Zooma seems to ignore the preferred datasource parameter

[henrietteharmse]

On RT the following issue has been logged (https://helpdesk.ebi.ac.uk/Ticket/Display.html?id=371346)

I noticed that ZOOMA appears to ignore the preferred datasource ordering
specified in its parameters. This can be demonstrated using two
datasources, which are both produced by us (EVA) after processing each
OpenTargets batch:

• *eva-clinvar *contains mappings, curated by ourselves, to ontologies
  such as are EFO, HP, ORDO and MONDO;
• clinvar-xrefs contains mappings extracted directly from ClinVar
  data and maps terms to MedGen and OMIM ontologies.

It appears that no matter which datasource preference I specify,
clinvar-xrefs is always preferred by ZOOMA. This is unfortunate, because
for the purposes of OpenTargets processing we would like to prioritise
existing eva-clinvar mappings.

Here's an example API query using only eva-clinvar datasource. An HP
term with a HIGH confidence is returned from that datasource:

$ curl -s --globoff
"https://www.ebi.ac.uk/spot/zooma/v2/api/services/annotate?propertyValue=Reduced+antithrombin+III+activity*&filter=required:[eva-clinvar],preferred:[eva-clinvar]"
| jq '.[0]["semanticTags"][], .[0]["confidence"]'*
"http://purl.obolibrary.org/obo/HP_0001976"
"HIGH"

Here's the same query including eva-clinvar and clinvar-xrefs
datasets. Despite eva-clinvar being explicitly prioritised over
clinvar-xrefs, MedGen and OMIM terms from the latter are returned:

$ curl -s --globoff
"https://www.ebi.ac.uk/spot/zooma/v2/api/services/annotate?propertyValue=Reduced+antithrombin+III+activity*&filter=required:[eva-clinvar,clinvar-xrefs],preferred:[eva-clinvar,clinvar-xrefs]"
| jq '.[0]["semanticTags"][], .[0]["confidence"]'**
*"http://identifiers.org/medgen/C0272375"
"http://identifiers.org/omim/613118"
"HIGH"

[henrietteharmse]

Mix up on the user side. This not an issue.