Open cmungall opened 1 year ago
cmungall
commented
1 year ago Here is how to regenerate using semsql:
create view charge as select * from statements where predicate='obo:chebi/charge';
create view inconsistent_charge as select c1.subject, c2.subject from charge as c1, rdfs_subclass_of_statement as s, charge as c2 where c1.subject=s.subject and c2.subject=s.object and c1.value != c2.value;
K-r-ll
commented
1 year ago Consider putreanine(1+) (CHEBI:180912) is a carboxylic acid anion (CHEBI:29067)
That a cation(1+) is an anion(1−) seems to be absurd, right? But carboxylic acid anion is defined as R-COO−, where R could bear any charge, including positive. The "net charge" of a carboxylic acid anion does make as much sense as its average mass, that is, none.
In case of entities such as monoatomic boron (CHEBI:33610), diatomic oxygen (CHEBI:33263), triatomic sulfur (CHEBI:33413) the average mass does make sense but the net charge doesn't.
IMHO the easiest fix to this problem would be simply to eliminate the "net charge" from the entries that do not have an INChI in them.
amalik01
commented
1 year ago Some of these charge inconsistencies are related to structures containing R-groups. This can be dealt with by having a comment in the charge section saying the Net Charge excl. R groups. We are currently doing this for the Average Mass and Monoisotopic Mass:
Others, i will manually check and fix.
cmungall
commented
1 year ago Some of these charge inconsistencies are related to structures containing R-groups. This can be dealt with by having a comment in the charge section saying the Net Charge excl. R groups.
Comments are not computable. Consider using a different property so tools can ignore this without string hacking.
cmungall
commented
7 months ago any progress on this?
amalik01
commented
7 months ago Hi @cmungall A quick solution would be to completely remove the charge field from all the parent classes that contain structures with R-groups. Will this be acceptable to you?
charge should be a functional inheritable property (see https://github.com/oborel/obo-relations/pull/723), but this is in conflict
charges-inconsistencies.tsv.txt
Example
cyanidin 3-O-(6-O-acetyl-beta-D-glucoside) (https://www.ebi.ac.uk/ols4/ontologies/chebi/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FCHEBI_131449)
charge +1
subclass
acetate ester
https://www.ebi.ac.uk/ols4/ontologies/chebi/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FCHEBI_47622?lang=en
charge 0