Open Dream Kit presence at the second RSE conference
The Manchester Museum of Science and Industry (MOSI) saw the second Research Software Engineering (RSE) conference on September 7-8th, 2017. Over 200 attendees gathered to discuss ways to better support and advance science via software, innovative computational tools, and policy making. There were more than 40 talks and 15 workshops covering a diverse range of topics, from community building to imposter syndrome, data visualization, and High-Performance Computing.
Attending the conference is an excellent opportunity to integrate within the international RSE community and appreciate how much this has grown over the last few years. All thanks to the great work done by RSEs within their institutions and their efforts to make software a valuable research asset. It will be, for certain, interesting to see how this will continue to grow and evolve as policy changes take place and more research councils, funding bodies, and research institutions acknowledge the importance of research software and its scientific impact.
OpenDreamKit member, Tania Allard, ran a hands-on workshop on Jupyter notebooks for reproducible research. This workshop focused on the use of Jupyter notebooks as a means to disseminate reproducible analysis workflows and how this can be leveraged using tools such as nbdime and nbval. Both nbdime and nbval were developed by members of the OpenDreamKit project as a response to the growing popularity of the Jupyter notebooks and the lack of native integration between these technologies and existing version control and validation/testing tools.
An exceptional win was that this workshop was, in fact, one of the most popular events of the conference and we were asked to run it twice as it was massively oversubscribed. This reflects, on one hand, the popularity of Jupyter notebooks due to the boom of literate programming and its focus on human-readable code. Allowing researchers to share their findings and the code they used along the way in a compelling narrative. On the other hand, it demonstrates the importance of reproducible science and the need for tools that help RSE and researchers to achieve this goal, which aligns perfectly with the goals of OpenDreamKit.
The workshop revolved around 3 main topics:
- Version control of the Jupyter notebooks
- Notebooks validation
- The basics of reproducible software practices.
The main focus was on how tools like nbdime and nbval can support people already using Jupyter notebooks but have struggled to integrate these with software best development practices due to a number of limitations on the existing tools. Then, we followed on other actions that can be taken to ensure that their data analysis workflows were reproducible and sustainable. This lead to a number of interesting discussions about the topic and allowed for the attendees to share their previous experiences regarding reproducibility and/or the lack thereof in different research areas.
We plan to run a set or workshops around reproducibility over the duration of the ODK project and we’ll make sure to report on them here too. Finally, all the materials are licensed under CC-BY and can be found in this GitHub repository .