Execution engines

Reportsrender comes with two execution engines:

  • Rmarkdown. This engine makes use of the Rmarkdown package implemented in R. Essentially, this engine calls Rscript -e “rmarkdown::render()”. It supports Rmarkdown notebooks (Rmd format) and python notebooks through reticulate.

  • Papermill. This engine combines papermill and nbconvert to parametrize and execute notebooks. It supports any programming language for which a jupyter kernel is installed.

Supported notebook formats

Reportsrender uses jupytext to convert between input formats. Here is the full list of supported formats.

So no matter if you want to run an Rmd file with papermill, an ipynb with Rmarkdown or a Hydrogen percent script, reportsrender has got you covered.

Hiding cell inputs/outputs

You can hide inputs and or outputs of individual cells:

Papermill engine:

Within a jupyter notebook:

  • edit cell metadata

  • add one of the following tags: hide_input, hide_output, remove_cell

    "tags": [

Rmarkdown engine:

  • all native input control options (e.g. results=’hide’, include=FALSE, echo=FALSE) are supported. See the Rmarkdown documentation for more details.

Jupytext automatically converts the tags to Rmarkdown options for all supported formats.

Parametrized notebooks

Papermill engine:


  • Add the tag parameters to the metadata of a cell in a jupyter notebook.

  • Declare default parameters in that cell:

input_file = '/path/to/default_file.csv'
  • Use the variable as any other:

import pandas as pd

Rmarkdown engine:


  • Declare the parameter in the yaml frontmatter.

  • You can set default parameters that will be used when the notebook is executed interactively in Rstudio. They will be overwritten when running through reportsrender.

title: My Document
output: html_document

  input_file: '/path/to/default_file.csv'
  • Access the parameters from the code:


Be compatible with both engines:

Yes it’s possible! You can execute the same notebook with both engines. Adding parameters is a bit more cumbersome though.

Example (Python notebook stored as .Rmd file using jupytext):

title: My Document
output: html_document
  input_file: '/path/to/default_file.csv'

```{python tags=c("parameters")}
    # try to get param from Rmarkdown using reticulate.
    input_file = r.params["input_file"]
    # won't work if running papermill. Re-declare default parameters.
    input_file = "/path/to/default_file.csv"

Sharing reports

Reportsrender create self-contained HTML files that can be easily shared, e.g. via email.

I do, however, recommend using github pages to upload and share your reports. A central website serves as a single point of truth and elimiates the problem of different versions of your reports being emailed around.

You can make use of reportsrender index to automatically generate an index page listing multiple reports:

Say, you generated several reports and already put them into your github-pages directory:

├── 01_preprocess_data.html
├── 02_analyze_data.html
└── 03_visualize_data.htmlp

Then you can generate an index file listing and linking to your reports by running

reportsrender index --index gh-pages/index.md gh-pages/*.html

For more details see Usage from command line and reportsrender.build_index()

Password protection

Not all analyses can be shared publicly. Unfortunately, github-pages does not support password protection.

There is a workaround, though:

As github-pages doesn’t list directories, you can simply create a long, cryptic subdirectory, e.g. t8rry6poj7ua6eujqpb57 and put your reports within. Only people with whom you share the exact link will be able to access the site.

Combine notebooks into a pipeline

Reportsrender is built with pipelines in mind. You can easily combine individual analysis steps into a fully reproducible pipeline using workflow engines such as Nextflow or Snakemake.

A full example how such a pipeline might look like is available in a dedicated GitHub repository: universal_analysis_pipeline. It’s based on Nextflow, but could easily be adapted to other pipelining engines.