------------------------+----------------------------------+ | | | | val(String) | Sample identifier | | | | +--------------------------+----------------------------------+ | | | | path | Reads not classified to belong | | | | | | to any of the taxa in the | | | | | | Kraken2 database | | | | +--------------------------+----------------------------------+ | | | | +--------------------------+-------------------------------------------------------------------+ | **txt** (Tuple) | | | | +--------------------------+----------------------------------+ | | | | val(String) | Sample identifier | | | | +--------------------------+----------------------------------+ | | | | path | Kraken2 report containing stats | | | | | | about classified and not | | | | | | classified reads | | | | +--------------------------+----------------------------------+ | | | | +--------------------------+-------------------------------------------------------------------+ Motivation ---------- I liked using the XML documentation blocks in VB.NET because it worked so well with IntelliSense. I often find myself scrolling through Nextflow code to remember what the form of the input tuple for a particular process was or how many outputs I need to account for. YAML seemed like a far superior language for documentation, and as most of my Nextflow projects were already using Sphinx, parsing the docstrings as part of my Sphinx documentation seemed like the logical thing to do. Why don't you just use a sidecar ``meta.yml`` file like nf-core does? --------------------------------------------------------------------- Honestly, because I started using my own format before realizing what the ``meta.yml`` file had in it. After some consideration, however, I like my system better and am not planning to add compatibility for ``meta.yml`` files. Reasons my system is better: * No need for sidecar files: everything is in one place * Tuple channels are noted as being different that the components that make them up, e.g. knowing that a process requires a tuple of ``val(prefix), file(reads)`` and a file of ``reference_genome`` is more informative than knowing that a process needs a ``val(prefix)``, ``file(reads)`` and ``file(reference_genome)``