Consider switching to https://identifiers.org/ URI or CURIEs

[matthias-samwald]

Looking at the currently used identifiers, it seems to be a mixture of OBO PURLs and ad-hoc solutions for some of the molecular entities. identifiers.org identifiers should make it possible to elegantly cover everything (either URI or CURIEs -- the latter would be more compact, but of course might be less ideal with OWL tools).

[callahantiff]

Hi, @matthias-samwald thanks for your interest PheKnowLator! 😄

We are currently using a mixture of URIs, those that come with each ontology (most are OBO PURLs) and the non-OBO PURLs we add for other sources. For the latter, we have opted to choose the primary identifier that is recommended for that source.

Happy to accommodate other solutions and your suggestions would certainly make things more elegant (and easier to write programs against 👩‍💻 ). I also can't think of any negative consequences to our current workflow for either option you suggest. Let me discuss this with the rest of the team and get back to you next week.

[callahantiff]

After taking time to think about this and to review available resources for identifier resolution, we have decided to give Bioregistry a try, which also happens to include identifiers.org as a provider. This will still enable us to address the same types of issues that were highlighted above and will more easily integrate with the existing Python framework.

Please let me know if you see any major issues with this. Thank you!