Closed dosumis closed 5 years ago
@cmungall this is also due to the ChEBI roles. We need to fix this issue more globally. Note we do have 'acetylcholine secretion, neurotransmission', which specifies context. I suspect this was added because of the potential hormone role, but the genus term should still not be using the role.
Agreed, we should never use the biological role branch of CHEBI, each should be replaced by a BP.
how about
hormone secretion = secretion and involved-in endocrine signaling
That is assuming hormone is always endocrine.. I guess not. Went down a worm (ho ho) hole here:
This works sometimes, but doesn't work for things like binding terms. I thought about the above suggestions, but is the secretion part of the signaling? It is not part of the signaling pathway by our definitions. At one point I suggested that we have a more global term, signaling and broke it down into the generation of the signal and the reception of the signal. The signaling pathways would be reception. If I recall correctly that idea was rejected.
Can we not use hormone secretion = secretion and upstream of-cell-cell signaling (to also avoid the endocrine issue) ?
cell-cell signalling ligand roles:
hormone . paracrine (local) . autocrine (self) . endocrine (via circulatory system) neurotransmitter (across synapse, regulating postsynaptic membrane potential)
Neurobiology terminology poss source of confusion. Term hormone not generally usex
Neuromodulator: paracrine signalling between neurons that modulates neuronal signalling. (My attempt at def - could probably be improved)
But... Some neuromodulators are released at synapses. Neurotransmitters sometimes work in paracrine fashion in nervous system.
On Fri, Mar 29, 2019, 12:50 PM pgaudet notifications@github.com wrote:
Can we not use hormone secretion = secretion and upstream of-cell-cell signaling (to also avoid the endocrine issue) ?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/17063#issuecomment-477986043, or mute the thread https://github.com/notifications/unsubscribe-auth/AAG4x-FEkJDf-AsZXrsxvEE6L6tKFvQdks5vbgwhgaJpZM4cE7oY .
As well as all the immune system-related roles. Sometimes by the same molecules and signaling pathways. For example dopamine can also act in the nervous system, the immune system or the renal system.
Quick fix: Please move 'acetylcholine sercretion' from under hormone secretion and 'acetylcholine transport' from under 'hormone transport'. If these are inferences deriving from CHEBI then this can be fixed by using the plain CHEBI term in formal defs - not the GOCHE term for the molecule with the role.
It would be great if you could do this soon as embarassing to have to explain to VFB users.
A more complete fix could look something like this:
cell-cell signaling . cell-cell signaling by hormone . . paracrine cell-cell signaling: cell-cell signaling between cells in close proximity. . . autocrine cell-cell signaling: signaling of a cell to itself via an extracellular ligand. . . endocrine cell-cell signaling: cell-cell signaling via a ligand transported through the circulatory system . neurotransmission (already in correct place) . cell-cell signaling by neuromodulator
hormone secretion EquivalentTo: secretion and part_of some 'cell-cell signaling by hormone'
CHEBI roles could (should?) be defined using these GO terms.
The quickest fix would be to remove the logical def of hormone secretion to remove the role term. We can't do it directly at the level of the acetylcholine because it includes both the roles and the chemicals.
OK. So as a quick fix, I will remove the logical definition of 'hormone secretion', which refers to the role. I will look at all of the currently inferred children of hormone secretion and assert the ones that I am confident have rigid roles as hormones. We need to systematically do this for other areas of the ontology, but this should be good here. Sorry it has taken so long @dosumis.
Thanks, this is the best short term course of action and I like DOS's proposal to use GO processes instead of CHEBI for hormone
On Wed, Apr 24, 2019 at 10:35 AM David Hill notifications@github.com wrote:
OK. So as a quick fix, I will remove the logical definition of 'hormone secretion', which refers to the role. I will look at all of the currently inferred children of hormone secretion and assert the ones that I am confident have rigid roles as hormones. We need to systematically do this for other areas of the ontology, but this should be good here. Sorry it has taken so long @dosumis https://github.com/dosumis.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/geneontology/go-ontology/issues/17063#issuecomment-486346229, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAMMOOFNKX6ZWYYZYXPFTTPSCK7NANCNFSM4HATXIMA .
Just to try to be consistent. I am doing the same for hormone transport. Epinephrine transport is asserted as a hormone transport. I am going to remove that one. I'm pretty sure it is also a neurotransmitter. I agree that we should use processes, with the secretion terms it might be safe to use a causal relation...
@krchristie, here is an example of how we solved one of these.
GO currently has:
This is incorrect. neurotransmission != hormone secretion and this is the only case where the two are tangled in GO right now. acetylcholine secretion should be removed from under hormone secretion. If acetylcholine acts as a hormone under some circumstances, then that should be represented by a new 'acetylcholine hormone secretion' term.
This is leading to incorrect classification of neurons as neurosecretory in FBbt which is a problem for VFB.
(More broadly this points to the need for better modelling of roles - the same molecule can play the role of a hormone in some situations and the role of a neurotransmitter in others), but I wouldn't want a general fix for this issue to get in the way of an easy quick fix).