More Than Just Data Sharing: Why EBRAINS’ Data Curation Enhances Research Impact

An interview with Jan Bjaalie about why publishing data on EBRAINS increases research impact.

The EBRAINS research infrastructure provides a data sharing platform for neuroscientists. Why do researchers decide to publish their data on EBRAINS?

We are observing a complete transformation from a focus on journal articles as the single primary outcome of research to recognition of the underlying scientific data as first-class research products. A shift of this kind is far from trivial. When the first research journals were launched in the 17th century, there was still a lot of scepticism within the scientific community towards publishing their work in this new format. In a similar way, today’s move towards open data sharing challenges long-standing traditions; yet, it is increasingly recognised as essential for generating valuable research products that reflect our technological advancements and modern analytical capabilities.

Many funding bodies now require researchers to publish their data online, and citeable DOIs of published data are becoming increasingly important, not least for young researchers. When we evaluate researchers these days, we look beyond publication lists and place emphasis also on transparent research practises, including data sharing. But in addition to providing DOIs, publishing data with EBRAINS offers other major benefits that increase research impact. Data are most valuable for scientific progress when they can be found, interpreted and effectively used by others. The EBRAINS infrastructure has a unique offering in this regard because it includes the services, including curation, that enable just that. 

"EBRAINS plays a central role in the general transition of research towards a data-centric approach."

EBRAINS has been designed specifically for the field of neuroscience, and it provides researchers with visibility within the community. While we tailor the system to the needs of neuroscientists, our pioneering approaches are generalizable to all research fields, ultimately benefiting the entire research community. So EBRAINS plays a central role in the general transition of research towards a data-centric approach, moving beyond journal articles to the underlying substance they contain.

How does EBRAINS compare to other data repositories?

There are data repositories that are quite niche, for very specific subdomains of neuroscience, and others are very general, covering all kinds of research domains. EBRAINS covers an area in the middle: It provides relevant services for a broad number of fields within neuroscience. EBRAINS also provides an option of attaching licences to the data which communicates the conditions for its use, and, in contrast to the more general repositories, EBRAINS makes use of specific metadata frameworks, harmonized descriptions, file organisation rules, and curation. This is not the standard for all systems, and it makes a fundamental difference.

"If you want your research to have impact, just simply publishing data is not enough."

Why is data curation so important?

One may consider publishing data somewhere as quickly as possible and presenting it in whichever way seems good to the researchers who collected them. However, if you want your research to have impact, just simply publishing data is not enough, because it is not a given that someone else will be able to understand your file structure, how your data were captured and what they really are. 

Data curation is a prerequisite to making data findable, correctly described, interpretable, comparable to other data, and ready for use with available tools and workflows. It ensures that strong, accurate metadata are correctly assigned to each research product and that data descriptions follow agreed, well-defined vocabularies and terminology frameworks. 

In EBRAINS, metadata are aligned with the openMINDS metadata framework and curated metadata are registered in the EBRAINS Knowledge Graph, so that datasets and related models, software, and workflows can be consistently discovered, compared, and reused. This approach is designed to scale dynamically, with terminologies and mappings continuously updated as science evolves and new concepts and terms are introduced. Data curation ensures that the rules are followed and the basic information about the data and its relations to other research products are in place. This is a huge advantage over general data repositories. 

"We have a vision for gradually transforming this process in a way that makes both research and publishing more efficient."

What does this curation service involve?

It usually starts with researchers sending us a curation request describing what they would like to share. The EBRAINS team will then guide the data provider through a stepwise process that includes supplying all required metadata, revising the folder structure that the data are stored in, and drafting a clear textual description of the data. Every step is carefully checked by an EBRAINS curator. These steps can take some time, but it is truly worth it to ensure the data to be most useful for the community. We are also working to gradually make all steps more automated. Finally, the data are uploaded into the storage system, all information is published through the EBRAINS platform and made discoverable through the Knowledge Graph.

Currently, most scientists contact us after their research has been completed, but we have a vision for gradually transforming this process in a way that makes both research and publishing more efficient: Researchers may already consider how they want to share their data before they start designing and performing their experiments, for example, by defining how to organise the data in terms of standardised file structures and formats and what kind of metadata should be collected throughout. This will not only facilitate data sharing, but it will also make a big difference for the research groups themselves. For example, it would become much easier to retrace and interpret the work of former students and postdocs, who have already left. If groups have good data organisation from the beginning, they will benefit from their own data to a greater extend. In my view, this is not an insignificant issue.

Interview conducted by Helen Mendes and Lisa Vincenz-Donnelly, edited for length and clarity.