Validate prefix maps against Semantic Farm

[cthoyt]

Motivation

LinkML doesn't support modelers in defining meaningful prefix maps.

Therefore, we have to look to external tooling to do so. I wrote about tooling in the Semantic Farm that supports modelers to define meaningful prefix maps: https://cthoyt.com/2026/01/06/bioregistry-linkml-validation.html

Summary of Changes

This PR adds the Semantic Farm LinkML prefix map validation / feedback workflow into the tests in this repository. It immediately surfaces several issues, some of which could be added as new prefixes in the Semantic Farm, and some of which might need to be reconsidered.

This PR also bumps the minimum python version from 3.9 to 3.10 - 3.9 is past end of life and software like Bioregistry (and soon, LinkML) don't support it.

Errors in DCAT-AP LinkML

$ bioregistry validate linkml --use-preferred https://github.com/nfdi-de/dcat-ap-plus/blob/main/src/dcat_ap_plus/schema/dcat_ap_linkml.yaml

gives:

prefix	uri_prefix	issue	solution
dcatap	http://data.europa.eu/r5r/	unknown CURIE prefix	added to Semantic Farm in biopragmatics/bioregistry#1869
vl	https://purl.eu/ns/shacl#	unknown CURIE prefix	replace with W3 definition of SHACL, see #76
iana	https://www.iana.org/assignments/	unknown CURIE prefix	Consider switching to the more specific CURIE/URI prefix pair iana.mediatype: https://www.iana.org/assignments/media-types/
epos	https://www.epos-eu.org/epos-dcat-ap#	unknown CURIE prefix	added to Semantic Farm in biopragmatics/bioregistry#1869
dcatap_linkml	https://w3id.org/nfdi-de/dcat-ap-linkml/	unknown CURIE prefix

Errors in DCAT-AP Plus

$ bioregistry validate linkml --use-preferred https://github.com/nfdi-de/dcat-ap-plus/blob/main/src/dcat_ap_plus/schema/dcat_ap_plus.yaml

gives:

prefix	uri_prefix	issue	solution
dcatap	http://data.europa.eu/r5r/	unknown CURIE prefix	added to Semantic Farm in biopragmatics/bioregistry#1869
vl	https://purl.eu/ns/shacl#	unknown CURIE prefix	this should be the W3 definition for SHACL, see #76
iana	https://www.iana.org/assignments/	unknown CURIE prefix	Consider switching to the more specific CURIE/URI prefix pair iana.mediatype: https://www.iana.org/assignments/media-types/
epos	https://www.epos-eu.org/epos-dcat-ap#	unknown CURIE prefix	added to Semantic Farm in biopragmatics/bioregistry#1869
dcatapplus	https://w3id.org/nfdi-de/dcat-ap-plus/	unknown CURIE prefix
SOSA	http://www.w3.org/ns/sosa/	non-standard CURIE prefix	Switch to preferred prefix: sosa
AFE	http://purl.allotrope.org/ontologies/equipment#AFE_	unknown CURIE prefix	Added allotrope.equipment in biopragmatics/bioregistry#1919, switch to using that instead of AFE
ex	http://example.org/	unknown CURIE prefix	use concrete examples, not made up ones

See Also

P.S. I made this PR biopragmatics/bioregistry#1866 to expose this functionality more nicely :)

[dalito]

I like the idea and allowed CI to run for you as new contributor.

...as the CI runs show such checks should be in place.

[dalito]

Before we merge this, we should fix #74 to keep CI green.

[HendrikBorgelt]

We also need to take a look at why we have some prefixes in our LinkML schema. epos and vl don't seem to be used, and I think @StroemPhi is still trying to drop AFE references. The prefix ex can be deleted, since we don't target this namespace, nor do we need it for your example data.

For the Prefix dcatap_linkml, I would like to get feedback from @dalito, @markdoerr, and @StroemPhi if we should even implement this. I opened an Issue at biopragmatics/bioregistry#1868, which we could use as a draft.

[StroemPhi]

As I just wrote to @HendrikBorgelt via RC, I'm for registering all our namespaces also in the Semantic Farm to make use of @cthoyt's automated checks. And I'm for snake_casing.
The Allotrope prefixes need to be present as long as we use AFO terms for the mapping. We can only drop those once we have alternatives in CHMO and similar ones.

[cthoyt]

regarding AFO, most of the subspaces were already in Semantic Farm. I added one for allotrope.equipment in biopragmatics/bioregistry#1919