matentzn commented 2 years ago

Prefix

Orphanet

Name

Orphanet

Homepage

https://www.orpha.net

Description

Orphanet is a unique resource, gathering and improving knowledge on rare diseases so as to improve the diagnosis, care and treatment of patients with rare diseases.

Example Identifier

79154

Regular Expression Pattern

[A-Z0-9][0-9]+

Redundant Prefix in Regular Expression Pattern

No response

Provider Format URL

http://www.orpha.net/ORDO/Orphanet_

Contributor Name

Nico Matentzoglu

Contributor ORCiD

0000-0002-7356-1779

Additional Comments

There is already a lowercase orphanet namespace but it uses the wrong URL prefix as well as we prefer this to be upper case due to convention.

cthoyt commented 2 years ago

The Bioregistry already has two related prefixes, that are not the same:

http://bioregistry.io/registry/orphanet which is the main orphanet vocabulary
http://bioregistry.io/registry/orphanet.ordo which is the Orphanet Rare Disease Ontology

ordo is already tagged as a synonym of orphanet.ordo.

It looks like this request is mixing parts of each, using the prefix from the main vocabulary but the provider format URL for the ORDO vocabulary. I'm not really sure where the confusion comes from, but maybe it's a systematic mistake propagated from the Biolink Model's Biocontext (see the "Metaregistry" heading in http://bioregistry.io/registry/orphanet.ordo). I manually curated the mappings from that resource and noted that they were mix and matching before

matentzn commented 2 years ago

I dont so much care about the prefix here, but the URL prefix needs to be added somewhere so I can run conversions correctly!

cthoyt commented 2 years ago

I mentioned in that other issue (https://github.com/mapping-commons/sssom-py/issues/161) that I recently actually added this URI prefix in biopragmatics/bioregistry@21906c3. Is that sufficient?

matentzn commented 2 years ago

It is! Thats the important bit.

dhimmel commented 1 year ago

The Bioregistry already has two related prefixes, that are not the same

I'm confused about the difference between ORPHA and ORDO terms, so coming here for help. And possibly finding some disambiguating description that we can add to Bioregistry to help users.

On ORDO, there's a 2014 publication, only on ResearchGate, titled "ORDO: An Ontology Connecting Rare Disease, Epidemiology and Genetic Data". My understanding of this paper is that the authors took Orphanet and converted it to an OWL ontology. I don't find any discussion of whether the identifiers carry over or new identifiers are assigned. The https://github.com/Orphanet/Orpha2Ordo repo mentioned in the paper is gone.

The official page for ORDO appears to be at https://www.orphadata.com/ordo/. Any more insights into how Orphanet (ORPHA) terms are difference from ORDO terms? Does every ORDO term have a corresponding Orphanet term, but with different identifiers?

I arrived here when we noticed EFO has some cross-references to orphanet and some to orphanet.ordo post Bioregistry normalization.

matentzn commented 1 year ago

It is my understanding that the identifier space for the two is one and the same. I cant say for certain that there are not some ids in Orphanet that are not in ORDO, but the last five years I am working under the assumption that the semantic space is one and the same. I will check with the Monarch folks and see if I find out more.

cthoyt commented 1 year ago

Please take a look at the terms that are prefixed with a C - these appear to work in ordo but not in Orphanet (https://www.ebi.ac.uk/ols/ontologies/ordo/terms?iri=http://www.orpha.net/ORDO/Orphanet_C023)

cthoyt commented 1 year ago

There's definitely two semantic spaces here, with distinct patterns. The resources that can resolve one or both are hard to understand. This feels reminiscent of the omim vs omim.ps discussion

dhimmel commented 1 year ago

So orphanet.ordo:C023 for "age of onset" exists only in ODRO because it's a property key that can be applied to diseases in Orphanet, but is not a disease itself?

Identifiers do appear shared between both. For example, both the following will get to you to "Rare dyslipidemia":

https://bioregistry.io/orphanet:101953 redirects to https://www.orpha.net/consor/cgi-bin/OC_Exp.php?Lng=EN&Expert=101953
https://bioregistry.io/orphanet.ordo:101953 redirects to http://www.orpha.net/ORDO/Orphanet_101953 which redirects to OLS

Is the identifier set for orphanet.ordo a strict superset of the identifier set of orphanet? @Orphanet / Marc Hanauer any guidance you could provide here would be much appreciated.

With respect to bioregistry, we use it to normalize prefixes to enable more comprehensive mapping between resources. I see why having two separate Orphanet namespaces is technically correct here, but it would be good to have a recommendation to improve data linking. For example, perhaps all orphanet IDs should be converted to orphanet.ordo.

dhimmel commented 1 year ago

EFO case study

Looking at EFO OTAR Slim v3.57.0 (a version of EFO with a disease focus), we observe the following counts of cross-references (xrefs after bioregistry normalization):

7240 to orphanet
55 to orphanet.ordo

All of the 55 orphanet.ordo xrefs appear to be valid orphanet IDs (table below). So for our use case, I'm tempted to convert these all to orphanet prefixes. Note that some EFO terms imported from Orphanet like Orphanet:107 actually xref themselves via orphanet.ordo:98702.

Another option would be to convert all orphanet prefixes to orphanet.ordo. However, it looks like EFO uses Orphanet and not ORDO as the prefix for the terms that they include from ORDO/Orphanet.

Expand for ORDO table

biopragmatics / bioregistry

Add prefix Orphanet #187

Prefix

Name

Homepage

Description

Example Identifier

Regular Expression Pattern

Redundant Prefix in Regular Expression Pattern

Provider Format URL

Contributor Name

Contributor ORCiD

Additional Comments

EFO case study