Q&A for Pistoia Debates Webinar on Benefits and Costs of FAIR

The EU report that is being referenced - is it the 'Turning FAIR Data into reality'?

'Turning FAIR Data into reality'? report for the EU Commission was mentioned by Alexandra & Drashti and is available at this public link.

The other EU report that is being referenced by James, Alexandra and Drashti?

The cost of not having FAIR research data by Price Waterhouse Coopers for the EU Commission is available at this public link.

Are semantic technologies the best approach to acheive FAIR or are there other technologies to look at?

Semantic technologies are often a good technical solution for implementation of FAIR because they support data integration through numerous mature standards such as ontologies, mappings, communication protocols and formats.

Can you please comment on the value for the pharma industry to have access to FAIR data from the public domain, as compared to proprietary data?

The pharma industry are more likely to be gain value from public and proprietary FAIR data. Conversly, data that is not FAIR is less likely to have value which is more likely to diminish faster over time.

Given that the case studies are convincing and the culture is changing is there any other "blocker" or would you say it is just a matter of time and FAIR is a given way forward?

Necessary culture change for FAIR implementation is likely to be be radical which will take time to implement. Priorities will need to be assigned and low hanging fruit should be sought to demonstrate value to management and end users.

If an organisation has made a commitment to incorporating FAIR into its overall data strategy, what are the key components that need to be considered to implement this in practice? If you were just starting to do FAIR, where would you start? If you were starting from scratch today on a global disease registry, what key three learnings would you implement?

Start by working out what each of the parts of FAIR mean to your use cases and user community: how will they find your data (through web browser, API, as part of another platform), what does findable and accessible mean to you (is it public, private within organisation, private within specific legally-defined group), interoperable and reusable could be taken across a spectrum of ‘fully reusable and interoperable with existing fully, richly marked up data’ to 'starting from nothing’. But do not try to "boil the ocean". This means pick your use cases and work out with your user community what’s needed. Making a lot of data partly FAIR is better than having just a handful of datasets 100% FAIR. Use public standards whenever possible and reasonable (OBO bio-ontologies, CDISC, etc). This future proofs the metadata to some extent. On disease specifically, there are a lot of competing standards so no easy answer. For example, SciBite use Orphanet and MeSH as our public disease bases because they cover most use cases but it’s not a hard and fast rule - they use SnoMed too because it’s also well used and they get asked to use EFO on some projects. Again, it depends on the use case; what integration with other marked up data is required?

Has anyone estimated the COSTS of making their R&D data FAIR? (amount of staff, $?)

To date no such cost estimates have been disclosed publicly which is discussed by Wise et al in a feature article published in Drug Discovery Today (doi.org/10.1016/j.drudis.2019.01.008).

Does applying FAIR retrospectively pay off, or is it better to focus on FAIR data production going forward?

Both are needed at present. Retrospective implementation (FAIRification) is being done selectively where clear business value justifies the cost. However, it is widely recognised that data management planning, guided by FAIR, is increasing in importance for reproducible science-based, drug discovery.

How much of the eTOX is publicly available? Are there plans to release more? Is the eTOX shared in a FAIR way? ie appropriate metadata etc? On eTOX - not relating to open/closed - putting it on the web does not equate to FAIR right? By what criteria have you measured eTOX to be FAIR?

eTOX (http://www.etoxproject.eu) is one of the data resoources being FAIRified under FAIRplus (https://fairplus-project.eu) where more details about availability can be found. It is important to understand that FAIR data can be closed due to confidentiality (e.g. patient or proprietary sensitivity). However, access authorisation should use open security standards according to the accessibility guidelines of FAIR.

How can using common data models (e.g. OMOP) help FAIR as well?

Usage of common data models e.g. OMPO, vocabulary and ontology standards will improve interoperability (the I of FAIR), which can be very challenging to implement.

What about FAIR for analysis?

Planned management of data and metadata guided by the FAIR principles is mare likely to allow better quality analysis by a wider range of tools and methodologies. In addition to this, the software development community are also embracing better practices for open software development. For example, the 4OSS recommendations are described by Jiménez RC, Kuzak M, Alhamdoosh M et al. 2017 Four simple recommendations to encourage best practices in research software. (doi: 10.12688/f1000research.11407.1).

Q1. Where is your workplace?

Biopharma

Agrifood

Tech

Academic

Other

23

2

33

7

6

71

32%

3%

46%

10%

8%

Q2. How mature is FAIR implementation in your workplace?

Maturity level 1

Maturity level 2

Maturity level 3

Maturity level 4

Maturity level 5

14

32

10

10

2

68

21%

47%

15%

15%

3%