Making your biomedical dataset more appealing

If you have a dataset that could be valuable for biomedical research, here are a few key points to consider from our perspective as data integrators and users in research-centric biotech and pharma projects.

Sharing this from my outbox, it is written from the perspective of a biomedical data integrator and user in small, mostly research-centric biotech/pharma projects that involve acquiring, centralising (databases/knowledge graphs), and feeding data into ML/AI workflows.

  If I had a large proprietary, potentially commercial, dataset with the potential to impact drug discovery, here's what I would do:

• Make myself (re-)acquainted with FAIR data principles. I would evaluate the costs and benefits of making the data more findable, accessible, interoperable and reusable.

• Publish. Domain specialists often require a reliable reference before engaging with data, ideally through a peer-reviewed publication. Publishing a use-case-driven analysis or modeling study can attract more attention than the dataset description alone.

• Make it visible in domain-specific aggregators. We list several ones we use at https://biokeanos.com/product#metasources. Some of them allow for direct submission.

• Open source. At least in part. Even a 10% lateral or vertical slice of the data can speed up the integration and drive the recognition. The right licensing is important. In smaller organisations, researchers are more likely to explore non-gated data. And who knows, maybe the data might even be used (and cited) without having to spend effort on persuading academics?

To reiterate - the points reflect a specific perspective shaped by experience in smaller, ML/AI research projects and may not necessarily translate to your circumstances.