New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Last Batch of Potential Modeling Issues in Uberon #1277

Open
ChrisOchs opened this Issue Oct 7, 2016 · 17 comments

Comments

Projects
None yet
2 participants
@ChrisOchs

ChrisOchs commented Oct 7, 2016

This is the last set of modeling inconsistencies/potential errors our domain expert identified in a random sample of Uberon concepts.

anterodorsal nucleus of medial geniculate body
Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.
Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

anterodorsal nucleus of medial geniculate body
It is interesting as many of the different parts of the brain and CNS are not part_of each other. For example, why isn't the diencephalon which is_a regional part of brain but not is_a part of brain. Could have eliminated a lot of redundancy?

decussation of trochlear nerve
Remove restriction: The decussation is just anatomic fact that each trochlear nerve crosses from one side to the other, therefore I'm not sure that it is correct to apply extends_fibers_into

immature eye
Remove restriction: The target eye primordium is used both for develops_from and has_development_contribution_from which is redundant. One should be removed or one of the values should be replaced. For example develops_from ecto-epithelium?

lateral forebrain bundle:
Change restriction: Refine part_of with forebrain instead of CNS

postganglionic sympathetic fiber
Change restriction: Might be my misunderstanding but this is post galglionic so it extends_fibers_from and not extends_fibers_to. See definition.

presumptive paraxial mesoderm
To be consistent with presumptive endoderm this should be part_of blastula and not mesoderm. Or vice versa.

subthalamic fasciculus
The subthalamic fasciculus runs from the subthalamic nucleus to the globus pallidus so I'm not sure it is part_of ventral thalamus or is_a ventral thalamic fasciculus.

postcommissural fornix of brain
Remove superclass: I think it is more correct to say that this is part_of the fornix than is_a and therefor the parent should be removed and added as a part_of relationship with the fornix of brain as the target.

piriform cortex layer 2a
Add restriction: add/refine part_of with rhinencephalon and/or telencephalon

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

anterodorsal nucleus of medial geniculate body
Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.
Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

How are you visualizing uberon? All the relationships here are currently correct and complete AFAICT:

t

Member

cmungall commented Oct 15, 2016

anterodorsal nucleus of medial geniculate body
Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.
Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

How are you visualizing uberon? All the relationships here are currently correct and complete AFAICT:

t

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

anterodorsal nucleus of medial geniculate body
It is interesting as many of the different parts of the brain and CNS are not part_of each other. For example, why isn't the diencephalon which is_a regional part of brain but not is_a part of brain. > Could have eliminated a lot of redundancy?

Sorry, not following, but this could be an artefact of how you are visualizing the ontology

Member

cmungall commented Oct 15, 2016

anterodorsal nucleus of medial geniculate body
It is interesting as many of the different parts of the brain and CNS are not part_of each other. For example, why isn't the diencephalon which is_a regional part of brain but not is_a part of brain. > Could have eliminated a lot of redundancy?

Sorry, not following, but this could be an artefact of how you are visualizing the ontology

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

immature eye
Remove restriction: The target eye primordium is used both for develops_from and > has_development_contribution_from which is redundant. One should be removed or one of the values should be replaced.

This is a more subtle issue but again I think this comes down to the tool being used. This is a hidden GCI pattern, which is a powerful axiom construct for inference. But most users should not need to worry about it, if the tool only shows the existential graph, the seemingly weaker axiom will not be there.

We clearly need to provide better guidelines for developers of portals and GUIs!

I recommend OLS

http://www.ebi.ac.uk/ols/ontologies/uberon/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FUBERON_0010312

unfortunately it does not show develops-from in the hierarchical view, but it cleanly separates the Equivalence axiom up top from the some-values-from relationships

For example develops_from ecto-epithelium?

We already have this path from eye primordium

Member

cmungall commented Oct 15, 2016

immature eye
Remove restriction: The target eye primordium is used both for develops_from and > has_development_contribution_from which is redundant. One should be removed or one of the values should be replaced.

