workflow.template.pacta

This project is archived for future reference, but no new work is expected in this repository.

This repository can be used as a template for all PACTA workflow repositories. It includes typical workflow scaffolding, and an explanation of good practices for developing a workflow.

How to use this template

When you are starting a new workflow, you may choose to:

• Click "New Repository" on GitHub
• Select this repository as a "template"

After this, you should submit a PR that changes the names in the following files to suit the name of your new repository:

• *.Rproj file
• README.md

You can then begin to work on your workflow using the demo.R file as a basis, and document how to run the workflow using the README.md.

Guidelines for writing a reproducible workflow (WIP)

These are a handful of guidelines for some best practices in developing a reproducible workflow repository. Please consider this when writing one yourself!

Consider Logically Separating Your Code into the "Data Science Pipeline"

A good way to keep your scripts organized is to think about them in terms of the "data science pipeline". It structures a typical data analysis path as follows:

• Import
• Tidy
• Do the following iteratively:
  • Transform
  • Visualize
  • Model
• Communicate (when you are happy with the output)

In the tidyverse, there are individual packages that can help you with each step of this process:

Version Control

Don't commit data or other sensitive information to version control. Data should be written to and read from an external source

Organize Your Project:

Create a clear folder structure for your project. Separate code (which can live in this repository) from outputs like: data, figures, and reports.

Consider using R projects (.Rproj) for a more organized environment.

Manage Dependencies:

At the beginning of your script, explicitly state all required R packages and their versions. You can use, for example, the renv package to manage package dependencies.

Always try to keep dependencies to a minimum, and remove those that are unused.

Always Manipulate Data Through a .R or .Rmd File!

If you want to be truly reproducible, then STOP OPENING EXCEL!

Use External Config Files

Store parameters, configurations, or settings in separate config files (e.g., a config.yml or .env) for easy modification without altering the code.

It is a good idea to document paths here, for example.

Clean Output:

Remove unnecessary or temporary files and folders from your script's output to keep things tidy.

Create an R Script for Data Downloading:

If your analysis relies on external data sources, create a separate R script to download and prepare the data automatically.

Reproducibility Checklist:

Develop a checklist to verify that your script meets all reproducibility requirements before sharing it.