Dockerfile of writing paper by R Markdown

This is a repository of Dockerfile including packages and add-ins that are useful for writing articles with R Markdown in Rstudio. This Dockerfile is based on rocker/verse. Installing rstan conducted by the method described in antoine-sachet / rocker-ML.

日本語解説(Japanese)

Maintainer is Yoshihiko Kunisato (ykunisato@psy.senshu-u.ac.jp)

Keywords: psychology, cognitive science, rstudio, rstan, rmarkdown

Usage

1. Install "Docker Desktop"

2. Open "terminal"(Mac) or "Command Prompt"(Windows)

3. Type the following code to pull a Docker container. Change the "password" and "name_of_container" as you like.

Mac

docker run -e PASSWORD=password -p 8787:8787 -v $(pwd):/home/rstudio -d --name paperr ghcr.io/ykunisato/paper-r:latest-linux-arm64

or

docker run -e PASSWORD=password -e DISABLE_AUTH=true -p 8787:8787 -v $(pwd):/home/rstudio -d --name paperr ghcr.io/ykunisato/paper-r:latest-linux-arm64

Windows

docker run -e PASSWORD=password -p 8787:8787 -v "%cd%":/home/rstudio -d --name paperr ghcr.io/ykunisato/paper-r:latest-linux-arm64

4. Open the web browser and type "http://localhost:8787/" in the URL bar.

5. You will see the Rstudio on the web browser. Type rstudio in ID column and password that you set in password column.

List of R packages

AI assistant tools

In addition to the R packages, this image bundles tools that let you use AI assistants for coding and writing:

• R packages for LLMs: ellmer (calling LLM APIs from R), gander and chores (RStudio add-ins for AI-assisted coding), btw, and usethis.

Setting up the API key

To use these tools, set your own API key in ~/.Renviron.

1. In R, open the file with:

   usethis::edit_r_environ()

2. Add a line with your key (here, Groq) and save:

   GROQ_API_KEY="gsk_xxxxxxxxxxxx"
   

3. Restart R.

This image's Rprofile.site is configured so that when GROQ_API_KEY is set, gander automatically uses Groq's llama-3.3-70b-versatile model:

if (nzchar(Sys.getenv("GROQ_API_KEY"))) {
  options(
    gander.chat = ellmer::chat_groq(
      model = "llama-3.3-70b-versatile"
    )
  )
}

Changing the Groq model

To use a different model on Groq, change the model = "..." value. You can override the default per user (no rebuild needed) by adding the following to your own ~/.Rprofile (open it with usethis::edit_r_profile()), then restart R:

if (nzchar(Sys.getenv("GROQ_API_KEY"))) {
  options(
    gander.chat = ellmer::chat_groq(
      model = "openai/gpt-oss-120b"   # set to the model you want
    )
  )
}

You can check the models currently available on Groq with:

curl https://api.groq.com/openai/v1/models -H "Authorization: Bearer $GROQ_API_KEY"

or see the official list at console.groq.com/docs/models. Groq updates its lineup frequently, so check there for the latest and best-performing models.

To change the default for the whole image instead, edit the Rprofile.site block in the Dockerfile and rebuild.

To use a different provider altogether, set the corresponding API key (e.g. OPENAI_API_KEY, ANTHROPIC_API_KEY) and configure ellmer (chat_openai(), chat_anthropic(), etc.) accordingly.