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] Ontology Metrics in BioPortal

Jennifer Leigh Vendetti vendetti at
Mon Mar 25 17:58:07 PDT 2019

Hi Laurel,

On Mar 13, 2019, at 11:41 AM, Laurel Cooper <cooperl at<mailto:cooperl at>> wrote:

I have a question about how the Ontology Metrics are generated in BioPortal.

The metrics for the Plant Ontology ( are confusing, as it says there are 2002 classes and 214 with no definition.  The latest release of the PO has 1788 classes and they all have definitions.  So this is incorrect.

With regard to metrics calculation, we follow the same paradigm as the Protege ontology editor [1]. We consider the ontology definition as the set of axioms that are declared in the source ontology, as well as those declared in imported ontologies. Opening po.obo in Protege shows the same metrics numbers for classes (2002) and object properties (13):

[cid:5131FD87-9351-47BE-A63F-791750282532 at]

I wouldn’t classify this behavior as incorrect, but rather a design decision to show metrics for the source ontology + imports. It’s possible that other ontology editing and/or visualization tools use a different approach. If there’s an alternative display of metrics data that you like, feel free to send us screen shots. We could always enter a feature request in our tracker for additional types of metrics display.

I did some digging into this and determined where the count of 214 lacking definitions is coming from.  We have 203 alt ids from terms that were merged or replaced, and 11 terms being imported from NCBI.

1788 + 214= 2002, which is the total number of classes in Bioportal Metrics.

The term counting tool does not discriminate between the PO terms and the NCBI taxon terms which are imported into the PO.  And is counting the alt IDs as terms w/o definitions.

Is there a way to correct this?  Ideally the term counter would only count the number of classes in the main or active ontology and not the Alt iDs

We use the open source OWL API to load and parse ontologies [2]. The API contains an OBO format parser, specifically for handling OBO ontologies. It appears that when the parser encounters an alt_id tag in a term stanza, it creates a class to represent the value of the alt_id. These classes are marked as deprecated.

For example, in the following term stanza:

id: PO:0000002
name: anther wall
namespace: plant_anatomy
alt_id: PO:0006445
alt_id: PO:0006477

… the parser creates a class for “anther wall”, as well as 2 deprecated classes for "PO:0006445”, and "PO:0006477”. The code for calculating metrics in the OWL API counts all classes in the ontology signature, including those marked as deprecated.

The OWL API is external to Stanford and I wasn't involved in the development of the OBO format parser, so I can’t speak with any authority on the decision to create deprecated classes for alt_id values. I can venture a guess that it helps to facilitate certain types of functionality in ontology editing tools. For example, in Protege you have the ability to view deprecated classes in the class tree drawn with strikethroughs. In the case of OBO ontologies with alt_ids, you can find those deprecated classes in the tree, click on them, and view what term they were replaced by:

[cid:0394199B-4E71-4BB2-A4F4-B9BAB9E8EF62 at]

In order to provide any more detail here, I'd probably have to speak to the developers of the OBO format parser.

Kind regards,


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screenshot 2019-03-25 11.10.05.png
Type: image/png
Size: 77962 bytes
Desc: Screenshot 2019-03-25 11.10.05.png
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Screenshot 2019-03-25 17.41.09.png
Type: image/png
Size: 80283 bytes
Desc: Screenshot 2019-03-25 17.41.09.png
URL: <>

More information about the bioontology-support mailing list