upheno
upheno copied to clipboard
obophenotype
Review abnormalCellProliferation
abnormalCellProliferation.yaml
Currently PLANP would be unable to use this pattern since it is restricted to
anatomical entity: UBERON:0001062
The planarian anatomy ontology (PLANA) has 'anatomical entity' but it has a PLANA id and has UBERON:0001062 as a dbxref.
There are quite a few other anatomy ontologies that have the same design.
I suggest using entity: BFO:0000001
Also, this pattern uses 'inheres_in_part_of: RO:0002314'. All of the other patterns PLANP has used 'inheres_in: RO:0000052'. Would this be an issue?
Sofia
In order for anatomy patterns to interoperate, we need to align the external anatomy ontologies (such as wbbt, zfa, plana, xao) with UBERON. This is going to be a gradual process we will realise later in the process by curating xrefs between those ontologies and Uberon.
So with that goal in mind it is best to leave all the UBERON restrictions in there; you can use the pattern even without uberon alignment in any case; problems occur only when checking pattern conformance later.
Yep, this should work for any metazoan - we should think what our strategy is for including fungi and plants
Can someone explain the difference between 'inheres_in_part_of' and 'inheres_in'. I see that most of the patterns use 'inheres_in'. Will it be an issue to use both in one ontology?
`inheres in part of is just a chain of something that inheres in something that is part of something else. So if you had a triangular that inheres in a dorsal fin, and dorsal fin is part of body, you could say ‘triangular and inheres in part of some body’. But it doesn’t inhere in it; the body is not triangular.
I think in this pattern we would want to use 'quality' instead of 'process quality'
equivalentTo: text: "'has_part' some ('process quality' and ('inheres_in_part_of' some (cell_proliferation and ('occurs in' some %s))) and ('qualifier' some 'abnormal'))" vars: - anatomical_entity
@nicolevasilevsky can you give some explanations why? just learning pato :P
I am only basing this on the fact that MP uses quality instead of process quality - I actually don't know what is the appropriate term to use.
It doesn't look like MP uses process quality at all in any of the logical defs.
HP uses it in only two terms:
HP_0012647 'Abnormal inflammatory response' 'has part' some ('process quality' and ('inheres in' some 'inflammatory response') and ('has modifier' some abnormal))
HP_0040224 'Abnormality of fibrinolysis' 'has part' some ('process quality' and ('inheres in part of' some fibrinolysis) and ('has modifier' some abnormal))
I dont mind either way; my general feeling is to pick the constraints that most accurately reflect the intention of the pattern; if this pattern is about abnormal process quality, it should use that, no matter who is using it how at the moment; lets see what the others say.
It is important to use process quality here and it should be checked in other patterns as well. But let’s discuss if other mods are okay with this step as it might cause more work, because existing defs need to be revised
I agree that if entity= process a process quality should be used. This is how it was originally intended to be used as I understand it. This is the way ZFIN has used it and constrains annotations with this in mind.
DPO also uses children of 'process quality' for processes.
For this pattern, I don't think the name and definition will work well if you want to use a cell type e.g. neuroblast, rather than a location e.g. brain. I'm not sure if type and location style terms can be incorporated into the same pattern, so there might need to be another pattern, maybe "abnormalCellProliferationByCellType" and rename the current one "abnormalCellProliferationByLocation".
If I understand @cmungall and his OBO core efforts correctly, we will soon have some kind of minimal biological upper layer which has a concept called 'Anatomical structure', that encompasses both cells and anatomical entities/regions. However, it is indeed debatable whether something that affects all cells of a certain kind is the same as something that effects an anatomical entity; i mean AFAIK cells of one type can be kind of scattered everywhere and not really belong to a coherent anatomical structure. Food for thought! @dosumis @cmungall
Wouldn’t it technically be possible to state in the pattern that multiple classes can be possible here? E.g. „cell“ or „anatomical entity“?
Sent from mobile
Am 06.02.2019 um 13:02 schrieb Nico Matentzoglu [email protected]:
If I understand @cmungall and his OBO core efforts correctly, we will soon have some kind of minimal biological upper layer which has a concept called 'Anatomical structure', that encompasses both cells and anatomical entities/regions. However, it is indeed debatable whether something that affects all cells of a certain kind is the same as something that effects an anatomical entity; i mean AFAIK cells of one type can be kind of scattered everywhere and not really belong to a coherent anatomical structure. Food for thought! @dosumis @cmungall
— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.
No, by design DOSDP is taking the dead simple very seriously.. You will need to have two patterns.
a few things that may be relevant. In CARO cells are anatomical entities, I thought this was still true in the OBO Core (as per @matentzn comment above). Agree that process quality should be used here, proliferation is a process. It does seem like we need two patterns, one that is simply abnormal process, and one where an abnormal process is part of another anatomical entity (and thereby infer that there may be phenotypic outcomes upon that entity)? Is this what you mean @matentzn ?
Oh, I might have misread the pattern before. Yeah @mellybelly you are right, probably we need three patterns here:
- abnormalCellProliferation (or what @Clare72 calls more accurately abnormalCellProliferationByCellType) (domain: cell type)
- abnormalCellProliferationInALocation (domain: anatomical entity)
- abnormalCellProliferationByCellTypeInALocation (if necessary)
The domain should be as specific as possible, not anatomical structure in general! @Clare72 good catch.
I thought the problem is the missing class that unites cells and other anatomical entities!?
Not in this case; because if we distinguish the patterns as I suggest, we can have a distinct pattern for each one. But for other patterns where this is relevant: we will have a class soon, in OBO core.
equivalentTo: text: "'has_part' some ('process quality' and ('inheres_in_part_of' some (cell_proliferation and ('occurs in' some %s))) and ('qualifier' some 'abnormal'))" vars:
- anatomical_entity
We should align compound entities with source ontologies - using the relationships they use. This maximises inference.
In this case we should use whatever GO uses. GO has:
acinar cell proliferation: 'cell population proliferation' and (acts_on_population_of some 'acinar cell')
cell proliferation in forebrain: 'cell population proliferation' and ('occurs in' some forebrain)
We should therefore have two patterns reflecting this:
abnormal_proliferation_of abnormal cell proliferation_in
The range of the var in the first pattern should be 'cell'
Ideally the range of the var in the second pattern would be 'multicellular anatomical structure'. This is in CARO, but we need to check whether Uberon supports it. Failing that we can use 'anatomical structure' which subsumes cell but name write the pattern description and name the var to make clear we want multicellular structures only.
Discussion on Phenotype call:
All agree that we should follow GO and have two patterns as described above -
X cell proliferation: 'cell population proliferation' and (acts_on_population_of some 'X')
cell proliferation in X: 'cell population proliferation' and ('occurs in' some X)
- The compound of the two:
'cell population proliferation' and (acts_on_population_of some 'X') and ('occurs_in some X)
I am confused, should abnormal cell proliferation in an anatomical entity have the logical def: 'cell population proliferation' and (acts_on_population_of some 'X') and ('occurs_in some X)
?
here is one pattern: https://github.com/obophenotype/upheno/blob/master/src/patterns/dosdp-dev/abnormalCellProliferationInAnatomicalEntity.yaml
not sure if this is right, but it is a starting place