Reproducible Analysis and Documentation with R and RStudio

In this workshop we will focus primarily on two workflow tools. The first is using RStudio Projects and the ‘here’ package to create a project-oriented workflow in your R scripts. The second is using R Markdown for literate programming. We will briefly review R Markdown documents, which you can use to write complete papers. We will take a look at the new features in Visual R Markdown, such as citations and technical writing.

This is the third of four sessions in our series “Reproducible Research for Early Graduate Students.” The remaining session is “Sharing Your Data for Transparent and Reproducible Research.”

Teaching Slides are available, and mirror the content in this document.

Learning Objectives

In this workshop, you will learn:

• How to use RStudio Projects and why you would want to
• What R Markdown documents are and why you would want to use them
• How the here package can help with file paths once you start using R Markdown documents regularly

This is not a complete workshop on how to use the many options provided within R Markdown documents.

• Before the first session, make sure you have installed or updated R, RStudio, and the here package. Installing and updating instructions are here. When reviewing the instructions, substitute the here package whenever you see the tidyverse package.
• Visit my GitHub Penguins repository and download the repository. This repository contains files for a demo project. Click on the big green “Code” button, then “Download ZIP,” and save it somewhere that you can find on the day of the workshop and unzip the folder.

Project-oriented workflow with RStudio Projects

What not to do: setwd() and rm(list = ls())

The working directory is the place that R looks for your files. There is always one set. You can check with getwd().

A lot of R scripts start off like this:

setwd("C:/Users/jah2ax.ESERVICES/Box Sync/_R/research/penguins")

It uses the setwd() function to set the working directory. The problem with this is that there is a 0% chance that this will work on someone else’s computer, and will very likely break on your computer if (and when) you move your directory around. setwd() is fragile - it breaks! If you actually want this to work on someone else’s computer - like an advisor or a collaborator - then that line of code is not going to do it’s job (i.e., set the working directory to the proper folder) because the path won’t be any good on the other person’s computer. Same is true if you move your files around. This is annoying for everyone!

Similarly, you may have seen this line, often at the top of scripts: rm(list = ls()). That line is good at removing objects from your environment, but doesn’t give your a fresh R process, specifically, rm(list = ls()) does not detach libraries. There is a better way!

Organize work into projects

Aka “little p projects.”

First, let’s make sure we set up our work into projects.

Step 1: Organize your work into “projects.” That means all the files in a project are in a single directory, aka folder. Check out UVA Library’s workshop Organize Your Files and Metadata for Transparent and Reproducible Research for more details.

Step 2: When working on a project, set your working directory to that project’s main directory. Be intentional about using that working directory. Use the tools that we will cover below, and avoid using setwd() and absolute paths

Step 3: Be disciplined about using your files paths. The working directory is the main directory, and all other paths are relative to that directory.

What RStudio Projects do for you

Aka “capital P Projects.”

RStudio can help you solve all of these problems with RStudio Projects. When you launch an RStudio Project, RStudio will:

• Launch R/RStudio.
• Set the working directory to the project’s folder.
• Provide you with a fresh environment. That means there are no libraries loaded and no objects saved in your environment.

This solves the problems posed by both setwd() and rm(list = ls())!

Let’s say you are working on one project, and you want to switch to another project. When you want to open a new Project, RStudio will:

• Restart R.
• Set the working directory again.

If you want to run multiple projects at the same time, you can do that!

• Each project gets its own R process (environment), and the working directory is set correctly in each.
• Bonus: multiple projects feel like multiple instances of R and you can use whatever mode you like to switch between them, e.g. Command+Tab (Mac OS) or Alt+Tab (Windows)

Using RStudio Projects

As an example, we will imagine you are working on two projects: worms and penguins.

You can designate a new or existing folder as an RStudio Project. All that means is that RStudio leaves behind a file, e.g., worms.Rproj. This file stores specific settings for that project.

In RStudio, use File…New Project or click on the R cube in the upper right to get started. Create a fresh project called worms.

Once you do that, close RStudio. Now go to where you saved the worms project folder using Windows Explorer or Mac Finder. Double click on your worms.rproj file.

Now, click on the little down arrow next to the R cube in the upper right, and open the penguins project you downloaded.

Notice that you get a fresh environment when you switch back and forth.

Using Windows Explorer or Mac Finder, go to where you saved your new worms project and double click on the worms.rproj file. It will launch your project, and you can run multiple instances of R at the same time. You should have both worms and penguins projects open at the same time. They are running in completely separate environments.

