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] help with GAMUTS ontology

Jennifer Leigh Vendetti vendetti at stanford.edu
Mon Mar 5 14:25:33 PST 2018


Hi Ross,


On Mar 5, 2018, at 8:19 AM, Filice, Ross W <Ross.W.Filice at gunet.georgetown.edu<mailto:Ross.W.Filice at gunet.georgetown.edu>> wrote:

Hi – NCBO could you please help? I don’t think we received a response (though I may have missed it).
Thanks much,
Ross

From: Filice, Ross W
Sent: Thursday, February 22, 2018 2:47 PM
To: support at bioontology.org<mailto:support at bioontology.org>
Cc: 'Kahn, Charles E'
Subject: help with GAMUTS ontology

Hi – we are having a little trouble with our new, now public, ontology GAMUTS. We’ve uploaded the ontology in OWL format including mappings to the Orphanet Rare Disease Ontology (ORDO) and it successfully parses, we can annotate text using the web GUI successfully, but we are having several problems:

1) When we try to just browse the ontology, or if we annotate something and try to click through to the matching entity, we get error messages. If we browse other ontologies everything seems to work ok – and I am fairly sure this used to work ok. This error occurs whether I’m logged in as the owner or anonymously. If I use the API I can find all of the annotation data, relationships, ORDO mappings, etc – just seems to be in the web browser version.


What is the specific error message you see in the BioPortal UI?

One thing I noticed about the GAMUTS ontology is that it has a relatively flat hierarchy with a large number of classes at the root level (15,230). The Classes page (http://bioportal.bioontology.org/ontologies/GAMUTS?p=classes) takes roughly 8 or 9 seconds to load because BioPortal is fetching information for ~15K root classes in order to construct / display the class tree. Most of the ontologies in BioPortal don’t have such a large number of classes at the root level, and it’s a known issue that the Classes page wasn’t designed to be optimized for this case. This was recently discussed on the support list for another ontology (http://ncbo-support.2288202.n4.nabble.com/bioontology-support-BioPortal-Feedback-from-Andrey-Fedorov-td4656106.html#a4656113).

One option for the few ontologies in BIoPortal like this, is to have us manually flag the ontology in our system as “flat”. What this means is that end users are presented with a search box for terms on the Classes page instead of a tree structure. It also speeds operations like clicking through to a term from the Annotations page. For an example of what I mean, you can look at the RXNORM ontology (http://bioportal.bioontology.org/ontologies/RXNORM?p=classes).




2) Under “mappings” for GAMUTS it lists three mappings that we did not intend to include – these look like they were auto-populated using case-insensitive string matching. Can we flag something to exclude these mappings? They often are nonsensical.


Which page are you referring to? Mapping counts for the whole ontology like below:


[cid:03E0C7DC-32D8-4BCD-BA9B-E6494E4A2270 at stanford.edu]


Or, mappings for a particular class?




3) We also included mappings to ORDO entities using the owl#sameAs property – again, this seems to show up properly when using the API but is not reflected in the Mappings tab. Should we be specifying these mappings in a different way?


BioPortal doesn’t use the contents of an ontology source file to generate mappings (i.e., owl:SameAs, skos:ExactMatch, etc.). Instead, mappings are automatically generated using the following techniques:

1). SameURI - mappings are generated for classes with identical URIs.
2). LOOM - mappings are generated for classes where the preferred names or synonyms are similar (delimiters removed). More information about the LOOM algorithm is available in this article: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2815474/.
3). CUI - mappings are generated for classes with identical CUIs (a unique identifier used by the UMLS).

End users also have the option of generating mappings via the REST endpoint (http://data.bioontology.org/documentation#Mapping), which are persisted in our backend.

Could you point us to an example where you’re able to retrieve a mapping from the REST endpoint that isn’t displayed in the user interface?

Kind regards,
Jennifer

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20180305/4bc03bd3/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screenshot 2018-03-05 14.10.40.png
Type: image/png
Size: 63644 bytes
Desc: Screenshot 2018-03-05 14.10.40.png
URL: <http://mailman.stanford.edu/pipermail/bioontology-support/attachments/20180305/4bc03bd3/attachment-0001.png>


More information about the bioontology-support mailing list