Re: [Obo-relations] connected_to

[mejino@u.washington.edu <mejino@u.washington.edu>]

Additional comments below.

On Thu, 13 Dec 2007, Alan Ruttenberg wrote:
>
> Do we have consensus that the tract connections are to be considered
> distinctly? Granularity would seem to be an issue here. At least in
> the identification of what connects to what - in the case of tracts
> would it be tissue types/anatomical regions?

Is tract a collection of neurites only (cell parts)? 
Tissue (Portion of tissue) is defined as follows:
FMA: Anatomical structure which has as its parts cells of predominantly one type
and intercellular matrix.
CARO: Anatomical structure, which consists of a characteristic set of cell types arranged in some characteristic fashion within an intercellular matrix to form an aggregate.

Tract, like gray matter tissue and white matter tissue (tract is a white matter "tissue"), is not a tissue based on the above definitions but in fact is a 'cardinal tissue part' (collection of cell parts). This is indeed esoteric for a lot of people but consistency with the definitions is a must if we are to move on.

Therefore the complement of the neurite is the soma, the collection of which is generally regarded as the gray matter "tissue" (nuclei, nuclei of origin, etc). Is it the case then that the tract from which it 'projects from' is a continuity relation and to which it 'projects to' is an attachment?



I think it's worth
> reminder here of an issue that has come up recently with the
> definition of Object in BFO - namely that it is dependent on some
> definition of connectivity ("maximally self-connected") and it would
> seem that it is up to the user of BFO to define that connectivity.
> Somehow.




