A workflow to be an ace reproducible scientist

I. Bartomeus

Work within a project

• Rstudio projects are self contained.

• GitHub integration allow for version control: https://happygitwithr.com/

• GitHub allow for larger visibility and integration with other tools (see below).

Create the actual dynamic document

• Dynamic documents combine executable code and formatted text. Quarto is the current standard.

• Text is annotated using Markdown flavored syntax.

• They can easily be turned into webpages, presentations and interactive documents.

Create the actual dynamic document

• In /demo.qmd you can play with one of this documents.

• Or open a new empty Quarto document (upper-right corner in Rstudio).

Add a ReadMe and a licence

• Explain people what you are doing: README.md file.

• .md is better here because is automatically rendered by GitHub.

• Explain people what they can do with your project: Licence it!

• Other useful files: .gitignore, .Rproj

Licenses

• Add a LICENCE txt file

  • Data: Creative commons: CC-BY

  • Code: MIT

• Choose a licence

• Add it to README.md: e.g.  (which you get through this code: [![License: CC BY 4.0](https://img.shields.io/badge/License-CC_BY_4.0-lightgrey.svg)](https://creativecommons.org/licenses/by/4.0/)

If releasing code: Testthat and renv

• Defensive programming (examples) and maybe functionalize your code

• Unit testing (testthat)

  • Check impossible values vs improbable values

• Track software versions (Renv, also see Docker for more complete options)

• Code review / Pair programming

If releasing data, add Metadata

• Standardize coding (4_NitrogenPhosphurous > 4NP > 4)
• Never edit the master file
• Use tidy data structures and plain, standard and open formats (e.g. csv)
• Metadata (Dataspice: creates computer (JSON, EML) and human readable (html) metadata)

Make a release and link it to Zenodo

• Zenodo gives you permanent stable DOI

• Log in to https://zenodo.org/ (with your GitHub account).

• Go to /GitHub in the top main menu and update your repos. Flip the switch of your repo.

• In GitHub, go to Releases (right column) and create a new one.

Make a release and link it to Zenodo

• Versioning should use incremental numbers for minor and major updates (0.1, 0.2, 1.0, 1.2, …)

• Go back to Zenodo to collect your DOI and add it to README.md file.

• Think on adding a News.md

Make the document/webpage visible on github

• In GitHub go to Settings/Pages and enable webpage creation.

• Many options are possible, but a simple Deploy from branch, main in /docs should work.

• GitHub will look for a index.html and a .nojekyll file.

• You can link it to the GitHub “About” section in the right column.

Issues: Templates.

• Explain how to contribute and report bugs (e.g. Github issues)

Automation: github actions

Github actions e.g. https://github.com/ibartomeus/CropPollinationModels

Summary:

Readme.md with DOI, version, file description, relevant info, etc…

/Data with raw, clean and the relevant scripts.

/Analysis

/Manuscript

LICENCE

testthat

News.md

Metadata/

Renv/

All will change

• Tools evolve

• Plan accordingly (use standard open formats)

• Will we see living papers in our lifetime?

Acknowledgement:

• Previous workshops:
  • https://github.com/RadicalCommEcol/Open_science/blob/main/ORT_Science.qmd
  • https://github.com/ibartomeus/ReproMeta