This is a more subtle issue but again I think this comes down to the tool being used. This is a hidden GCI pattern, which is a powerful axiom construct for inference. But most users should not need to worry about it, if the tool only shows the existential graph, the seemingly weaker axiom will not be there.

We clearly need to provide better guidelines for developers of portals and GUIs!

I recommend OLS

http://www.ebi.ac.uk/ols/ontologies/uberon/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FUBERON_0010312

unfortunately it does not show develops-from in the hierarchical view, but it cleanly separates the Equivalence axiom up top from the some-values-from relationships

For example develops_from ecto-epithelium?

We already have this path from eye primordium

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

lateral forebrain bundle:
Change restriction: Refine part_of with forebrain instead of CNS

Good catch. In fact there is a deeper problem here. This was part of a batch of terms transferred when we merged in the teleost anatomy ontology. The neuro terms here have not been vetted in a multi-species context.

In this particular case, the string "lateral forebrain bundle" is a very generic name for what the structure represents - a zebrafish structure. The way this is indicated is not clear. The term has an annotation that says: This class was sourced from an external ontology (teleost_anatomy). Its definitions, naming conventions and relationships may need to be checked for compatibility with uberon

The same label is used in other species, and work will need to be done unpicking homology vs synonymy

There is also a lateral/basal forebrain bundle in reptiles and birds http://braininfo.rprc.washington.edu/centraldirectory.aspx?ID=2465

There is also a structure with this name in lungfish

in xenopus (XAO:0004559)

as well as in mammals!

This one is so interesting I made a new dedicated issue: #1280

Member

cmungall commented Oct 15, 2016

lateral forebrain bundle:
Change restriction: Refine part_of with forebrain instead of CNS

Good catch. In fact there is a deeper problem here. This was part of a batch of terms transferred when we merged in the teleost anatomy ontology. The neuro terms here have not been vetted in a multi-species context.

In this particular case, the string "lateral forebrain bundle" is a very generic name for what the structure represents - a zebrafish structure. The way this is indicated is not clear. The term has an annotation that says: This class was sourced from an external ontology (teleost_anatomy). Its definitions, naming conventions and relationships may need to be checked for compatibility with uberon

The same label is used in other species, and work will need to be done unpicking homology vs synonymy

There is also a lateral/basal forebrain bundle in reptiles and birds http://braininfo.rprc.washington.edu/centraldirectory.aspx?ID=2465

There is also a structure with this name in lungfish

in xenopus (XAO:0004559)

as well as in mammals!

This one is so interesting I made a new dedicated issue: #1280

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

decussation of trochlear nerve
Remove restriction: The decussation is just anatomic fact that each trochlear nerve crosses from one side to the other, therefore I'm not sure that it is correct to apply extends_fibers_into

and

postganglionic sympathetic fiber
Change restriction: Might be my misunderstanding but this is post galglionic so it > extends_fibers_from and not extends_fibers_to. See definition.

This relation is declared symmetric, and is intended to be used in either direction for any case where a fiber connected to X overlaps with Y.

But point taken, the def must be clearer, and we should be using the more precise relations defined by @dosumis in Osumi-Sutherland, et al (2012). A strategy for building neuroanatomy ontologies. Bioinformatics (Oxford, England), 1–10. http://dx.doi.org/10.1093/bioinformatics/bts113

Member

cmungall commented Oct 15, 2016

decussation of trochlear nerve
Remove restriction: The decussation is just anatomic fact that each trochlear nerve crosses from one side to the other, therefore I'm not sure that it is correct to apply extends_fibers_into

and

postganglionic sympathetic fiber
Change restriction: Might be my misunderstanding but this is post galglionic so it > extends_fibers_from and not extends_fibers_to. See definition.

This relation is declared symmetric, and is intended to be used in either direction for any case where a fiber connected to X overlaps with Y.

But point taken, the def must be clearer, and we should be using the more precise relations defined by @dosumis in Osumi-Sutherland, et al (2012). A strategy for building neuroanatomy ontologies. Bioinformatics (Oxford, England), 1–10. http://dx.doi.org/10.1093/bioinformatics/bts113

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

presumptive paraxial mesoderm
To be consistent with presumptive endoderm this should be part_of blastula and not mesoderm. Or vice versa.

