This blog is part of the event report series on International Open Data Day 2017. On Saturday 4 March, groups from around the world organised over 300 events to celebrate, promote and spread the use of open data. 44 events received additional support through the Open Knowledge International mini-grants scheme, funded by SPARC, the Open Contracting Program of Hivos, Article 19, Hewlett Foundation and the UK Foreign & Commonwealth Office. This event was supported through the mini-grants scheme under the Open Research theme.
Brainhack Global 2017 consisted of 40 satellite events around the world on March 2nd, 3rd and 4th 2017. Brainhack is a unique hackathon and unconference that brings together researchers with disparate backgrounds to collaborate on open science projects in neuroimaging.
A map of all the Brainhack Global 2017 satellite sites.
We had more than 40 participants at the Cambridge satellite event of Brainhack Global representing early career researchers from multiple university departments and research institutes. Over three days we supported each other as we learned new skills and developed analyses to investigate neuroimaging data.
The organising committee worked hard to foster a warm and friendly atmosphere. We know how hard it is to go outside their comfort zone and we wanted to make sure that everyone felt welcome. We had a strict code of conduct and made it very clear that everyone was welcome, no matter your race, gender, level of coding ability, or choice of programming language.
The talks that kicked off our first two days together were a great opportunity to get excited about new areas of research. They inspired participants to consider how we can all play a role in the future of big data and open science in neuroimaging.
František Váša and Jakob Seidlitz gave an excellent tour of the fundamentals of network neuroscience and introduced some of the freely available datasets that researchers could utilize to carry out this type of research. I gave a presentation on how to make your results reproducible and Dr Niko Kriegeskorte showed how his lab are using deep neural networks to understand visual perception.
Of particular note for open data advocates was Dr Darren Price’s presentation of the Cambridge Centre for Ageing and Neuroscience (Cam-CAN) dataset. You can apply for access to the data from 700 adults, aged between 18 and 88 years old, who were scanned using structural and functional Magnetic Resonance Imaging (MRI) and magnetoencephalography (MEG). These participants also completed a large battery of behavioural tasks and questionnaires outside of the scanner. There’s more information about the study here.
We also had an excellent discussion with members of the Wellcome Trust’s open research team on incentives and roadblocks for early career researchers who want to open their academic workflow. We covered some of the key reasons that Open Data is important:
- Open data allows us to meet one of the core principles of the scientific method, that someone else could reproduce your results. It’s what separates science from magic.
- Open data also means you can get more out of a dataset. Re-using the data for a different purpose is a more efficient use of all the money spent on collecting it. A return on investment to the funder, which for Brainhack Cambridge participants is often the UK tax payer, is especially important when the datasets are expensive/difficult to collect.
- Open data leads to better innovation and collaboration. By bringing together ideas from many different disciplines to understand the data from many different points of view, a diverse group of people can analyse the data in ways that you might never have imagined.
There are, however, major challenges associated with sharing human brain imaging data:
- It may be possible to identify individual people from their data. It can be very difficult to anonymise some data sets and sensitive information such as their history of mental health difficulties or intelligence measures should be protected. It can be very hard to know how to best navigate the ethics of sharing data and there may even be different requirements in different countries.
- Some human neuroimaging datasets are large (a few hundred gigabytes per acquisition) and therefore many existing repositories are not suitable. Although members of Brainhack Global were working on the Brain Imaging Data Structure project it is not yet widely adopted. This means it is difficult to organise data in a way that it is interpretable to other researchers.
- Not only are the datasets large, but as it is very expensive to collect brain imaging data, there are many stakeholders and collaborators. It is unlikely that an early career researcher will be able to make the decision to share the data from their study.
- There is a steep learning curve associated with learning new skills, software or platforms. Adding an additional open data burden on PhD students and postdocs may require a lot of dedicated time that they simply do not have.
We all worked hard to help each other address this last challenge – that it takes too much time to figure out new techniques on your own – during the open hacking time at Cambridge Brainhack. It was a great opportunity to share what we know and learn from each other. We got some fantastic feedback that our participants left feeling motivated to develop their open science skills, and confident that they weren’t the only one struggling. We’re excited to continue to build our community together, locally, and around the world.
We are incredibly grateful to SPARC and Open Data Day for their mini-grant that provided lunch on the last day for our hungry, hard-working participants. We would not have been able to hold Cambridge Brainhack without their support, along with our other fantastic sponsors: The Wellcome Trust, Overleaf, PLOS, Mozilla Science Lab, MRC Cognition and Brain Sciences Unit and the University of Cambridge Department of Psychiatry.