FAIR Principles and Open Science

The information given in the video above is also available as an info slide deck:

Click here for the Info slide deck "FAIR Data in fundamental research"

Practicing Open Sience means to practice science in such a way that it is transparant, accessible and reproducible.

Being "FAIR" is a method to help you to practice Open Science. Producing FAIR outputs will improve the discoverability, reusability, and long-term availability of your scientific products.

However, practicing Open Science does not mean by default that all your research data are available for everybody. Some data may simply not be shared with everybody, for example because of the privacy of participants or because of intellectual property rights.
In such cases, you can create openly accessible FAIR metadata and keep the actual data-files under access control.

Watch our video regarding metadata in (fundamental) research to find out what metadata are, how you benefit most and how to produce metadata in an efficient way:

The information given in the video above is also available as an info slide deck:

So, the main principle of Open Science is to be:

Therefore, consider making as much as possible elements of your (meta)data available, but within the restrictions of the law, to support Open Science.

Nowadays, the GO FAIR initiative keeps the record up to date how each individual principle can be achieved.

As you can imagine, complying to the suggested guidelines and standards can be more challenging for some than for others.

You can look up relevant standards, polices, and more on FAIRsharing.org; look up possible repositories and archives and their standards on re3data.org; and available bioinformatics resources on the Elixir bio.tools registry.

With every standard or best practice you implement in your workflow, software- and hardware set-up, and output sharing/publishing you will increase the FAIRness of your work.

The FAIR principles movement desires to create a world where humans and machines alike can easily interact with data, code, and other (digital) objects via the world wide web.

That means that in an first instance the metadata and files need to be standardized, harmonized, and finalised by humans. This completed version then is shaped into a published/ archived data-set with the aid of standardised and/ or certified data repositories and archives. Thanks to the technical set-up this published/ archived data-set now can be discovered, viewed, and interacted with by humans and machines.

However, the true aspiration of the FAIR principles is that machines can easily interact and operate with metadata and data. Humans would use machines in the future to search, combine, analysis, share, and save files that are located in different resources around the globe. Of course future machines will be able to follow through certain tasks and processes on their own as well.

In order to achieve this, machine actionable standards needs to be implemented and maintained.

By submitting your files in commonly used file-formats, accompanied by the necessary documentation in a readme-file, to a standardised and/or certiefied data archive or repository (like Zenodo.org) you can achieve a certain level of FAIRness.

The more standards the archive/ repository applies to, the better the FAIR compliance scoring will be. The picture above illustrates what standards/ features a data archive/ repository should offer you in order to achieve a good level of FAIR compliance.

Bottom line: this is a good first step to publish and archive your output in a FAIR format. Yet, the contents of the actual files are not machine-actionable in this way. The machine can crawl and index all metadata, but won't be able to interact with the more important information within the files of this data-set. However, a human can use the original files and apply them in any set-up.

One way of achieving FAIRness on file-content-level is to convert your original files into linked data. This can be done with transforming your files from e.g. CSV-format (comma sperated value) to RDF-format (resource description framework).

This subsequent means that your data now exists in the shape of a knowledge graph, and is expressed in RDF triplets that have the simple syntax: subject -> predicate -> object.

This graph database now consists of facts extracted from the original files, and can be enriched with more facts based of auxiliary data, documentation, publications etc. Applying ontological terms, or even implementing ontologies is the next step in increasing the FAIRness of this resource.

The Ontology Lookup Service (OLS) as well as FAIRsharing.org help you find appropriate ontological terms to use, or whole ontologies to implement.

Bottom line: this is a challenging step. You either need expertise on how to create RDF-graphs, and apply SPARQL in your team, or have access to the resources and services that can help you with this transformation. Thanks to the nature of ontologies, machines can interact and operate very well with this FAIR resource. Please keep in mind, that once your files are transformed into linked data all future infrastructure that wants to interact with it has to be able to handle linked data. In this scenario the original files (i.e. CSV) are no longer applied nor executed.

Increasing the FAIRness of software, code, and scripts in principle follows the same workflow mentioned above for data. You can both improve things on metadata and content levels. On fair-software.nl you will specific guidance.