2021 Toronto Workshop on Reproducibility

A little over a week ago I had the opportunity to attend and present at the Toronto Workshop on Reproducibility. It was a full 2 days packed with talks on various aspects of reproducbility: metascience, case studies, software tools, etc. Rohan Alexander and his colleagues at the University of Toronto did a great job organizing the ambitious online event. Not only did the event go smoothly (despite the attempts of my web cam to sabotage my presentation, as you’ll see below), but the recordings were edited and posted to YouTube almost immediately!

In my talk I explained how the R package workflowr helps you make your data analysis projects more organized, reproducible, and shareable. Most exciting for me was that this was the first time I was able to include workflowr.io in a public presentation.

As always, my slides are available for reuse and remixing with attribution under the CC-BY license. Check out the repository workflowr-presentations to access the slides for this and past presentations.

I also want to highlight a few of the other talks that I found interesting, with a focus on software tools for reproduciblity:

• Meghan Hoyer and Larry Fenn presented Project organization with DataKit. DataKit provides project templates for data analysis projects and also provides functionality to integrate version control.

• Julia Schulte-Cloos presented on her new R package reproducr, which provides an extensive custom R Markdown output format makes it easier for researchers to write their manuscripts reproducibly.

• Simeon Carstens presented Reproducible software environments for data analysis with Nix. Nix is a purely functional package manager, which is a different approach than other package managers you may be familiar with (e.g. APT, conda, renv). It provides lots of reprocibility guarantees, a large collection of packages ready to install, integration with Docker, and more. Check out his demo project to learn more.