Dynamic Documents with R Markdown

You might be used to cutting tables or plots from your statistical package, and pasting them into a Word document. Now, imagine that your advisor suggests you make changes to your analysis. Did you remember to update the table correctly in Word? What code generated that table in the first place?

Manual copy-and-paste is a huge opportunity for errors and confusion. It’s super tedious and no one wants to spend their time doing it or figuring out what code made what table. R Markdown solves that problem for you. R Markdown allows you to embed R code into narrative text, and then format that text into PDF, HTML, Word, and so on. Now when you need to make changes, you can make changes directly to the code embedded in the R Markdown document. No more cutting and pasting! This is better for reproducibility and transparency, and more simply, makes your life easier.

As noted in R Markdown: The Definitive Guide, with R Markdown, you can:

• Compile a single R Markdown document to a report in different formats, such as PDF, HTML, or Word.
• Create notebooks in which you can directly run code chunks interactively.
• Make slides for presentations (HTML5, LaTeX Beamer, or PowerPoint).
• Produce dashboards with flexible, interactive, and attractive layouts.
• Build interactive applications based on Shiny.
• Write journal articles.
• Author books of multiple chapters.
• Generate websites and blogs.

Visual Editor

Visual markdown editing is available in RStudio v1.4 or higher. This brings some of the features and ease-of-use that GUIs provide, while still always maintaining the source code that is the .Rmd file (R Markdown file).

To switch to visual editing, click on the compass in the upper right of the editor pane.

One of the highlights of the new Visual Editor is how easy it is to insert citations with Zotero.

References:

here package

The here package enables easy file referencing with project-oriented workflows. Paths are created relative to the top-level directory. It is most useful when you work with R Markdown documents a lot, as it smoothes out some of the idiosyncrasies of how R Markdown looks for files.

The here package always points to the root directory, which is typically where we set the working directory with our RStudio Project.

here::i_am("repro_analysis_R_RStudio.Rmd")
## here() starts at C:/Users/jah2ax.ESERVICES/Box Sync/_R/workshops/workshops_teaching/repro_analysis_documentation_R/website

References:

Demo for R Markdown and the here package

We are going to use the penguins demo project to explore the here package and R Markdown documents simultaneously. penguins has the following structure:

## C:/Users/jah2ax.ESERVICES/Box Sync/_R/workshops/workshops_teaching/penguins
## +-- analysis
## |   +-- apa.csl
## |   +-- references.bib
## |   +-- report-fail.Rmd
## |   \-- report-success.Rmd
## +-- data
## +-- penguins.Rproj
## +-- prepare
## |   \-- penguins.R
## \-- README.md

Note that this implements the essential “little p” project structure, by separating content into clear subdirectories.

Let’s take a look at the penguins project now. The instructions for downloading the project are in “Get Ready” section of this page. (Some of the penguins demo project is taken from the here() project demo files, specifically the file structure, and the sections related to here.)

References

Barrett, Malcolm. 2018. “Why Should I Use the Here Package When I’m Already Using Projects?” Malcolm Barrett. https://malco.io/2018/11/05/why-should-i-use-the-here-package-when-i-m-already-using-projects/.
Bryan, Jennifer, and Jim Hester. n.d.a. “Chapter 2: Project-Oriented Workflow.” In What They Forgot to Teach You About R.
———. n.d.b. “Chapter 3: Practice Safe Paths.” In What They Forgot to Teach You About R.
Grolemund, Garrett, J. J. Allaire. n.d. R Markdown: The Definitive Guide.
Müller, Kirill, and Jennifer Bryan. n.d.a. “Here Vignette.” https://cran.r-project.org/web/packages/here/vignettes/here.html.
———. n.d.b. “Using Here with Rmarkdown Vignette.” https://cran.r-project.org/web/packages/here/vignettes/rmarkdown.html.
Richmond, Jenny. n.d. “How to Use the ‘Here‘ Package.” http://jenrichmond.rbind.io/post/how-to-use-the-here-package/.
RStudio. n.d.a. “R Markdown: Introduction.” https://rmarkdown.rstudio.com/lesson-1.html.
RStudio. n.d. “Visual R Markdown.” https://rstudio.github.io/visual-markdown-editing/#/.
RStudio. n.d.b. “Visual R Markdown: Citations.” https://rstudio.github.io/visual-markdown-editing/#/citations.
Wickham, Hadley, and Garrett Grolemund. n.d. “8 Workflow: Projects.” In R for Data Science.