FAIR data

Let’s say you’re at your desk, reading a paper. It’s great, groundbreaking even. In fact, it has exactly the data you need to answer a very important question! You scroll down to the supplementary material, heart racing:

“Data from this study are available from the author upon reasonable request.”

Heartbreaking. Of course there is a chance that your reasonable request will be answered by a lovely new colleague, but there is also a chance that the author in question has disappeared off the face of the earth. In which case, you’re faced with having to spend significant time and money to rerun a perfectly good experiment. 

If you’ve been in science for any length of time, you’ve likely experienced this heartbreak. So how do we start to heal? 

The answer is FAIR data. FAIR is:

Let’s dive in to see what each of these means.

1. Findable

It’s somewhere in that 10-foot tall stack of lab notebooks. Not good enough. Findable data ensures that both your data and metadata are easy to find for both humans and computers. This means that all data and metadata are assigned global unique and persistent identifiers that are linked to each other and easily searchable. While this will save you a lot of searching through dusty notebooks, it is especially important if you want your data to be AI ready. Your machine learning models can’t train on data if they don’t know where to access it. 

How do we create findable data?

At Ometa Labs, we assign every individual file, metabolite, and spectrum a unique global identifier called a Universal Spectrum Identifier (USI). USIs not only make it easy to find a specific spectrum across our platform, but also link spectra to the files they came from. We also provide a metadata databasing system based on the ReDU framework. Finally, we assign each analysis a unique identifier that is searchable across the platform. These unique identifiers are linked to the exact parameters used to perform the analysis, making it easy to search old analyses and reproduce results.

2. Accessible

Once (meta)data is found, it needs to be easily accessed. Easily accessed does not mean emailing someone. If you’re working in academia, this access should be free and open (with appropriate authentication or authorization when necessary). If you’re in industry, that data should at least be accessible to other researchers within your organization. An example of accessible data is the MassIVE database, which allows users to browse and download over 700 TB of publicly-available mass spectrometry data after creating a free account.

How do we create accessible data?

We provide a suite of built-in tools to facilitate secure access to data and analyses.* For instance, we use those unique IDs to make it easy to search and share results. Just copy/paste the unique URL for your analysis, data, or visualization and share it with anyone within your organization. Users can also search databases of shared data and analyses to access, visualize, reproduce, and download results. Finally, if you are a fan of emailing, all analyses are linked to the unique user who created them, so you always know who to bother.

3. Interoperable

Gone are the days of small batch analysis. In the world of AI and deep learning, it’s important that your data can be integrated with all your other data. It’s also important in the world of normal humans, since nothing is worse than trying to compare results between data formats that are completely different. Interoperability includes formalized vocabularies for referencing data and metadata, links between data, and consistent databasing systems that work across platforms.

How do we create interoperable data?

Interoperability is one of the reasons we care so much about using open source file formats for our mass spectrometry data. Proprietary file formats are not compatible with data from other instruments. By using open file formats, you can analyze data from all your instruments using exactly the same tools and methods, with results output in the same format.

Tired of trying to find results on the one computer that has only copy of that one program? Ometa’s server-based integrated storage and analysis platform ensures interoperability by providing users access to the same data and results from anywhere. With unique logins for each user, researchers from across your organization can find, connect, and analyze data in a way that is shareable, traceable, and reproducible.

4. Reusable

Ultimately, the goal of making data findable, accessible, and interoperable is so it’s able to be reused. This could be by yourself, a colleague, or your organization’s newest shiny AI model. Making data reusable includes making sure data is accurately described, has appropriate usage rights, is associated with detailed provenance, and meets community standards.

How do we create reusable data? 

Every one of our analyses comes with a detailed list of every input parameter and file that was used to create the results. This list is available both on our platform and as an easy-to-download file that can be quickly added to electronic lab notebooks. Analysis data is stored and identified in a central location that enables easy reuse. We’ve even created clone button that allows you to reproduce an analysis with exactly the same files and settings in a single click. Finally, using open source file formats and USIs are part of the way we adhere to community standards for reusability. 

Finally, creating FAIR data is a process that shouldn’t end after going through a few bullet points. While we’re proud of how our platform adheres to these principles, there are many ways in which we are planning to further improve. If you are a client (or future client) and have suggestions or questions about how to make your data more FAIR, let us know!

*Since we work with private companies, accessible means access for users within an organization. If you’re an academic or non-profit researcher and are looking for the open access version of these tools, check out GNPS2.

Want to learn more?

The GO FAIR initiative grew out of the original paper on FAIR data and has a ton of helpful tools and resources.

Have more questions? Contact us.