Query Selector Duplicates

[sstemann]

In what drugs may treat: developmental and epileptic encephalopathy

The first choice runs MONDO: 0100062

• https://ui.ci.transltr.io/results?q=f005fd9f-fdc4-4768-9e51-4567bbff20c7
• https://arax.ncats.io/?r=f005fd9f-fdc4-4768-9e51-4567bbff20c7
  

The second choice runs DOID :

• https://ui.ci.transltr.io/results?q=d161651c-b7eb-44dc-9201-3fcafdf9afb9
• https://arax.ncats.io/?r=d161651c-b7eb-44dc-9201-3fcafdf9afb9

Similar issues with gene, what chemicals may downregulate echs1

echs1 (first choice) https://ui.ci.transltr.io/results?q=5a4c6e4c-291e-4a0f-a136-109c75a3e11f

• https://arax.ncats.io/?r=5a4c6e4c-291e-4a0f-a136-109c75a3e11f
• NCBIGene:140547

echs1 (second choice) https://ui.ci.transltr.io/results?q=84be79f3-1c73-4059-bd2b-01f5c8749684

• https://arax.ncats.io/?r=84be79f3-1c73-4059-bd2b-01f5c8749684
• NCBIGene: 93747 mouse

echs1 (third choice) https://ui.ci.transltr.io/results?q=497f3c8a-1887-460b-8c9b-56bf2dedd23f

• pending so saving this before i lose anything
• NCBIGene: https://pubchem.ncbi.nlm.nih.gov/gene/8624884

echs1 (fourth choice) https://ui.ci.transltr.io/results?q=d0a74a5a-ce74-494e-8a91-867ff379efae

• https://arax.ncats.io/?r=d0a74a5a-ce74-494e-8a91-867ff379efae
• NCBIGene: 1892 human

[sierra-moxon]

potentially related issues:

• the autocomplete UI should show users if they are matched on a synonym and/or display some indication of why the same string shows up many times (in the case of a gene with the same symbol in many species) Show match text, or show designation of matching on a synonyms in addition to the "real" disease name in the UI autocomplete #44
• when two identical strings are found in the autocomplete, they can sometimes result in different search results (because under the hood, they are different identifiers) Semicircular Canal Dehiscence Syndrome returns two identical results in the autocomplete, but different results on search #51

[gaurav]

I'm planning to generate some kind of report of identically named concepts from the Name Resolver for issue TranslatorSRI/NameResolution#40

[sierra-moxon]

@gaurav - ok to assign this to you for a bit while you generate the report? That might help us figure out the problem or how widespread the problem could be?

[gaurav]

That is A-okay!

[sstemann]

UI team is also working on fixes for this

[sierra-moxon]

[gaurav]

Andy:

• Turning off everything that isn't MONDO helps, because we would expect a single ontology have consistent names.
• Adding in HP might cause this problem to return.
• Therefore, identifying cliques with identical preferred names isn't a high priority for us right now.

[Genomewide]

Screen shot showing the fix for the above description by @gaurav

this is currently in CI.

[sstemann]

the disease and chemical query inputs are improved, i'm not sure about the "What genes are up/down regulated by: [chemical]"? search.

Looks like one is HMDB:HMDB0259186 (https://ui.test.transltr.io/results?l=Triclopyr&i=HMDB:HMDB0259186&t=3&q=938bddae-c95f-4a88-adf3-3712a5a3674e) and the other is PUBCHEM.COMPOUND:41428 (https://ui.test.transltr.io/results?l=Triclopyr&i=PUBCHEM.COMPOUND:41428&t=3&q=b5a43e9a-6e26-4c22-b7ee-231df4c1e645)

From node norm it looks like HMDB:HMDB0259186 isn't matching to anything. Can we tell if this is a systematic issue with HMDB CURIES? Does HMDB have chemicals that other sources don't that need to be included in MVP2?

{
"HMDB:HMDB0259186": {
"id": {
"identifier": "HMDB:HMDB0259186",
"label": "TRICLOPYR"
},
"equivalent_identifiers": [
{
"identifier": "HMDB:HMDB0259186",
"label": "TRICLOPYR"
}
],
"type": [
"biolink:SmallMolecule",
"biolink:MolecularEntity",
"biolink:ChemicalEntity",
"biolink:NamedThing",
"biolink:Entity",
"biolink:PhysicalEssence",
"biolink:ChemicalOrDrugOrTreatment",
"biolink:ChemicalEntityOrGeneOrGeneProduct",
"biolink:ChemicalEntityOrProteinOrPolypeptide",
"biolink:PhysicalEssenceOrOccurrent"
]
}
}

[sstemann]

@dnsmith124 i'm still seeing this use case in CI - should this go back to name resolver?

[dnsmith124]

I think so @sstemann

Paging @gaurav! See above

[gaurav]

@cbizon is working on Drug Conflation that should help with this. We're hoping to be able to report on this late next week. Thanks for reporting this!

I think it would also be helpful if we could add links to let people look up these concepts in an ontology: what do you think, @dnsmith124?

[dnsmith124]

@gaurav I've actually already implemented linkouts for NCBIGENE, PUBCHEM, HMDB and CHEMBL, should be live on CI by next week!

[sstemann]

@gaurav i'm back on the Disease selection, when i type "develop" the first responses i get are "Developmental and Epileptic encephalopathy, 40" , but when i type out to "development" the list updates, i'm also wondering about when a disease has variations that are numbered, how the numbering is ordered. this might be more related to #85

[sstemann]

for drugs the link out seems to be more spotty

[sstemann]

seems improved in prod 9/6

[Genomewide]

@dnsmith124 I think we can work on the ordering of the autocomplete a bit. One tune-up as suggested above is the numbered subclasses come back in order. Also, any exact match comes to the top. I am thinking of a full exact match to the string. So, 'Alz' would exactly match all the numbered Alzheimer diseases but 'Alzheimer diseases' may only match one?

Related is also having the human gene show up at the top of the list.