search

geneontology / termgenie

Web-application for pattern-based ontology class generation
7 stars 6 forks source link

Seeing inferred parents in mp.termgenie #59

Closed hdietze closed 9 years ago

hdietze commented 9 years ago

Original issue 59 created by hdietze on 2015-02-09T17:28:20.000Z:

When we create a new term we can't see the parents in the tg interface (in contrast to go)

I assume because we are in owl. Just seeing the owl axioms is better than nothing. But if there could be some textual output, e.g.

'Your term "venom gland development" would be placed under "exocrine gland development"'

hdietze commented 9 years ago

Comment #1 originally posted by hdietze on 2015-02-09T17:28:39.000Z:

<empty>

hdietze commented 9 years ago

Comment #2 originally posted by hdietze on 2015-02-09T17:50:25.000Z:

Yes, unfortunately that is a side effect of not asserting the inferences. The current implementation does not have the data structures (server and js-client wise) to show inferences separately. A rough time estimate to implement: Three days (half a day for the server, one day for the client, rest testing and debugging)

hdietze commented 9 years ago

Comment #3 originally posted by hdietze on 2015-02-09T23:28:59.000Z:

<empty>

hdietze commented 9 years ago

Comment #4 originally posted by hdietze on 2015-02-09T23:31:14.000Z:

After investigating the issue more, this turns out to be a weird issue. Only after removing full ChEBI from the loading the reasoning is working properly again. This means the abnormal X levels is also de-activated.

Keeping this issue open until we find a permanent solution.

hdietze commented 9 years ago

Comment #5 originally posted by hdietze on 2015-02-25T22:25:19.000Z:

The issue with the missing inferences is spreading. I can reproduce the missing super classes with ELK in Protege. Attached is the real ontology with two classes without inferences: 'greater spleen glucose level' and 'abnormal spleen glucose level'

hdietze commented 9 years ago

Comment #6 originally posted by hdietze on 2015-02-28T14:40:41.000Z:

If i change the axiom on 'abnormal spleen glucose level' to use the Pato 'quality' term instead of the BFO one, and add "and ('has modifier' some abnormal))" to the 'greater spleen glucose level' then I get the correct inferences. Note that I had to change the label of the the Pato 'quality' term in order to do this.

So, seems two things:

  1. need to ensure everything gets the ('has modifier' some abnormal) portions of the axiom
  2. It is hard to choose the correct 'quality' term in P4 because the PATO one shares a label with a BFO term (which was incorrectly used in this case).
hdietze commented 9 years ago

Comment #7 originally posted by hdietze on 2015-03-02T16:20:10.000Z:

TG should auto-add the 'abnormal' axiom for anything in HP or MP that is a child of abnormal phenotype

hdietze commented 9 years ago

Comment #8 originally posted by hdietze

Made the change in TermGenie. Inference tests are now working, once we always use the has modified' some abnormal. See revision r1213 for template change.