Muscle names (labels)

[RDruzinsky]

I am submitting this issue to get some comments on a problem that we have encountered in the FEED AO. Homologies between oro-pharyngeal muscles in mammals and "more primitive" vertebrates are uncertain. Unfortunately, some of the names are identical.

For example, "hyoglossus" is used in mammals, birds, reptiles, amphibians, etc. Putting this muscle into a single class implies, I believe, homology, and I don't want to do that. In addition, it is impossible to create a logical definition that is true across vertebrates. For the moment I have created one class, called "hyoglossus muscle" (http://purl.obolibrary.org/obo/MFMO_0000064) for the mammalian muscle and another, called "m. hyoglossus" (http://purl.obolibrary.org/obo/MFMO0000203) for the "reptilian" or, more properly, the "non-mammalian" class. I feel that this is a very poor way to handle this, but I don't have any bright ideas. Do you? Thanks, Robert

[cmungall]

There is no implication of homology unless there is an entry in the homology table (I will loop you into a discussion with phenoscape & bgee curators on this shortly, am about to start working on the homology issue more intently, would like to collect FEED use cases for homology cc @hlapp )

The first thing to look at is the definition:

def: "A muscle that attaches to the hyopid bone and to the tongue and is innervated by cranial nerve XII" [FEED:rd]

Does this definition apply to non-mammals? If not, then this class should get a mammal taxon constraint, and then the problem is purely terminological.

If it does, then we have a design decision and some options.

The way that is most consistent with how we have done things thus far would be to retain UBERON:0001572 ! hyoglossus muscle as a taxon generic class as defined by its definition. We have the option of making further subclasses. Ideally these would have non-taxon discriminating features even if they correlate with homologous groupings.

[fbastian]

Hi Robert,

as classes in Uberon do not imply homology, you might be interested in our effort to capture homology information, see http://sourceforge.net/p/bgee/wiki/Similarity%20annotation/ and http://svn.code.sf.net/p/bgee/code/trunk/release/similarity/similarity.tsv.

We welcome contributions, and hope to collaborate with the Monarch initiative.

[hlapp]

There is no implication of homology unless there is an entry in the homology table

Technically this is correct, but I think it's better to consider this from the fact that most scientists who apply Uberon apply with the assumption that if a Uberon term is applied to two anatomical elements in two different organisms, then these two elements are biologically "the same". If the two organisms are from two different species, that assumption implies homology.

Uberon is purposely engineered such that making this assumption is typically well founded ("uncontroversial homology is built in"), so one question here is how to deal with the hyoglossus situation such that keeps the principles of the current design.

[fbastian]

In such cases, we use a negative annotation to assert that it is NOT homologous among vertebrates, and two positive annotations, to assert that it is homologous in mammalian on the one hand, and homologous in non-mammalian on the other hand.

[RDruzinsky]

Hi, Thanks for the invitation. I would like to participate in the efforts to capture homology. And, I think that the issue that I have raised should be central to at least some of your discussion.

The problem is not, precisely, the one addressed by Hilmar. Young morphologists have to become comfortable dealing with structures that have the same names in different taxa but are not necessarily homologous. This gives people many opportunities to write numerous papers arguing about the homologies.

The problem in Uberon is not just with the names, it is with the logical definitions. Consider the hyoglossus, as mentioned in the original post:

In mammals the muscle may be defined, uniquely, as: "muscle and innervated_by CN XII and attaches_to some tongue and attaches_to some hyoid_bone." But non-mammals do not have a hyoid bone. The traditional definition for a "non-mammalian" hyoglossus muscle would be: muscle and innervated_by CN XII and attaches_to some tongue and attaches_to some 1st ceratobranchial.

So, we could just do what I suggested by using slightly different names for the mammalian and non-mammalian muscles, and give a taxon constraint to the mammalian class, as Chris suggested.

This does not get to the thorny issue of "homology." Since the 1st ceratobranchial does not form part of the mammalian hyoid, is the muscle homologous to the mammalian hyoglossus?? It probably is, but how do we describe that? Thanks,Robert

Robert E. Druzinsky, Ph.D. Clinical Associate Professor Dept. of Oral Biology College of Dentistry University of Illinois at Chicago 801 S. Paulina Chicago, IL 60612 druzinsk@uic.edu

Office: 312-996-0406 Lab: 312-996-0629 Website: www.peerj.com/RobertDruzinsky

On Wed, Aug 20, 2014 at 8:22 AM, fbastian notifications@github.com wrote:

In such cases, we use a negative annotation to assert that it is NOT homologous among vertebrates, and two positive annotations, to assert that it is homologous in mammalian on the one hand, and homologous in non-mammalian on the other hand.

— Reply to this email directly or view it on GitHub https://github.com/obophenotype/uberon/issues/548#issuecomment-52776696.

[fbastian]

In our table, we chose to assign a "confidence code" to the homology assertions, to express uncertainty. Also, we provide one evidence code for each assertion (as in GO annotations).

In your case, we might have an assertion: NOT homologous in vertebrates, based on ECO:0000060 positional similarity evidence, confidence low; and a contradicting assertion: homologous in vertebrates, based on ECO:0000071 morphological similarity evidence, confidence low (for instance)

[RDruzinsky]

Interesting. I will have to learn more about this. Thanks!