>
> -Alan
>
> On Dec 12, 2007, at 2:49 PM, David Sutherland wrote:
>
>> Hiya,
>>
>> Thanks for the comments and defintitions everyone - very helpful.
>>
>> On Wed, 2007-12-12 at 10:05 -0800, Melissa Haendel wrote:
>>> Hi all,
>>> a couple of points in response to the previous emails:
>>>
>>> Cristian, first, a disclaimer, I am no logician and don't follow
>>> all the
>>> logic of your proposal. That said, I wonder about the use of fiat
>>> boundaries in defining two things as connected. The boundary
>>> between two
>>> potentially connected neurons is not fiat, it has a specific
>>> morphology
>>> and boundary.
>>
>> Yep. If one neuron is synapsed to another, the bona-fide boundary
>> between the two could be the synaptic cleft, or the pre-or post
>> synaptic
>> membrane.
>>
>>> The case for joints between bones is similar. If two
>>> things must share a fiat boundary and not a bona fide one, then
>>> neurons
>>> and bones are not connected, but perhaps something else - like
>>> attached?
>>>
>> I agree.  The FMA paper Barry sent a link to outlines a v.reasonsable
>> approach along thse lines (although not formalised?).
>>
>> Interestingly though, I think the connected_to relation defined in the
>> doc Chritian sent works for the connection between an axon tract or a
>> single axon and a neural compartment.  In fly brains, axon tracts
>> follow
>> glial rich boundary regions between synapse rich neural compartments.
>> If we consider axons leaving a tract and entering a neural compartment
>> to have a distal part which is part_of that comparment, then there
>> is no
>> bona-fide boundary in the axon which can define the point where this
>> part relation begins - only a fiat boundary based on criteria external
>> to that axon.
>>
>> i.e.- looks like Alan is right to suggest we need separate
>> relations for
>> synaptic 'connections' and axon tract to neural compartment
>> 'connections'.
>>
>> Cheer again,
>>
>> David
>>
>>> Alan, good point: "What inferences should follow from "connected_to".
>>> Is connected_to transitive? If I am reading Cristian's definition
>>> right
>>> one would suppose so. However, I wouldn't think that we would want to
>>> infer that because neuron A is connected_to neuron B, and neuron B is
>>> connected_ neuron C, that A is connected_to C, or else the whole
>>> nervous
>>> system would be connected, and then whats the point?
>>>
>>> Additional comments about paula's email below.
>>>
>>> cheers
>>> melissa
>>>
>>> David Sutherland wrote:
>>>> Hi Paula,
>>>>
>>>> Thanks for the comments.
>>>>
>>>> On Wed, 2007-12-12 at 10:12 -0600, Paula Mabee wrote:
>>>>
>>>>> Hi David and Melissa,
>>>>>
>>>>>  My comments don't pertain to your high level questions about
>>>>> the nature of
>>>>> the relation, so I didn't cc: the list or others here.
>>>>>
>>>> Hope you don't mind, but I've added the list back as cc because I
>>>> think
>>>> you've brought up a very important point about formal relations
>>>> which is
>>>> closely related to the problems are grappling with.
>>>>
>>>>
>>>>> I wondered about this relation just last night while thinking
>>>>> about the
>>>>> proposed addition of "joint" as an anatomical cluster (on ZFA
>>>>> and TAO
>>>>> tracker currently).  As proposed, "joint" is defined as a
>>>>> particular kind of
>>>>> connection, and there will be lots of joints (e.g. Ethmoid bone -
>>>>> prevomer
>>>>> bone joint).  Similarly, in the case of muscle-bone connections
>>>>> (e.g.
>>>>> insertions, origins), one could define "insertion" as a
>>>>> particular kind of
>>>>> connection, and then define the many insertions in the body
>>>>> (e.g. biceps
>>>>> muscle - radius bone).
>>>>>
>>>>> I had wondered about using connected_to for joint: ethmoid bone
>>>>> connected_to
>>>>> prevomer bone and biceps muscle_connected to radius bone, but
>>>>> then we lose
>>>>> the information about the nature of the connection.
>>>>>
>>> One could certainly use logical definitions to define a joint as any
>>> bone part connected_to another bone part, and a muscle attachment
>>> site
>>> as any bone part connected_to any muscle part, etc. But see below,
>>> joints aren't necessarily only a connection between two bones.
>>>>> Not sure that this relates to your issues, but it seems to me that
>>>>> connected_to is a very low information relation, and like
>>>>> Melissa, I think
>>>>> it needs some refinement.  I'm trying to think about the queries
>>>>> that need
>>>>> to recover information about connectedness and for our purposes
>>>>> it seems
>>>>> that the proposed is better.
>>>>>
>>>>>  Paula
>>>>>
>>>>>
>>>>> PROPOSED:
>>>>> Adopt Joint definition from FMA: Joint ... consists of two or
>>>>> more adjacent
>>>>> bones or cartilages, parts of which are interconnected by organ
>>>>> parts that
>>>>> consist of various types of connective tissue.
>>>>>
>>>>> Joint is_a anatomical cluster.
>>>>> Ethmoid -prevomer joint is_a joint
>>>>>
>>> We had considered joints as anatomical clusters previously because
>>> they
>>> are more than just two bones coming together, they have specific
>>> morphology, ECM, and often ligaments and tendons that all together
>>> make
>>> up the joint.