I agree, this is now fixed in the editors version. Thanks!

Member

cmungall commented Oct 15, 2016

presumptive paraxial mesoderm
To be consistent with presumptive endoderm this should be part_of blastula and not mesoderm. Or vice versa.

I agree, this is now fixed in the editors version. Thanks!

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

subthalamic fasciculus
The subthalamic fasciculus runs from the subthalamic nucleus to the globus pallidus so I'm not sure it is part_of ventral thalamus or is_a ventral thalamic fasciculus.

good. I removed the link to VT. The inferred link to VTF will disappear as a consequence

Member

cmungall commented Oct 15, 2016

subthalamic fasciculus
The subthalamic fasciculus runs from the subthalamic nucleus to the globus pallidus so I'm not sure it is part_of ventral thalamus or is_a ventral thalamic fasciculus.

good. I removed the link to VT. The inferred link to VTF will disappear as a consequence

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

postcommissural fornix of brain
Remove superclass: I think it is more correct to say that this is part_of the fornix than is_a and therefor the parent should be removed and added as a part_of relationship with the fornix of brain as the target.

I agree. It looks like this came from an early effort align with the FMA (which has it as is_a fornix of neuraxis, part-of fornix of forebrain)

This is now fixed, along with pre-commisural. Thanks!

Member

cmungall commented Oct 15, 2016

postcommissural fornix of brain
Remove superclass: I think it is more correct to say that this is part_of the fornix than is_a and therefor the parent should be removed and added as a part_of relationship with the fornix of brain as the target.

I agree. It looks like this came from an early effort align with the FMA (which has it as is_a fornix of neuraxis, part-of fornix of forebrain)

This is now fixed, along with pre-commisural. Thanks!

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

piriform cortex layer 2a
Add restriction: add/refine part_of with rhinencephalon and/or telencephalon

we have the (inferred) part-of to telencephalon already

can you clarify the rhinencephalon part? In uberon we use UBERON:0013201 ! olfactory pathway

is is just 2a specifically that links here, or the piriform cortex more generally?

Member

cmungall commented Oct 15, 2016

piriform cortex layer 2a
Add restriction: add/refine part_of with rhinencephalon and/or telencephalon

we have the (inferred) part-of to telencephalon already

can you clarify the rhinencephalon part? In uberon we use UBERON:0013201 ! olfactory pathway

is is just 2a specifically that links here, or the piriform cortex more generally?

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Oct 15, 2016

Member

Many thanks for these. I would like to give credit for this - this can happen in different ways - if you comment on this ticket then the credit will flow to the relevant terms in the ontology automatically, linked to your github (so you will get credit @ChrisOchs but I'd also like to extend this to the domain specialist that prepared these)

Alternatively, if you give me a name or ORCID I will add manually.

It's also clear we need a good way of browsing the ontology that doesn't give confusing redundancies. I recommend OLS as the best web tool for browsing uberon.

Member

cmungall commented Oct 15, 2016

Many thanks for these. I would like to give credit for this - this can happen in different ways - if you comment on this ticket then the credit will flow to the relevant terms in the ontology automatically, linked to your github (so you will get credit @ChrisOchs but I'd also like to extend this to the domain specialist that prepared these)

Alternatively, if you give me a name or ORCID I will add manually.

It's also clear we need a good way of browsing the ontology that doesn't give confusing redundancies. I recommend OLS as the best web tool for browsing uberon.

@cmungall cmungall self-assigned this Oct 15, 2016

@ChrisOchs

This comment has been minimized.

Show comment
Hide comment
@ChrisOchs

ChrisOchs Nov 1, 2016

Chris,

Thanks. I only posted these, our MD domain expert actually did all of the reviewing. So a acknowledgement to the Structural Analysis of Biomedical Ontologies Center at NJIT would be fine :)

Did you get a chance to look at #1243? That's the bigger list of changes we first reported.

I've forwarded these remarks to our domain expert for his feedback.

