News
FAIRplus project ends: ‘FAIR-by-design is the way to go’
Four years of FAIRplus has provided academia and industry with tools to start implementing the FAIR Principles. ‘We can now move beyond the need to convince people about the benefits of FAIR, and actually start implementing it.’
31 December marked the last day of the FAIRplus project, a four-year € 7.8 million IMI project with 21 partners from academia and industry. FAIRplus aimed to improve the availability and usefulness of life science data, i.e. making it more FAIR (see explanation below). ‘After four years, I’m proud with what we achieved’, says Jan-Willem Boiten, Lygature’s Data Portfolio Lead and responsible for the project’s communication and outreach. ‘We really reached a next level with FAIR, going from “why” to “how”. An important goal of FAIRplus was to develop a practical process for FAIRification, going beyond the sometimes idealistic and abstract idea of FAIR.’
Thus, FAIR is not a goal in itself, Boiten stresses, but a means to reach your business or research goals. It provides an answer to the significant challenges when it comes to research data, which are heterogeneous by nature: standards and ontologies are often not in place or different versions are used between research institutions and industry, different languages are used (both verbal as well as in programming), metadata are absent or incomplete, data are saved in proprietary formats, etc. These are all obstacles preventing datasets to be fully exploited and making cross-comparison of datasets difficult. ‘Researchers could therefore miss pivotal correlations or important insights’, Boiten explains.
‘An important goal of FAIRplus was to develop practical tools, going beyond the sometimes idealistic and abstract idea of FAIR’
FAIRplus was a unique IMI project in the sense that it involved working with data generated in other IMI projects. Small groups of data experts, bringing in the necessary expertise and technical skills (‘squad teams’), assessed many IMI projects over the course of the FAIRplus project. The final selection of about twenty projects, which were mainly public-private partnerships in drug development and medical research, was based on a systematic process factoring in societal impact, availability of data, and scientific value, among others. The insights obtained from these FAIRification processes were then used to develop the practical tools and guidelines for future research.
One of the most important assets developed in FAIRplus is the FAIR Cookbook, a collection of FAIR ‘recipes’ that each address different challenges during a FAIRification process. ‘The Cookbook aims to lower the bar for a researcher to start their own FAIRification’, Boiten explains. ‘It provides you with indicators of FAIRness, a data maturity model, and an idea of the skills required to improve FAIRness.’
The Dataset Maturity (DSM) model is another key result from the FAIRplus project. It is a comprehensive reference framework for improving FAIRness in research datasets. The DSM model presents five levels of FAIR, each characterised by increasing requirements across three categories: content-related, representation and format, and hosting environment capabilities. This way, the DSM model helps assessing the level of FAIRness of a dataset.
Other notable outcomes of the FAIRplus project are the Fellowship Programme (a hands-on training in FAIR data management), a flyer to promote the FAIR Cookbook, an interactive digital flowchart to help first-time FAIR users (in the final stage of development), and a collection of use cases. These use cases each represent the FAIRification process of a single IMI project, for example from eTOX, APPROACH, and COMBINE. ‘Not only the success stories’, Boiten points out, ‘but also the pitfalls, difficulties, and challenges one can expect to encounter when starting a FAIRification journey.’
‘FAIR is not a goal in itself, but a means to reach your business or research goals’
The use cases resulted in a list of seven ‘lessons learned’. The most important one after four years of FAIRifying existing datasets? The answer is clear: try to avoid it and choose FAIR-by-design instead. ‘The FAIRplus project had to work with projects that already started or were already finished’, Boiten explains. ‘It has become very clear that although FAIRification of existing datasets is often doable, the result is less optimal than when you implement the FAIR principles before you start generating your data. It simply saves a lot of time.’
Boiten believes FAIRplus has clearly demonstrated the benefits of FAIR. ‘I hope we no longer have to put a lot of energy in convincing senior managers and policy makers about the added value of FAIR. We can move beyond that and use our tools to implement FAIR where it can make the biggest difference. Not only in single projects, but in the wider research community in academia and industry.’
The FAIR Principles
The key acronym in the FAIRplus project is ‘FAIR’, which stands for ‘Findable, Accessible, Interoperable, Reusable’. Familiarizing yourself with FAIR comes with learning novel nouns and verbs like FAIRness and FAIRification, reflecting the level of data being FAIR and the process of making data FAIR, respectively. The theory of FAIR was first published in 2016 by Wilkinson et al in Nature and has since then steadily gained ground in the life sciences research community. The FAIR Principles are the beginning of a solution to the vast increase in research data and the need to make it findable and accessible for the broader research community (academia and industry), interoperable between different digital platforms and software, and reusable for future research.