Sarah Gibson - Using Binder for Reproducible Research

Introduction

The final talk of the day was given by Sarah Gibson who is based at The Turing Institute. She is also a maintainer of Project Binder, which is part of the Jupyter family. For those that don’t know, JUPYTER produces tools that support the main open source languages Julia, Python and R, as well as many others.

Reproducibility

Like Daniel Nüst in one of the day’s earlier talks, Sarah began by speaking about journal articles not being scholarship in their own right but an advert for it. The true scholarship comes from the complete software development environment and the set of instructions that generated the figures that are ultimately part of the article. If you want to read more about this approach visit The Turing Way: https://the-turing-way.netlify.app/

This brings the subject of Binder to the forefront. All of the prose, code and visualisation are put together in a public repository, such as GitHub. From here it can be made ‘Binder Ready’ which also means describing the software dependencies and resources required to reproduce the analysis.

Binder automates much of the process of creating the software environment making it available in a browser window. Once created the Binder project and analysis can be reproduced with a single click. To demonstrate this point Sarah used the work of the LIGO Scientific Collaboration to detect and show the first gravitational wave. You can see this research for yourself here: http://bit.ly/GravWaveBinder

Conclusion

One of the biggest challenges faced by the event’s participants was preparing the software environment ahead of reproducing the work. Many struggled with software libraries, packages and dependencies. Using Binder can easily overcome these making reproducibility much, much easier.

The Turing Way host 'getting started' articles for using Binder with the more popular programming languages:

You can follow Sarah on Twitter: @drsarahlgibson.

Return to article index