Actions

Ontolog Forum

Session Ubergraph
Duration 1 hour
Date/Time 22 Feb 2023 17:00 GMT
9:00am PST/12:00pm EST
5:00pm GMT/6:00pm CST
Convener Gary Berg-Cross

Ontology Summit 2023 Ubergraph

Helping scientific researchers make better use of ontologies

Agenda

  • Ubergraph: A SPARQL endpoint with many OBO ontologies loaded and pre-reasoned with simple triples materialized
  • Presenter: Jim Balhoff
  • Slides
  • Video Recording

Conference Call Information

  • Date: Wednesday, 22 Feb 2023
  • Start Time: 9:00am PST / 12:00pm EST / 6:00pm CST / 5:00pm GMT / 1700 UTC
  • Expected Call Duration: 1 hour
  • Video Conference URL
    • Conference ID: 837 8041 8377
    • Passcode: 323309
  • Chat Room

The unabbreviated URLs are:

Attendees

Discussion

[12:03] RaviSharma: Hello welcome everyone,

[12:12] RaviSharma: what is the significance of word Axioms in other ontologies vs just used in other ontologies?

[12:17] Sue: @Ravi the axioms have formal logic in which the terms from other ontologies are used to define the term in the focus ontology. It is one way to use terms from one ontology in another ontology. Another example of use would be as cross-references.

[12:17] Mike Bennett: Why equivalent class: Usually one would make the class a subclass both of it's usual (taxonomic) parent and of the restriction. the intersection is implicit

[12:19] RaviSharma: Jim you showed both ways, which one has less errors? Say in constructing KG.

[12:20] James Overton: @Mike: We have a standard operation `relax` that converts equivalence axioms to simpler subclass axioms <http://robot.obolibrary.org/relax>. But I think the point stands that querying OWL in SPARQL is hard, and Jim is building toward presenting a more convenient solution.

[12:23] Tiago Lubiana: Is there a way to differentiate stated/inferred relations in Ubergraph? E.g. one might be interested in retrieving only direct superclasses of a term, but all superclasses in the tree are inferred

[12:24] James Overton: relation-graph was added to ROBOT 1.9.2 very recently, as part of the <http://robot.obolibrary.org/extract#subset> command.

[12:24] Chris Mungall: @Tiago - these are distinguishable the named graph the triple belongs to

[12:28] Alexander Diehl: Is the query language for Ubergraph SPARQL or just SPARQL-like?

[12:28] Mike Bennett: @James thanks - makes sense.

[12:29] James Overton: @Alexander: Jim is showing actual SPARQL. He's using a trick to define prefixes for commonly used terms such as `cell` and `has_output`, to make the SPARQL more readable. [12:29] James Overton: @Alexander: Jim is showing actual SPARQL. He's using a trick to define prefixes for commonly used terms such as `cell` and `has_output`, to make the SPARQL more readable.

[12:31] RaviSharma: Jim do these include GO relations mapping all gene pairs?

[12:32] Chris Mungall: @Ravi - no, this is just the ontology

[12:33] Asiyah: Is there a list of the 53 ontologies included in the Ubergraph?

[12:39] Robert Rovetto: what is the visualization tool used for those graph images?

[12:40] Chris Mungall: https://github.com/INCATools/obographviz

[12:42] Asiyah Lin - NIAID: Is there a list of 53 ontologies included in the Ubergraph?

[12:42] Nico Matentzoglu: This query gives you all the phenotypic abnormalities linked to the urine sodium level: https://api.triplydb.com/s/J1QkfjIAw

[12:43] James Overton: The README has a list of the included ontologies: https://github.com/INCATools/ubergraph

[12:43] Sierra Moxon: I’ve used this file to determine which ontologies are in Ubergraph https://github.com/INCATools/ubergraph/blob/master/ontologies.txt

[12:43] Nico Matentzoglu: Or you check Ubergraph: https://api.triplydb.com/s/0TLkJlDcA

[12:43] James Overton: Sierra's answer seems like a more complete list. Thanks!

[12:43] Asiyah Lin - NIAID: Thank you both, James and Sierra! and Nico!

[12:44] RaviSharma: Jim have you checked the lost relations namely other than those that execute SPARQL and those that go in the error file?

[12:54] RaviSharma: Nico should we not be concentrating on dealing with harmonizing and aligning in terms of levels of depth to achieve better semantic interoperation? Namely by domain overlap handling?

Reply from Gary Berg-Cross: Current automation of alignment and harmonization is limited but useful. Some needs to be done manually but in that process we may come to understand what semantic depth means and automate a portion of it.

[13:00] Asiyah: Is there a way to crowd source people's mappings happened everywhere?

[13:01] James Overton: @Asiyah: Yes, SSSOM is part of Mapping Commons: https://github.com/mapping-commons

[13:01] Sierra Moxon: I'm curious if the "confidence score" has a restricted range?

[13:03] Nico: yes, its basically %

[13:05] RaviSharma: I asked about a way of controlling the ambiguities as we do while developing software, to which Nico replied that one could filter or selectively map terms from e,g, SnoMed so that two way connections are somewhat possible. The presentations showed to us the improving maturity in ontology integration, KGs and SPARQL use. Thanks to speakers.

[13:11] Raymond Lee: Thanks!

Resources

Previous Meetings

 Session
ConferenceCall 2023 02 15OBO Dashboards
ConferenceCall 2023 02 08ODK
ConferenceCall 2023 02 01ROBOT
... further results

Next Meetings

 Session
ConferenceCall 2023 03 01Panel
ConferenceCall 2023 03 08Modules and Patterns
ConferenceCall 2023 03 15Design Patterns
... further results