>>>>> ethmoid bone part_of ethoid-prevomer joint
>>>>> prevomer bone part_of ethmoid-prevomer joint
>>>>>
>>>>>
>>>> Is this a legal use of part_of?  I presume the ethmoid bone is not
>>>> completely part_of the joint.  It's my understanding that his is
>>>> what
>>>> part_of, as defined in the RO, means.  Perhaps you need to say
>>>> distal part of ethomid bone part_of ethoid-premover joint?
>>>>
>>>> This is precisely the problem we have for part_of and located_in
>>>> applied
>>>> to the nervous system.  A whole neuron or a whole axon tract may
>>>> be so
>>>> spread out that we can only capture the most trivial/gross
>>>> part_of or
>>>> located_in information (to brain or CNS).  The question is, do we
>>>> need
>>>> to define terms for bits of tracts and bits of specific neurons or
>>>> neuron types in order to capture this more specific part
>>>> information, or
>>>> are there
>>>> (a) relations we can use to capture partial location/part relations
>>>> (e.g.- partially_located_in)
>>>> &/OR
>>>> (b) ways to refer to the relevant bits of an axon tract or neuron
>>>> using
>>>> post-composition within a logical definition
>>>> - e.g.- some way to refer to 'distal part of X' as a genus terms,
>>>> where
>>>> X is some type for which we already have a term in the ontology.
>>>>
>>> This approach uses spatial qualities in combination with the part_of
>>> relation. One could also use them in combination with the
>>> connected_to
>>> relation: 'distally connected_to'.
>>>
>>>> Cheers,
>>>>
>>>> David
>>>>
>>>>
>>>>> ALTERNATIVE:
>>>>> Ethmoid connected_to prevomer
>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>>> On 12/12/07 9:07 AM, "Melissa Haendel"
>>>>> <Ha...@uo...> wrote:
>>>>>
>>>>>
>>>>>> Hi David,
>>>>>>
>>>>>> We too have have been considering this relation to record
>>>>>> muscle and
>>>>>> ligament attachment. For this particular application,
>>>>>> directionality is
>>>>>> not an issue, but certainly for numerous others it would be. It
>>>>>> seems like
>>>>>> one might want to further refine the connectedness, perhaps
>>>>>> using some of
>>>>>> the relational spatial qualities? For example, a given neuron
>>>>>> might be
>>>>>> dorsally connected_to another neuron. But maybe this is a bit
>>>>>> much to bite
>>>>>> off at once.
>>>>>>
>>>>>> -Melissa
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>> Hi all,
>>>>>>>
>>>>>>> I'm working with a bunch of neurobiologists right now on
>>>>>>> fixing up the
>>>>>>> neuroanatomy part of the Drosophila anatomy ontology.  They
>>>>>>> are very
>>>>>>> keen to record connectivity - initially on the gross level of
>>>>>>> axon
>>>>>>> tracts (connectivity of individual neurons being, of course, a
>>>>>>> very
>>>>>>> large dataset).
>>>>>>>
>>>>>>> I noticed that RO proposed contains the undefined relation:
>>>>>>>
>>>>>>> [Typedef]
>>>>>>> id: OBO_REL:connected_to
>>>>>>> name: connected_to
>>>>>>> !is_symmetric: true {instance_level=true}
>>>>>>> def: "" []
>>>>>>>
>>>>>>> I have a few questions about this:
>>>>>>>
>>>>>>> (i) Is it likely to be defined soon & if so would it be unwise
>>>>>>> or not to
>>>>>>> begin using it now?
>>>>>>> (ii) Does the "!is_symmetric" in this stanza indicate that
>>>>>>> this is
>>>>>>> intended as an asymmetric relation at the instance level? (!
>>>>>>> indicating
>>>>>>> negation).  Directionality could certainly be useful for
>>>>>>> recording
>>>>>>> neural connectivity, but it seems to me that a symmetric
>>>>>>> relation would
>>>>>>> be more generally applicable.
>>>>>>>
>>>>>>> Cheers,
>>>>>>> David
>>>>>>>
>>>>>>> [PS - Bill: cc'ing you as I figure this is an issue you'll
>>>>>>> have already
>>>>>>> thought about alot - any suggestions most welcome.]
>>>>>>>
>>>>>>>
>>>>>>> -----------------------------------------------------------------
>>>>>>> --------
>>>>>>> SF.Net email is sponsored by:
>>>>>>> Check out the new SourceForge.net Marketplace.
>>>>>>> It's the best place to buy or sell services for
>>>>>>> just about anything Open Source.
>>>>>>> http://sourceforge.net/services/buy/index.php
>>>>>>> _______________________________________________
>>>>>>> Obo-relations mailing list
>>>>>>> Obo...@li...
>>>>>>> https://lists.sourceforge.net/lists/listinfo/obo-relations
>>>>>>>
>>>>>>>
>>>
>>
>>
>> ----------------------------------------------------------------------
>> ---
>> SF.Net email is sponsored by:
>> Check out the new SourceForge.net Marketplace.
>> It's the best place to buy or sell services
>> for just about anything Open Source.
>> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/
>> marketplace
>> _______________________________________________
>> Obo-relations mailing list
>> Obo...@li...
>> https://lists.sourceforge.net/lists/listinfo/obo-relations
>
>
> -------------------------------------------------------------------------
> SF.Net email is sponsored by:
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services
> for just about anything Open Source.
> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
> _______________________________________________
> Obo-relations mailing list
> Obo...@li...
> https://lists.sourceforge.net/lists/listinfo/obo-relations
>