Correction of logical issues related to negation in CL

[addiehl]

James Overton @jamesaoverton discovered a number of logical errors in the CL related to conflicting has_plasma_membrane_part and lack_plasma_membrane_part relations in terms in class-subclass relationships.

From James, his method:

1. all object properties removed except for:
   • has plasma membrane part
   • lacks_plasma_membrane_part
   • has_low_plasma_membrane_amount
   • has_high_plasma_membrane_amount
2. has_low_plasma_membrane_amount and has_high_plasma_membrane_amount mapped to just 'has plasma membrane part'
3. all "lacks_plasma_membrane_part some X" axioms replaced by "'has plasma membrane part' only (not X)"

HermiT reasons over this in a few seconds. After reasoning in Protege there are 14 classes under owl:Nothing. It would be better not to remove all the other object properties (step 1), but we need HermiT to handle the negation, and HermiT was failing for me on the full Cell Ontology. Instead of debugging CL right now, I focused on just the membrane parts.

My analysis: There are 14 classes, but actually fewer errors.

For instance a single error involving 'CD27 molecule' accounts for these three classes being in OWL:Nothing double negative memory B cell IgG-negative double negative memory B cell IgG-positive double negative memory B cell

Also, a single error involving 'CX3C chemokine receptor 1' accounts for these five classes being in OWL:Nothing Gr1-high classical monocyte MHC-II-negative classical monocyte lymphoid MHC-II-negative classical monocyte lymphoid MHC-II-negative non-classical monocyte (this class suffers from three errors, actually, though only two are caught by the reasoner) [appears below as well] MHC-II-positive classical monocyte

Also, a single error involving 'CX3C chemokine receptor type 2' accounts for these five classes being in OWL:Nothing Gr1-low non-classical monocyte MHC-II-high non-classical monocyte MHC-II-low non-classical monocyte MHC-II-negative non-classical monocyte lymphoid MHC-II-negative non-classical monocyte (this class suffers from three errors, actually, though only two are caught by the reasoner) [appears above as well]

There are two other classes with similar errors as well Kit-positive, integrin beta7-high basophil mast progenitor cell langerin-positive dermal dendritic cell

So actually there are only 5 errors (+1 error in the construction of the logical definition for 'lymphoid MHC-II-negative non-classical monocyte' in specifying its parent as 'MHC-II-negative classical monocyte' and asserting its parent correctly as 'MHC-II-negative non-classical monocyte' in its subclass relationship).

I propose to fix these errors in the short term by removing a single logical assertion from the equivalence axiom of a superclass that is in conflict with a more specific opposite assertion in a subclass.

For ‘langerin-positive dermal dendritic cell’ I will remove the assertion from the class itself, since it doesn't make sense to remove CD11b from ‘CD11b-positive dendritic cell’.

[addiehl]

Removed ('has plasma membrane part' some 'CD27 molecule') from 'memory B cell' equivalence class.

Removed (has_low_plasma_membrane_amount some 'CX3C chemokine receptor 1') from 'classical monocyte' equivalence class.

Removed (lacks_plasma_membrane_part some 'C-C chemokine receptor type 2') from 'non-classical monocyte' equivalence class.

Error in the construction of the logical definition for 'lymphoid MHC-II-negative non-classical monocyte' corrected.

Removed ('has plasma membrane part' some 'mast/stem cell growth factor receptor') from 'basophil mast progenitor cell' equivalence class.

Removed (lacks_plasma_membrane_part some 'integrin alpha-M') from 'langerin-positive dermal dendritic cell' equivalence class. Adjusted the textual definition as well to reflect this change.