Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Classification lost in -simple version when named EC pairs present in the master file #54

Open
ShahimEssaid opened this issue Feb 2, 2015 · 3 comments

Comments

@ShahimEssaid
Copy link

From dosu...@gmail.com on March 04, 2013 03:31:22

= What steps will reproduce the problem? =

ontology-release-runner --reasoner hermit flybase_controlled_vocabulary_xp.obo --allow-equivalent-pairs --no-subsets --simple --allow-overwrite --outdir oort

You can grab the master file from here:
http://kyala.gen.cam.ac.uk/svn/ontologies/ontologies/trunk/flybase_controlled_vocabulary_xp.obo

It contains named pairs of equivalent classes from ChEBI AND FBcv.

= What is the expected output? What do you see instead? =

I expected the -simple versions to have FBcv terms with classification to root, including some inference from ChEBI. Instead, all classification of these terms is lost, including asserted classification from FBcv.


= What version of the product are you using? On what operating system? =

Ubuntu 2.6.32-45-server
OWLTOOLS SVN Revision: 1132

Original issue: http://code.google.com/p/owltools/issues/detail?id=54

@ShahimEssaid
Copy link
Author

From HDie...@lbl.gov on March 04, 2013 13:55:13

Thank you for providing the concise example. I can explain why OORT produces the result.

During the reasoning step, we do not create (redundant) axioms with the FBcv terms and chemicals. For the reasoner they are equivalent and thus OORT does not add any new axioms. For example 'carbamate' (CHEBI:13941 and FBcv:0000538), it is a subclass of 'amino-acid anion' (CHEBI:37022) and chemical (FBcv:0000525). As the ChEBI term is more specific it does not add any FBcv-related axioms.
Now during the simplification step, we more or less remove all axioms with a non FBcv id. There is no reasoning step involved. So we do not add the axiom 'carbamate' is_a chemical (FBcv:0000525), which is true in the original.

The solution is not quite clear yet as this seems to be a new feature. Do we need to rework the mechanism to create a simple ontology, or do we do a separate rewriting step of axioms?

Labels: -Priority-Medium Priority-High

@ShahimEssaid
Copy link
Author

From dosu...@gmail.com on March 05, 2013 13:51:47

Thanks for the quick response!

I understand why we're not getting any extra inference directly from the ChEBI classification. But we seem to be losing all classification for the FBcv terms.

  • e.g. you pointed out that "'carbamate' (CHEBI:13941 and FBcv:0000538)is a subclass of 'amino-acid anion' (CHEBI:37022) and chemical (FBcv:0000525)." But it ends up as a root class in the -simple file. I'll settle for a fix that keeps the FBcv classification for now.

@ShahimEssaid
Copy link
Author

From HDie...@lbl.gov on March 05, 2013 14:52:52

The problem is that the ChEBI hierarchy is more specific than the FBcv one. During the reasoning step all direct relations to 'chemical'l (FBcv:0000525) are considered to be redundant. Even if these axioms were in the ontology, they would be removed.
We have no mechanism (at this point) to signal that we want these relations to be asserted.
I'll have to talk to Chris to decide how and where to put code for this.

As a potential work around, I have added the option '--skip-remove-redundant' to skip the remove-redundant step in OORT. See revision r1167 .

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

1 participant