Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[bioontology-support] [BioPortal] Feedback from Emma Vos

Samson Tu swt at stanford.edu
Thu Aug 13 18:41:46 PDT 2020


There may be some inferencing involved for EBI hierarchy to display blood as a a part of hematopoietic system. However, that’s not the issue.  The issue is that Bioportal hierarchy is a pure is-a hierarchy, so it doesn’t display the part-of relation as a link in the hierarchy. The EBI hierarchy is a mixed is-a/part-of hierarchy, so you see these part-of descendants.

Once Bioportal correctly processed the ontology, it shouldn’t have any missing terms. It’s just that you are not going to find ‘blood’ as a (part-of) child of "hematopoietic system” in Bioportal’s display hierarchy, but only an “is-a” child of "haemolymphatic fluid”.

With best regards,
Samson


On Aug 13, 2020, at 7:57 AM, John Graybeal <jgraybeal at stanford.edu<mailto:jgraybeal at stanford.edu>> wrote:

Hello Frans,

I think you have slightly misinterpreted the basis of the discussion, in that almost all of us are semantic engineers, not medically trained. So we were only using the medical terms in question as a means of understanding what these three tools are displaying to users, given the UBERON model that has been defined and their own semantic approaches.

I think the inferred classification that is displayed in Protege might be a bit more precisely described than using the blended terms in your final paragraph, but I'd leave it to someone more semantically expert than myself to get it exactly right. (Samson?)

But in regard to the medical understanding, we must send you to the builders of the UBERON ontology.

John



On Aug 13, 2020, at 12:15 AM, frans at semantoya.nl<mailto:frans at semantoya.nl> wrote:

Hi Jennifer and Eva et al,
With great interest I’ve been following this discussion from a professional point of interest.
The UBERON classification is based on anatomical properties.
In this respect the ‘hematopoietic system’ is most likely to be classified as part of ‘bone marrow’, because this is the place where most of the stem cells  resides.
The ‘blood’ class as we know it in daily practice is distinct from bone marrow because it flows around between in all compartments like ‘vascular system’ and ‘bone marrow’ system .
‘Venous blood’ could anatomically be considered as the ‘venous part of’ the ‘vascular system’ (UBERON: distinct class); because this system in general can be divided into arterial, venous and capillary blood compartment; On the other hand the properties of biomarkers in venous blood are distinct from those in capillary and arterial  blood samples and  for this reason we report the lab results in conjunction with the type of sample. So from a propetrt point of view venous blood is a subclass of blood as a substance.
So I conclude that the inferred classification that ‘venous blood’ isa/partof  ‘blood’ isa/partof ‘hematopoietic system’ might be misleading or even incorrect.  Or did I misinterpret the basics of this discussion?

Best regards
Frans van der Horst


Van: bioontology-support <bioontology-support-bounces at lists.stanford.edu<mailto:bioontology-support-bounces at lists.stanford.edu>> Namens Jennifer Leigh Vendetti
Verzonden: donderdag 13 augustus 2020 02:57
Aan: Samson Tu <swt at stanford.edu<mailto:swt at stanford.edu>>
CC: bioontology-support at lists.stanford.edu<mailto:bioontology-support at lists.stanford.edu>; Emma Vos <emma at thehyve.nl<mailto:emma at thehyve.nl>>
Onderwerp: Re: [bioontology-support] [BioPortal] Feedback from Emma Vos

Thanks for that correction Samson, and apologies to all that I navigated down the wrong path in my last response.

Jennifer




On Aug 12, 2020, at 4:17 PM, Samson Tu <swt at stanford.edu<mailto:swt at stanford.edu>> wrote:

Hi,


On Aug 12, 2020, at 10:29 AM, Jennifer Leigh Vendetti <vendetti at stanford.edu<mailto:vendetti at stanford.edu>> wrote:

With regard to the display of class hierarchies, there’s a distinct difference between an “asserted" hierarchy and an “inferred" hierarchy. BioPortal is displaying the asserted class hierarchy. If you open the uberon.owl file that OLS serves in the Protege ontology editor, you can see that the “venous blood” class is also show as a leaf term, i.e., in the position that it was manually defined in the asserted hierarchy (screen shot below). You can also see in the same screen shot that the “venous blood” class has a subClassOf axiom:

'part of' some 'hematopoietic system’

This means that "venous blood" will be shown as a subclass of "hematopoietic system” in any instance where an *inferred* hierarchy is displayed. In order to display an inferred hierarchy, you need to perform reasoning on an ontology. I don’t know the internal of the OLS software - however, I’m guessing that somewhere in their pipeline they perform a reasoning task that allows them to display terms in the context of an inferred hierarchy.

The hierarchy that BioPortal displays isn’t incomplete and/or incorrect. The differences you see between the two pieces of software have to do with asserted vs. inferred hierarchies.


Actually, I think the difference is not between asserted vs inferred hierarchies, but between the kinds links displayed in the Bioportal and EBI hierarchies.The Bioportal hierarchy is a strict is-a hierarchy, whereas the EBI hierarchy is a mixed is-a/part-of hierarchy. Note the P icon in the EBI hierarchy, indicating that blood is part of hematopoietic system, whereas venous blood is a kind of blood. Bioportal correctly shows that hematopoietic system has no is-a child.

<PastedGraphic-1.png>


With best regards,
Samson


_______________________________________________
bioontology-support mailing list
bioontology-support at lists.stanford.edu<mailto:bioontology-support at lists.stanford.edu>
https://mailman.stanford.edu/mailman/listinfo/bioontology-support

========================
John Graybeal
Technical Program Manager
Center for Expanded Data Annotation and Retrieval /+/ NCBO BioPortal
Stanford Center for Biomedical Informatics Research
650-736-1632  | ORCID  0000-0001-6875-5360




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20200814/57023a8d/attachment-0001.html>


More information about the bioontology-support mailing list