Unobsolete excitatory and inhibitory neuron and fix current excitatory/inhibitory neuron terms

[cmungall]

Previously it was decided that we obsolete excitatory neuron as it was too hard to populate

• Excitatory neuron currently only has retinal bipolar neuron under it #1540

I am sympathetic to the argument - we want to avoid a spaghetti ontology with lots of incompleteness but I think we have to balance this with making an ontology that reflects what domain scientists expect. (also, when decisions are made on CL calls it is vital that we retain the full provenance of decisions, this is totally non apparent when we look at CL now)

Additionally, even if we want to stand by our original decision, we should implement it completely. We still have these terms:

• CL:4023068 ! thalamic excitatory neuron An excitatory neuron that has its soma located in the thalamic complex.
• CL:4023039 ! amygdala excitatory neuron No def

These IMHO are odd groupings and the argument they are hard to populate also holds

The textual definition of the first now refers to a genus that doesn't exist, which is bad...

and the current logical definitions are more specific than either the label or the definition

[Term]
id: CL:4023039 ! amygdala excitatory neuron
intersection_of: CL:0000540 ! neuron
intersection_of: RO:0002100 UBERON:0001876 ! amygdala
intersection_of: RO:0002215 GO:0061535 ! glutamate secretion, neurotransmission

https://www.ebi.ac.uk/ols4/ontologies/cl/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FCL_4023039

[Term]
id: CL:4023068 ! thalamic excitatory neuron
intersection_of: CL:0000540 ! neuron
intersection_of: RO:0002100 UBERON:0010225 ! thalamic complex
intersection_of: RO:0002215 GO:0061535 ! glutamate secretion, neurotransmission

https://www.ebi.ac.uk/ols4/ontologies/cl/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FCL_4023068

Things are even odder on the inhibitory side

id: CL:4023079
name: midbrain-derived inhibitory neuron
def: A GABAergic inhibitory neuron that is derived from the midbrain
intersection_of: CL:0000540 ! neuron
intersection_of: RO:0002202 UBERON:0001891 ! midbrain

(neuronanddevelops fromROsomemidbrainUBERON)

https://www.ebi.ac.uk/ols4/ontologies/cl/classes/http%253A%252F%252Fpurl.obolibrary.org%252Fobo%252FCL_4023079

The text definition is more specific than the label, and the logical definition is even more broad, and the only reason we aren't classifying all midbrain derived neurons here is due to the paucity of brain-region specific developmental relations...

There are other terms that are implicitly inhibitory by label:

• CL:0000498 inhibitory interneuron An interneuron (also called relay neuron, association neuron or local circuit neuron) is a multipolar neuron which connects afferent neurons and efferent neurons in neural pathways. Like motor neurons, interneuron cell bodies are always located in the central nervous system (CNS).
• CL:0008015 inhibitory motor neuron A motor neuron that is capable of directly inhibiting muscle contraction.

these don't have logical definitions. The definition of the first seems to a definition of the parent?

are these not similarly hard to populate

Recommended actions

• unobsolete inh/exc neuron
• add a note to the class that we are working on making classification complete
• fix the 3 specific terms above to have consistent labels, text definitions, and logical definitions
• start coming up with a plan to better populate exc/inh neurons (this may be hard but this should not be a blocker to obsoletion)

Recommended general actions for the future

Some general guidelines that should make their way into the CL editors guide:

• perfection is enemy of the good
• when obsoleting terms, leave a reason that a biologist will understand, and include provenance trail (e.g. links back to ticket)
• if decisions are made on CL calls, try and reflect this in the ticket rather than just saying "decided on call"
• when making decisions, make them consistently, rather than partially, otherwise we end up with a patchwork
• labels, text definitions, and logical definitions should not all be saying wildly different things
  • Read the SPS paper frequently, and memorize S11

[dosumis]

@bvarner-ebi commented on Slack

A related question: are there considerations when adding classes (like excitatory/inhibitory (E/I) neuron) where the classes can never be defined in isolation?

In this example, there are no intrinsic properties of a cell type that alone could define it as an excitatory neuron. While glutamate is mainly an excitatory neurotransmitter in the CNS, this is dependent on the receptor of the post-synaptic cell. There are metabotropic glutamate receptors which are primarily inhibitory, so a neuron that releases glutamate is not de facto excitatory. Whether or not the cell is E/I is determined by external factors. If a glutamatergic neuron is adjacent to a cell with ionotropic glutamate receptors and another cell with metabotropic mGluR2/3 receptors, is it E or I?
With these considerations, it seems using a class like this would not be appropriate for annotating single cell data.

[dosumis]

Suggested definitions:

A neuron that, when it releases neurotransmitter across synapses, completely or predominantly excites cells it is synapsed, changing their membrane potential towards their action potential firing threshold.

A neuron that, when it releases neurotransmitter across synapses, completely or predominantly inhibits cells it is synapsed to, moving their membrane potential away their action potential firing threshold.

[ghost]

A neuron that, when it releases neurotransmitter across synapses, completely or predominantly excites cells it is synapsed, changing their membrane potential towards their action potential firing threshold.

A neuron that, when it releases neurotransmitter across synapses, completely or predominantly inhibits cells it is synapsed to, moving their membrane potential away their action potential firing threshold.

@dosumis, thank you for the suggested text definitions.
Do these classes exclude cell types that are excitatory/inhibitory via electrical synapses? If not, would it be appropriate to remove the reference to neurotransmitter in the definitions?

Consider these following proposed logical definitions:

CL:0008030 'excitatory neuron'
neuron and ‘capable of’ some ‘positive regulation of transmission of nerve impulse’

CL:0008029 'inhibitory neuron'
neuron and ‘capable of’ some ‘negative regulation of transmission of nerve impulse’

[github-actions]

This issue has not seen any activity in the past 6 months; it will be closed automatically in one year from now if no action is taken.