Interoperable and reusable data

Where does this story come from?
Cognitive neuroscience

Tell us your horror story, what happened?
In 2016, our cognitive neuroscience/AI lab started working on an ambitious plan to create a unique fMRI dataset. We quickly decided that we wanted to share this resource with the world, but we didn't fully consider what was needed for that. When we finished the project in 2018 and proudly sent it to a well-respected data journal, we thought we had created a useful dataset for the world. We were taken aback by the many pages with suggestions for changes and requests for extra documentation. One of the (very sensible) requests was to put the data into a standard structure, which required us to reprocess and rename every single of our thousands of files. If only we'd thought properly about this before starting the project..

Did you find a solution? How did this situation end?
Because of the unique nature of our dataset, it took us a very long time to put all the data in the standard format and apply the standard naming conventions. We are still working on addressing some of the other requests from our reviewers.

Was there a lesson learned? How could this horror be avoided?
If we would have carefully considered the requirements for an interoperable and reusable dataset (e.g. standard formats and naming conventions), we could have saved ourselves a great deal of work. This was a very valuable lesson on the importance of proper research data management for us. 

Data horror_03

Photo by Mika Baumeister on Unsplash