Workflow
There’s a famous blog post about workflows in R1 about a talk Jenny Bryan gave that included this slide:
If the first line of your R script is
I will come into your office and SET YOUR COMPUTER ON FIRE 🔥.
If the first line of your R script is
I will come into your office and SET YOUR COMPUTER ON FIRE 🔥.
The problem with setwd()
setwd()changes the working directory, leading to potential issues in collaboration and reproducibility- You and I don’t have the same file structure!
- For example, my current working directory is
- It’s also really annoying to move around files and folders, even if it’s just you using them
R Projects
- R projects provide a structured and organized way to work on projects in R
- R projects encapsulate all project-related files and settings into a single directory
- RStudio makes it easy to work with R projects
R Projects
my-project/
├─ my-project.Rproj
├─ README
├─ data/
│ ├── raw/
│ └── processed/
├─ R/
├─ results/
│ ├── tables/
│ ├── figures/
│ └── output/
└─ docs/An
.Rprojfile is mostly just a placeholder. It remembers various options, and makes it easy to open a new RStudio session that starts up in the correct working directory. You never need to edit it directly.A README file can just be a text file that includes notes for yourself or future users.
I like to have a folder for raw data – which I never touch – and a folder(s) for datasets that I create along the way.
Benefits of R Projects
- Isolation: Each project has its own workspace, separate from other projects
- Reproducibility: Projects ensure that code and data are self-contained and portable
- Collaboration: Projects facilitate collaboration by sharing the entire project directory
Creating an R Project
- Open RStudio and go to File > New Project, or click on the projects button in the upper-right corner of RStudeio.
- Choose a project location (New Directory, Version Control, Existing Directory).
- Specify the project directory and create the project.
- Choose the project type (e.g., regular project, Quarto website, Bookdown book)
The {here} Package
- The
{here}package simplifies file path management within R projects. - It provides a consistent and reliable way to reference files within the project directory.
Benefits of the {here} Package
- Simplicity : Avoids the need for manual file path manipulation with
setwd()andpaste()functions - Flexibility : Works seamlessly even when scripts are run from different locations
- Portability : Ensures that file paths work consistently across different systems
Using {here} in your project
- Install the
{here}package:install.packages("here"). - Load the
{here}package in your R script:library(here). - Use the
here()function to reference files within your project directory.
I actually like to use here::here() rather than loading the package with library(here) every time.
Using {here} in your project
Construct file paths with reference to the top directory holding your
.Rprojfile.here::here("data", "raw", "data.csv")for me, here, becomes"/Users/l.smith/Documents/Presentations/reproducible-epi-SER/data/raw/data.csv"But if I send you my code to run, it will become whatever file path you need it to be, as long as you’re running it within the R Project.
Start fresh
rm(list = ls())removes objects from your environment, but it doesn’t unload the packagesYou can easily create errors of the type
Error in... could not find function...because your code runs just fine until you restart R and realize you never included the rightlibrary()callsRestart your R session early and often as you write code!
Conclusion
- Embrace the project-oriented workflow in R for better organization, reproducibility, and collaboration
- Use R projects to encapsulate project-related files and settings
- The
{here}package helps with reliable and portable file path management
Additional Resources
- Jenny Bryan’s blog post: Project-oriented workflow
- Another blog post by Malcolm Barrett
{here}package documentation: