Levelling Up Reproducibility: Containerised RStudio with Docker Compose and renv

As data scientists and researchers, we constantly strive for more robust and reproducible workflows. We've explored tools like Docker, literate programming, conda, and Quarto in the past.

Today, we dive into a powerful combination: containerized ?????????????? using Docker Compose and renv. This setup elevates reproducibility by ensuring consistent environments and package management across different systems. You can find the whole project on our GitHub page.

The Power Trio: Docker Compose, ??????????????, and renv

Let's break down the key components of this reproducible analytics setup:

1. Docker Compose: Simplifies multi-container application management.
2. ?????????????? Server: A web-based integrated development environment (IDE) for R.
3. renv: An R package for project-specific dependency management.

By combining these tools, we create a portable, version-controlled environment that is easily shared and reproduced.

Setting Up the Environment

Here are the key files that make up our reproducible ?????????????? environment.

The Dockerfile

Our Dockerfile starts with the ????????????/?????????????? base image and installs necessary system dependencies:

This Dockerfile ensures all necessary system-level dependencies are installed, sets up renv, and configures the ?????????????? Server environment.

The docker-compose.yml File

Docker Compose lets us define and run multi-container Docker applications. Here's our ????????????-??????????????.??????:

This configuration:

• Builds the ?????????????? container using our Dockerfile.
• Maps port 8787 for web access.
• Sets an environment variable for the RStudio password.
• Mounts volumes to persist data and configurations.

The Magic of renv

The real power of this setup comes from renv, which manages project-specific R package dependencies. By using renv:

1. All collaborators use the same package versions.
2. The project becomes more portable and reproducible across different systems.
3. Package conflicts between projects are avoided.

The ????????.???????? file, which is mounted in the Docker container, tracks all package versions used in the project.

Getting Started

To use this setup:

1. Clone the GitHub repository containing these configuration files.
2. Run ./??????????????_??????????.???? from that repository to create the necessary directories that are mounted in the YAML file.
3. Execute ????????????-?????????????? ???? --?????????? to build and start the RStudio container.
4. Access RStudio by navigating to ????????://??????????????????:???????? in your web browser.

Benefits and Best Practices

1. Version Control: Include your Dockerfile, ????????????-??????????????.??????, and ????????.???????? in version control to track environment changes.
2. Portability: Easily move this setup between different machines or cloud environments.
3. Consistency: Every team member works in an identical environment, reducing "works on my machine" issues.
4. Isolation: Projects are isolated, preventing package conflicts between different analyses.
5. Reproducibility: Docker and renv ensure that your analysis can be exactly reproduced, even years later.

Conclusion

By leveraging Docker Compose, containerized ??????????????, and renv, we've built a robust, reproducible environment for R-based data analysis. This setup enhances collaboration and ensures that our work stands the test of time, allowing it to be easily verified or expanded upon by others in the scientific community.

Reproducibility is about more than just code—it's about the entire computational environment. With this approach, we're one significant step closer to truly reproducible data science.