In a previous article, I checked out how ontologies will help deal with a few of the life sciences’ huge knowledge challenges by FAIRifying knowledge, making it Findable, Accessible, Interoperable and Reusable. Ontologies – human-generated, machine-readable fashions of a website – will help to make knowledge FAIR and usable from the purpose of creation. This reduces the time scientists spend trying to find data, avoids duplicated experimental work, and makes knowledge “machine prepared” to energy AI and machine studying tasks.
Nevertheless, deciding to implement ontologies into your knowledge administration practices might be daunting. It may also be a tough promote to enterprise stakeholders because the ROI is commonly not rapid. On this article, I define the challenges and provide 10 pointers for kickstarting your ontologies journey.
A enterprise, cultural and scientific problem
Implementing any ontology is a specialised activity. Efficiently doing so requires knowledge collated from many sources to be constantly formatted, structured and harmonized. In any data-heavy subject this can be a problem. However within the life sciences this problem is especially acute – with knowledge sources together with printed literature, experimental knowledge, and affected person and scientific information, containing graphs and tables, biomedical photos, social media knowledge and voice recordings.
Life science organizations should additionally take into account enterprise and regulatory necessities. Firms wish to guarantee any ontology complies with strict governance processes and has sturdy model management to provide a visual audit path, while additionally needing a system agile sufficient to make adjustments simply. Constructing a community of ontologies that may concurrently permit these ranges of flexibility and management is tough and time consuming.
As well as, there may be rising demand for ontologies to be extra “democratic” by permitting a spread of customers throughout the enterprise to contribute to their growth. This widens the pool of information that feeds an ontology, so it’s extra correct and displays the wants of its customers. Nevertheless, this requires a shift in cultural mindset – now not “it’s my lab and my knowledge”; quite “it’s the corporate’s knowledge and its FAIR”.
The ultimate, and doubtlessly most crucial, problem is proving the worth of ontologies and FAIR tasks to stakeholders. As with every large-scale, complicated venture, ROI is medium-to-longer time period, and ontology tasks could also be in danger within the brief -term. Subsequently, to maximise the success of your ontology venture, these are 10 issues to keep in mind:
1. Uncover what’s in place already
Earlier than planning a brand new venture, knowledge groups ought to establish what ontologies are already getting used inside their group – be that public ontologies or bespoke terminologies created in-house. Constructing on present work accelerates progress and offers early wins to current to stakeholders.
- Rebuild, reuse, recycle
Work on life science ontologies has been ongoing for a lot of many years, which implies there may be an present open-source framework to attract on. Public ontologies, equivalent to MeSH from the NIH are an incredible start line. Utilizing what’s already out there as a basis to your personal ontologies is a straightforward solution to make tangible progress.
- Discover your FAIR champions
The businesses I’ve recognized to have essentially the most success are those that have “FAIR champions” who perceive the challenges mentioned above. FAIR champions don’t should be specialists in semantics or knowledge science, they must be tenacious, dedicated to the venture and able to enthusing stakeholders round targets and milestones.
4. Create a URI technique
Uniform Useful resource Identifiers (URIs) needs to be established in the beginning of any ontologies journey. URIs present a method of finding and retrieving assets on a community – just like net handle URLs. URIs are tough to vary as soon as they’re in place as they denote the distinctive ID for an entity. A typical URI technique from the outset reduces the prospect of error and will increase standardization throughout the enterprise.
- Map sparingly
Mapping ontologies is a time-consuming and never-ending activity, with a always transferring goal as ontologies evolve with our understanding of the life sciences. Attempt to restrict mapping wherever attainable by limiting to a small variety of ontologies (ideally one!) per area and never bringing in or creating a brand new ontology the place one is already in use for that space.
6. Simplify your ontology choice
Minimizing the variety of ontologies used reduces the burden of getting to maintain them in sync, or map between them. Deciding on public ontologies additional simplifies integration of private and non-private knowledge. For instance, in case your area is illnesses, you would possibly use Mondo Disease Ontology to cut back your workload.
- Begin small and iterate
You possibly can’t deal with all your knowledge without delay. It takes too lengthy to see returns and moreover, it’s most likely not possible. Begin with one use case at a time – prototyping to see what works and utilizing these learnings to iterate. Knowledge entry tasks equivalent to assay registration are a great start line as they have already got a selected construction. It could possibly be a easy swap from getting into free textual content to picking from a drop-down listing of assays out of your area ontology of alternative. This makes knowledge FAIR from the outset; a normal listing ensures the knowledge is constantly recorded, interoperable and facilitating future reuse.
- Don’t let the size of the drawback put you off
A corporation doesn’t want a mannequin of the whole technique earlier than getting began on an ontology venture. As talked about, iterative successes are key. For instance, consolidating lists of phrases and importing them centrally the place folks can contribute, or beginning in an space that you already know already has comparatively good knowledge administration that may be constructed on to indicate worth rapidly.
- Discover the enterprise worth
One of many challenges with any knowledge administration endeavor is that enterprise worth is medium to long run. To win funding and make sure the venture strikes forwards, discover the short-term impacts and hyperlink them to enterprise outcomes. For instance, present that making use of an ontology to bioassay creation has diminished time spent trying to find knowledge by X variety of hours. Or present that utilizing an ontology has made it attainable to reuse worthwhile datasets that had been beforehand siloed away. Tangible outcomes should be shared early and infrequently with enterprise leaders.
- Empower subject material specialists
It’s important to empower and belief your subject material specialists. That features each your knowledge scientists and your area specialists who can give you the relationships and information of the sector to develop the ontology correctly. Give them the best instruments to do the job, and a sensible timeframe through which to ship.
Driving future innovation
Using ontologies in knowledge administration is prime to driving future innovation. Transformational life science leaders are spending time and assets on embedding sturdy knowledge practices. They know that when scientists are in a position to make environment friendly use of knowledge being produced, the trail to new discoveries is accelerated. False begins, useless ends or being on the flawed monitor are extra widespread than they must be. This may be demotivating and irritating.
Shortening the drug discovery lifecycle is not only worthwhile when it comes to shareholder worth and profit to sufferers, it can additionally enhance workforce productiveness. Scientists are extra engaged when they’re assured that the trail they’re pursuing will both finally achieve success or will “fail quick”. With the best technique and experience, organizations can use ontologies to make sure they’re on the forefront of latest breakthroughs.