Remove revision links from CW-WikiPathways linkset

[stain]

In the linkset ConceptWiki-WikiPathways we find two kind of links, unversioned and versioned:

<http://www.conceptwiki.org/concept/39a2bdd4-4e0e-4c60-80cb-08cb217a1708> skos:exactMatch 
     <http://www.wikipathways.org/index.php/Pathway:WP1022> .
<http://www.conceptwiki.org/concept/39a2bdd4-4e0e-4c60-80cb-08cb217a1708> skos:exactMatch 
     <http://www.wikipathways.org/index.php/Pathway:WP1022_r67325> .

..This is great provenance as it means that cw:39a2bdd4-4e0e-4c60-80cb-08cb217a1708 matched against WP1022 in revision r67325, but the revisioned link is not so useful in IMS results as it does not work with the /pathway API calls as IMS refuses to do WikiPathways to WikiPathways mappings.

Note that the API calls however return the revisioned wikipathway URIs as their subject, and use dc:identifier to link to the unversioned identifier. (This could probably be using pav:hasVersion instead).

For example, https://beta.openphacts.org/1.5/pathways/byCompound?uri=http%3A%2F%2Fwww.conceptwiki.org%2Fconcept%2F83931753-9e3f-4e90-b104-e3bcd0b4d833&app_id=161aeb7d&app_key=bbcba81896020f0b95e3dd35b55e3345&_format=ttl
returns

<http://rdf.wikipathways.org/Pathway/WP1252_r69118> void:inDataset <http://www.wikipathways.org> ;
                                                    foaf:page <http://www.wikipathways.org/instance/WP1252_r69118> ;
                                                    dc:identifier <http://identifiers.org/wikipathways/WP1252> ;
                                                    dc:title "Amino acid conjugation of benzoic acid"@en ;
                                                    ns0:organism <http://purl.obolibrary.org/obo/NCBITaxon_10090> ;
                                                    dct:hasPart <http://identifiers.org/hmdb/HMDB00123> .

The WP revision loaded in the RDF cache might not match the one in the ConceptWiki-WikiPathways - we know this will be outdated for many pathways as soon as we update the WP linkset in the RDF cache.

This bug suggests to simply remove the WP[0-9]+_r[0-9]+ links and only keep the unversioned ones in the CW-WikiPathways mapping.

It might longer-term make sense to have a pav:hasVersion linkset generated from the WikiPathways - this could then still be used with IMS and would match the API URIs.

[stain]

Perhaps @egonw and @AlasdairGray have views on this..? I remember we discussed this before. :)

[egonw]

@Chris-Evelo, what do you remember about this? You once said something about CW indexing not just pathway titles but also the ontology terms linked to pathways... do you know how this exactly got implemented in (and by whom) CW? Has this been written up at some point?

[danidi]

Christine and me tried to find Pathway Ontology terms for each of the Pathways that were available in the curated subset at that time. Where not available, we tried to add GO terms instead. See the list here https://docs.google.com/spreadsheets/d/1c0sE__bJQDHO6lNs1R5OBi4vX-GMEebpAkN0H6HDf7o/edit?pli=1#gid=0. These were then added directly to the Wikipathways pathways.
I'm not sure about how exactly the data was added to Conceptwiki, but there are two different sources available for pathways: Wikipathways and pathway ontology. Compare http://www.conceptwiki.org/concept/index/dfccdb4a-1472-4761-bed0-3b3ea3849b03 and http://www.conceptwiki.org/concept/index/0e1dcdb4-b345-4fb6-99a0-040ec64b827d. For the Wikipathways links, I think the reason to keep the version number was that the name in Wikipathways could change anytime. As Conceptwiki is no longer updates, this leaves us with older revision numbers here, but they also map to the current version via the identifiers.org linkset.

[Chris-Evelo]

That is what I remember too, except I think Christine actually suggested the pathway ontology to be extended with the extra terms and that that was indeed done. As far as I know the URIs for the pathway ontology classes are in the WikiPathways RDF. So we should only need the ConceotWiki terms for the labels. I do think we should be able to do that without a version number.