A weird idea on using OBO Foundry metadata to standardise ontology level metadata

[matentzn]

The idea is this:

We generate standard ontology metadata for all ontologies from the OBO Foundry metadata. For example, we translate the ddpheno.md metadata file in the OBO Foundry metadata to OWL like this: https://raw.githubusercontent.com/information-artifact-ontology/ontology-metadata/metadatata/src/metadata/obo/ddpheno.owl.

Every ontology that wants to does then follows these steps to include the metadata:

• Create an import module to its own ontology metadata:

https://raw.githubusercontent.com/information-artifact-ontology/ontology-metadata/metadatata/src/metadata/obo/ddpheno.owl

• Merge the OBO ontology metadata like this upon release (something like this):

robot merge -i $(SRC) -i imports/ontology-metadata.owl --include-annotations true -o $*.owl

This will add all the standard metadata (license, title, description, creators, issue trackers etc) into the ontology.

Advantages,

1. we can standardise ontology-level metadata rapidly (which properties we use for which metadata item)
2. propagate stuff (new/updated metadata) quickly (for example, we add a new metadata item for grant references)
3. generally ensure that ontology metadata is rich (all the stuff OBO foundry deems important from the metadata records) and
4. in-sync with OBO foundry at all times (no diverging titles or descriptions).

Try this command to see how it would look like:

robot merge -I http://purl.obolibrary.org/obo/ddpheno.owl -I https://raw.githubusercontent.com/information-artifact-ontology/ontology-metadata/metadatata/src/metadata/obo/ddpheno.owl --include-annotations true -o ddpheno-with-metadata.owl

Sounds a bit crazy but I kinda like it.

WARNING:

The metadata properties I used for https://raw.githubusercontent.com/information-artifact-ontology/ontology-metadata/metadatata/src/metadata/obo/ddpheno.owl are not thought through - this is just an example. We will decide how to map OBO metadata to annotation properties.

[graybeal]

Another advantage:
5. downstream users and apps (BioPortal, AgroPortal, linked data users, semantic researchers) are encouraged to see, compute with, bring forward in their product, and standardize on OBO metadata properties.

[jamesaoverton]

The status quo is that the OBO registry contains data about each ontology project, which gets updated periodically, and the OWL files contain data about each release, each of which is frozen in time. There is also data associated with each term, since terms get imported into other projects. Here's an old document trying to make those distinctions clear.

There is plenty of room for improvement, but I don't think that just collapsing the distinction between project and release data will make things better. The current Dashboard checks are meant to help keep the relevant data in sync.

[matentzn]

Not sure I understand, I am not suggesting collapsing release metadata (versionIRI, versionInfo, etc) with project metadata (title, description, license), nor does this have anything to do with term-level metadata - I am just suggesting that since we are making an effort to standardise project-level metadata on the OBO foundry level, we may as well just merge that directly into our releases..

[jamesaoverton]

@matentzn Sorry, I misunderstood. I'm fine with better tooling for syncing selected values from the registry to a project.

[matentzn]

The next step would be to map all that metadata to standard vocabs. I used @jonquet's MOD 2 as a reference, which is also used afaik by Agro and BioPortal (@graybeal):

OBO metadata	Mapped Property	Example
title	dcterms:title	Mondo Disease Ontology
description	dcterms:description	A semi-automatically constructed ontology that merges in multiple disease resources to yield a coherent merged ontology.
homepage	foaf:homepage	https://monarch-initiative.github.io/mondo
license.url	dcterms:license	http://creativecommons.org/licenses/by/4.0/
contact	dcat:contactPoint	"Nicole Vasilevsky ([email protected]), GitHub: @nicolevasilevsky, ORCiD: 0000-0001-5208-3432"
depicted_by	foaf:logo	https://raw.githubusercontent.com/monarch-initiative/mondo/master/docs/images/mondo_logo_black-stacked-small.png
tracker	doap:bug-database	https://github.com/monarch-initiative/mondo/issues
mailing_list	doap:mailing-list	https://groups.google.com/group/mondo-users
publications.id	dcterms:bibliographicCitation	https://www.medrxiv.org/content/10.1101/2022.04.13.22273750
repository	doap:repository	https://github.com/monarch-initiative/mondo
preferredPrefix	idot:preferredPrefix	MONDO

I only mapped the metadata elements that I think could be useful for consumers of the ontology, maybe I included too much, maybe I missed something.

[jamesaoverton]

The OBO registry is converted to RDF (see registry/) and mappings for all (?) of these are defined in registry/context.jsonld.

[matentzn]

Excellent, I didnt know that :) Thanks!