Robert E. Druzinsky, Ph.D. Clinical Associate Professor Dept. of Oral Biology College of Dentistry University of Illinois at Chicago 801 S. Paulina Chicago, IL 60612 druzinsk@uic.edu

Office: 312-996-0406 Lab: 312-996-0629 Website: www.peerj.com/RobertDruzinsky

On Wed, Aug 20, 2014 at 10:40 AM, fbastian notifications@github.com wrote:

In our table, we chose to assign a "confidence code" to the homology assertions, to express uncertainty. Also, we provide one evidence code for each assertion (as in GO annotations).

In your case, we might have an assertion: NOT homologous in vertebrates, based on ECO:0000060 positional similarity evidence, confidence low; and a contradicting assertion: homologous in vertebrates, based on ECO:0000071 morphological similarity evidence, confidence low (for instance)

— Reply to this email directly or view it on GitHub https://github.com/obophenotype/uberon/issues/548#issuecomment-52796799.

[cmungall]

On 20 Aug 2014, at 8:26, Robert Druzinsky wrote:

Hi, Thanks for the invitation. I would like to participate in the efforts to capture homology. And, I think that the issue that I have raised should be central to at least some of your discussion.

The problem is not, precisely, the one addressed by Hilmar. Young morphologists have to become comfortable dealing with structures that have the same names in different taxa but are not necessarily homologous. This gives people many opportunities to write numerous papers arguing about the homologies.

The problem in Uberon is not just with the names, it is with the logical definitions. Consider the hyoglossus, as mentioned in the original post:

In mammals the muscle may be defined, uniquely, as: "muscle and innervated_by CN XII and attaches_to some tongue and attaches_to some hyoid_bone." But non-mammals do not have a hyoid bone. The traditional definition for a "non-mammalian" hyoglossus muscle would be: muscle and innervated_by CN XII and attaches_to some tongue and attaches_to some 1st ceratobranchial.

According to this, the problem becomes a terminological / presentation issue. If the definition references the hyoid bone, and the hyoid bone is restricted to mammals then the definition is restricted to mammals. It just remains how we present this via the labels.

However, it's not this simple, as the above isn't consistent with our current representation. We currently use the class "hyoid bone" generically, up to amphibians.

This is partly justified by the follow medium confidence gene expression homology assertion:

HOM:0000007 historical homology UBERON:0001685 hyoid bone 7742 Vertebrata RAW ECO:0000075 gene expression similarity evidence CONF:0000004 medium confidence assertion from single evidence DOI:10.1038/nrn1221 Santagati F, Rijli FM, Cranial neural crest and the building of the vertebrate head. Nature Reviews Neuroscience (2003) [In vertebrates] Hoxa2 expression in NCCs is necessary for the patterning of hyoid skeletal elements. bgee ANN 2013-09-05

The hyoid could probably do with work, starting with improving the wikipedia definition: http://purl.obolibrary.org/obo/UBERON_0001685

So, we could just do what I suggested by using slightly different names for the mammalian and non-mammalian muscles, and give a taxon constraint to the mammalian class, as Chris suggested.

possibly - it looks like we have to get the skeletal elements right first and build up from there

(I think the skull in general is hard but the pharyngeal elements should hopefully be more straightforward)

This does not get to the thorny issue of "homology." Since the 1st ceratobranchial does not form part of the mammalian hyoid, is the muscle homologous to the mammalian hyoglossus?? It probably is, but how do we describe that?

We don't necessarily need to explicitly describe everything. If we do restrict hyoid to mammals, and end up with parallel representations then we can retrieve parallel structures using more sophisticated query techniques.

For example, the non-mammal hyoglossus would score highly in a semantic similarity query to mammal hyoglossus, based on general location and innervation and tongue attachment, and attachment to a skeletal derivative of some anterior pharyngeal arch. This is even without explicit statements of homology.

It partly depends on how we want to make use of homology in presentation of the ontology, data and querying.

Thanks,Robert

[RDruzinsky]

I think that it would be very cool to get to the point, for example, that the attachment to the 1st ceratohyal would exclude mammals, by virtue of the fact that the 1st ceratohyal is not a part of the mammalian hyoid bone.

The reference that you cited, I believe (HOM:0000007 historical homology UBERON:0001685 hyoid bone 7742 Vertebrata RAW ECO:0000075 gene expression similarity evidence CONF:0000004 medium confidence assertion from single evidence DOI:10.1038/nrn1221 Santagati F, Rijli FM, Cranial neural crest and the building of the vertebrate head. Nature Reviews Neuroscience (2003) [In vertebrates] Hoxa2 expression in NCCs is necessary for the patterning of hyoid skeletal elements. bgee ANN 2013-09-05)

shows that the hyobranchial apparatus, as a whole, is derived from neural crest, but I don't think that it parses out the individual components of the apparatus and their phylogenetic histories.

[fbastian]

It is true that this assertion is not well supported: it is an HOM:0000007 "historical homology" assertion, based on one single evidence ECO:0000075 "inferred from gene expression similarity", while this is actually equivalent to HOM:0000003 "homocracy" ("Similarity that is characterized by the organization of anatomical structures through the expression of homologous or identical patterning genes").

So this assertion need to be reviewed. The good thing is, we can automatically detect such problems with the formalism adopted. We would be happy to integrate any other source of evidence you could provide.

[gouttegd]

WARNING: This issue has been automatically closed because it has not been updated in more than 3 years. Please re-open it if you still need this to be addressed addressed addressed – we are now getting some resources to deal with such issues.