Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.Sign up
Document design patterns for named nerves #298
The representation of nerves and their parts varies wildly from ontology to ontology. We should create a shared set of design patterns for nerves.
This item serves as a draft for this design pattern. The text here will be altered on the basis of comments entered below.
Anyone is welcome to comment. Particularly interested in comments on nerve innervation from David OS and @RDruzinsky
the focus of this articles is vertebrates, but much can be learned from work done on invertebrates. See in particular the representation of neuron connectivity in VFB and FlyBase:
A strategy for building neuroanatomy ontologies
A fairly standard subdivision into cranial and spinal:
Note however, that not every CN is a true PNS nerve. More on this later.
Nerves can be subdivided
We focus here on 1 as this is where some inconsistencies currently lie. However, our answers to 1 depend to a large extent on 2, as nerves must be defined in terms of cellular components such as axons.
The open question here is roots - are roots considered part of the nerve, part of the CNS (ie brain for CNs), or do they overap? We have to be careful as if the CNS and PNS are considered not to overlap, and nerves are considered entirely part of the PNS we get incoherent ontologies quickly.
We must also consider axon tracts here. E.g.
What is the relationship to the brain structure sending the signal? In spatial terms the answer depends in part on how we handle roots above, and must be guided by relationships at the level of cell components
The RO relation needs generalizing to be applicable to muscles, skin etc
Is the relationship between the placodes develops_from or has developmental contribution from?
For cranial nerves (CNs) here is a subset of the current hierarchy
Do we consider a nerve a branching structure or a tube-like structure? I.e. is a branch of nerve X also a part of nerve X?
The answer here should be informed by (but not necessarily identical to) how we treat branching on blood vessels, etc. See #170
Differences in branching may also provide problems for a multi-species representation. CNs are relatively well preserved, but named branches may tend to be lineage specific?
Remember this isn't truly a PNS nerve
Treatment in other ontologies
FMA has an (undefined) "central zone of nerve"
We assume that "central zone" refers to the zone that is part of the CNS. Thus a nerve - in FMA - is not constrained to be purely PNS (unless we allow CNS and PNS to overlap, which may be an odd decision)
The same hierarchy with parthood relations
It's not clear why these are only named for a subset of the cranial nerves.
We can also look at the terminology used:
Note that FMA considers "X nerve fibers" "central part of X" and "X tract" as synonyms. Is this correct?
Nerve fibers surely extend into the PNS portion (unless the FMA considers the CNS "tunneling through" the nerves which might be an interesting approach)
Note that NIFSTD creates separate classes for nerve fibers and tracts.
Here are some nerve fibers in NIFSTD:
Note that in NIFSTD the CNS and PNS part-overlap:
The ON root is part of the brain and part of the PNS
Note the same pattern is not followed for all roots.
Digging up this issue since it is the cause of many incompatibilities with NIFGA (while these terms are deprecated I want to use them to make sure that when we transition the representations are at least minimally compatible).
At the moment my primary concern is with the optic nerve wrt the cranial nerves. The the simplest way to resolve this issue seems to be to treat 'cranial nerves' as a more colloquial term that includes the optic nerve and the 'peripheral cranial nerves.' This would allow us to treat the optic nerve separately without having to accommodate a definition of nerve that is both central and peripheral. The alternative seems to be to split nerve into a peripheral and a central class (since it is not reasonable to model the optic nerve as a white matter tract).
what we ended up doing but not documented in this ticket was a compromise: CN retains a nerve classification, but we introduce a new grouping UBERON:0034713 ! cranial neuron projection bundle
here is a subset:
it's not ideal. The weirdly named grouping is more inclusive than you might want as it contains other projection bundles beyond the classic ones. But everything is consistent at least.
I'm open to your suggestion. If implemented how would you feel about