search

pombase / fypo

Fission Yeast Phenotype Ontology
Creative Commons Attribution 4.0 International
14 stars 6 forks source link

Redo logical definition for 3 classes #3696

Open matentzn opened 4 years ago

matentzn commented 4 years ago

'increased H2A.Z level at centromere inner repeat' 'increased H2A.Z level at subtelomere' 'normal H2A.Z level at centromere'

Again @mah11, wait for after we are done.

mah11 commented 4 years ago

Note to self: several others fit the general case of 'level of some protein at some SO-identifiable location'. @matentzn do we have a pattern for that (or a placeholder model LD to use in the meantime)?

matentzn commented 4 years ago

You could check here. https://github.com/obophenotype/upheno/tree/master/src/patterns/dosdp-patterns

Maybe this one: https://raw.githubusercontent.com/obophenotype/upheno/master/src/patterns/dosdp-dev/abnormallyIncreasedLevelOfChemicalEntityInLocation.yaml ?

For the normal one, you could use the abnormal equivalent: https://raw.githubusercontent.com/obophenotype/upheno/master/src/patterns/dosdp-patterns/abnormalLevelOfChemicalEntityInLocation.yaml

And simply use "normal" as a modifier instead of abnormal?

matentzn commented 4 years ago

Dont worry about the Protein vs Chemical entity discussion now.. all of these fall under Molecular entity, which really is the scope of these patterns.

mah11 commented 4 years ago

Maybe ... we've got a bunch like that for locations like 'cell', 'vacuole', 'chromatin', etc. The additional feature for the H2A.Z terms and the other ones I've spotted are that the location is ('chromatin' and at some 'sequence feature'), i.e. not located to all chromatin, just the part of chromatin where the DNA has a sequence specified by a given SO term. Can the pattern use a compound construction like that for the location?

matentzn commented 4 years ago

Do you want to try to create a definition that looks something like the ones in the pattern, and then amend it to what you think it should look like, and post it here? This sounds like we need a slightly more specific pattern for what you want, but its better if I can see the definition you would suggest before making a comment..

mah11 commented 4 years ago

All righty, I've had a go ... really flailing around here, not even sure what to call it ... maybe abnormalLevelOfChemicalEntityInChromatinLocation or abnormalLevelOfChemicalEntityAtSequenceInChromatin ? Anyway ...

equivalentTo:
  text: "'has_part' some ('amount' and ('inheres_in' some (%s and ((some %s 'in' ('part_of' some %s)))) and ('has_modifier' some 'abnormal'))"
  vars:
    - chemical_entity
    - sequence feature
    - location

note that only a few options would make sense for the value of $location - chromatin, chromosome, maybe transcript (RNA molecule)

I'm not sure I got the nesting right, and I don't know what in RO might work for the "in" relation (it's completely made up for this attempt). 'Sequence feature of' or 'subsequence of' or ..??

another idea, with different nesting and variable order:

equivalentTo:
  text: "'has_part' some ('amount' and ('inheres_in' some (%s and ('part_of' some (%s at %s)))) and ('has_modifier' some 'abnormal'))"
  vars:
    - chemical_entity
    - location
    - sequence feature
matentzn commented 4 years ago

Looping in @dosumis for after he is out from his grant hole.