rworkflows: taming the Wild West of R packages

Author(s): Brian Schilder,Alan E Murphy

Affiliation(s): Imperial College London

Social media: https://twitter.com/BMSchilder

Despite calls to improve reproducibility in science, achieving this goal remains elusive even within computational fields. >50% of R packages are currently distributed exclusively through GitHub, a code repository that does not require software to adhere to any coding standards or even run at all. This has contributed to the scientific landscape becoming the “Wild West” in terms of code usability and reproducibility. To address this we developed rworkflows; a suite of tools to make robust continuous integration accessible to R package developers of all skill levels. Once implemented within an R package, any change to its code on GitHub automatically triggers a workflow to launch virtual machines across multiple operating systems, install all software dependencies, run code checks, and deploy a ready-to-use container to DockerHub. rworkflows makes reproducible best practices easy to implement, even for novice programmers. By making this suite freely available (https://github.com/neurogenomics/rworkflows), we hope to take a step closer towards taming the Wild West of R packages.