[EPIC] - HubMap requests 2511

[dosumis]

This ticket is intended to track a large number of requests arising from HubMap (Nov 2025).

Breakdown:

• "Missing" CL terms (@zhengj2007 is reviewing which of these are missing mappings)
  • Missing mappings - fix by updating tables. Priority 1. (Some of these might be borderline composable)
  • Missing PNS terms. Priority 1. Extend existing template + add Agentic review.
  • Review remaining for clear missing cell types (evidence that they are distinct - priority 1) vs cross-products (TBD)
    • @zhengj2007 to review to pull out sets of related terms to make tickets, making sure that the AS term and any available references are present - e.g. mouth table has a whole set of salivary gland terms to add; There are many possible new terms to add for spleen.
• Terms from Azimuth cross-walks
  • Aziumth v2 - these have references and so should be prioritised. Use agents + paper.
  • Pan tissue Azimuth - Hold for now. Lack of refs mean we will need a new strategy.

@zhengj2007 to post links/tables with examples

Folder for sharing (not open to world) https://drive.google.com/drive/u/4/folders/1Wm35YzgKHNSa1R-Q2UlmsaiC60RqhPNN

[dosumis]

@zhengj2007 - Would you be able to add these tables to the drive link? Ta

[dosumis]

Next steps:

• For each unmapped term, can we get the most specific AS term + references. These can be found in the ASCT+B tables that can be downloaded from links in unmapped term sheet, but much easier to get Bruce to generate reports with extended SPARQL queries.

[dosumis]

Example of defining terms using DeepSearch:

This table entry:

https://purl.humanatlas.io/asct-b/mouth |   | CT | https://purl.org/ccf/ASCTB-TEMP_demilune-acinar-cell | demilune acinar cell

Corresponds to this row in an the mouth ASCT+B table:

sublingual gland | UBERON:0001832 |   |   | demilune acinar cell

Using this information to generate a refernced definition via deepsearch.

https://www.perplexity.ai/search/system-role-you-are-an-expert-mgVrXEyqSRqP6OYtYuaVbQ#0

@zhengj2007 to ask Bruce to script extraction of most specific AS (Uberon) and all referenced for each entry.

Minimal aims for term addition:

Location, general classification (look at what axes are available) and references from Azimuth or ASCT+B table + minimal def.

We could potentially add more detailed definitions via Deepsearch later.

[dosumis]

Hi @nicolevasilevsky - we are reviewing potential new terms table from you. I think ideally we need source papers e.g. I can see a lot of tonsil term requests from Azimuth, Azimuth tonsil v2

https://azimuth.hubmapconsortium.org/references/#Human%20-%20Tonsil%20v2 has this reference paper.

Would you be able to add these?

There are some cases where we could do ROBOT templates (e.g. cycling terms), but in other cases I think we need to take a more standard approach, e.g. we need to be careful not to end up with duplicate terms for dark and light zone cells given different nomenclatures.

[bherr2]

I have created a new query that returns more information about each unmapped ontology term, including the table, row number, and references for each unmapped term.

You can download the csv here: https://apps.humanatlas.io/api/grlc/hra/unmapped-asctb-term-list-with-refs.csv

[nicolevasilevsky]

Hi @nicolevasilevsky - we are reviewing potential new terms table from you. I think ideally we need source papers e.g. I can see a lot of tonsil term requests from Azimuth, Azimuth tonsil v2

https://azimuth.hubmapconsortium.org/references/#Human%20-%20Tonsil%20v2 has this reference paper.

Would you be able to add these?

There are some cases where we could do ROBOT templates (e.g. cycling terms), but in other cases I think we need to take a more standard approach, e.g. we need to be careful not to end up with duplicate terms for dark and light zone cells given different nomenclatures.

Hi @dosumis - sure thing, I can add specific references for the NTRs I added in my spreadsheet. I'll work on this and aim to get it done in the next week.

Just to clarify, this is different than what @bherr2 provided in the table above, right?

[zhengj2007]

@nicolevasilevsky Yes. It is different from the list of terms that Bruce provided. I will email you later today for details of new terms you proposed.

[nicolevasilevsky]

Thanks @zhengj2007! Sounds great. 😄

[dosumis]

@bherr2 Many thanks for the file. Could you provide the SPARQL query and remind me of the endpoint? Ideally we need both references and most specific AS. Thanks!

[bherr2]

Query: https://github.com/hubmapconsortium/ccf-grlc/blob/main/hra/unmapped-asctb-term-list-with-refs.rq
Endpoint: https://apps.humanatlas.io/sparql
It queries against this named graph: https://purl.humanatlas.io/collection/hra

This query includes the most specific AS, CT and references. It also links back to the actual row from the ASCT+B table it came from.

Thanks,
Bruce

[nicolevasilevsky]

@zhengj2007 and all - I created a new tab in this spreadsheet that lists all of the potential new term requests from the crosswalk files. I added groupings in column K, where applicable (in some cases, more than one group was applicable, so I separated those by a comma.)

I included references for each of the potential new term requests in column K and applicable synonyms in column L.

This is a first pass and I welcome feedback and am happy to revise as needed. 😺