ChrisOchs commented Nov 1, 2016

Chris,

Thanks. I only posted these, our MD domain expert actually did all of the reviewing. So a acknowledgement to the Structural Analysis of Biomedical Ontologies Center at NJIT would be fine :)

Did you get a chance to look at #1243? That's the bigger list of changes we first reported.

I've forwarded these remarks to our domain expert for his feedback.

@ChrisOchs

This comment has been minimized.

Show comment
Hide comment
@ChrisOchs

ChrisOchs Nov 7, 2016

Re: anterodorsal nucleus of medial geniculate body

We have the same display as your figure. Our domain expert believes that the relationships that he mentioned were incorrect are incorrect.

Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.

Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

In regards to: piriform cortex layer 2a

we have the (inferred) part-of to telencephalon already

can you clarify the rhinencephalon part? In uberon we use UBERON:0013201 ! olfactory pathway

is is just 2a specifically that links here, or the piriform cortex more generally?

Was this changed in a more recent version? We don't see the inferred part of in Protege or our tool in the Jan 2016 Uberon release.

ChrisOchs commented Nov 7, 2016

Re: anterodorsal nucleus of medial geniculate body

We have the same display as your figure. Our domain expert believes that the relationships that he mentioned were incorrect are incorrect.

Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.

Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

In regards to: piriform cortex layer 2a

we have the (inferred) part-of to telencephalon already

can you clarify the rhinencephalon part? In uberon we use UBERON:0013201 ! olfactory pathway

is is just 2a specifically that links here, or the piriform cortex more generally?

Was this changed in a more recent version? We don't see the inferred part of in Protege or our tool in the Jan 2016 Uberon release.

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Nov 14, 2016

Member

Re: anterodorsal nucleus of medial geniculate body

We have the same display as your figure. Our domain expert believes that the relationships that he mentioned were incorrect are incorrect.

sorry, I'm not following. Are you saying:

  1. that you are confirming your belief these were incorrect
  2. that your initial report of incorrectness was incorrect?
Member

cmungall commented Nov 14, 2016

Re: anterodorsal nucleus of medial geniculate body

We have the same display as your figure. Our domain expert believes that the relationships that he mentioned were incorrect are incorrect.

sorry, I'm not following. Are you saying:

  1. that you are confirming your belief these were incorrect
  2. that your initial report of incorrectness was incorrect?
@ChrisOchs

This comment has been minimized.

Show comment
Hide comment
@ChrisOchs

ChrisOchs Nov 15, 2016

Sorry for the ambiguity.

Option 1. The domain expert believes that the relationships are incorrect.

ChrisOchs commented Nov 15, 2016

Sorry for the ambiguity.

Option 1. The domain expert believes that the relationships are incorrect.

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Nov 15, 2016

Member

which of the two relationships are incorrect?

Member

cmungall commented Nov 15, 2016

which of the two relationships are incorrect?

@ChrisOchs

This comment has been minimized.

Show comment
Hide comment
@ChrisOchs

ChrisOchs Nov 21, 2016

For the concept: anterodorsal nucleus of medial geniculate body

Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.

Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

ChrisOchs commented Nov 21, 2016

For the concept: anterodorsal nucleus of medial geniculate body

Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.

Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

@cmungall

This comment has been minimized.

Show comment
Hide comment
@cmungall

cmungall Nov 21, 2016

Member

For the concept: anterodorsal nucleus of medial geniculate body

Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.
Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

Please see my comments here, and the graph visualization:
#1277 (comment)

You are either looking at a very old version of the ontology, or there is an issue with how you are visualizing the relationships.

Member

cmungall commented Nov 21, 2016

For the concept: anterodorsal nucleus of medial geniculate body

Remove restriction: The inherited value part_of dorsal plus ventral thalamus is misleading. This is related only to the dorsal thalamus.
Change restriction: The value of part_of brain gray matter should be refined with gray matter of diencephalon

Please see my comments here, and the graph visualization:
#1277 (comment)

You are either looking at a very old version of the ontology, or there is an issue with how you are visualizing the relationships.

cmungall added a commit that referenced this issue Nov 24, 2016

Various fixes for #1277
> Brodmann (1909) area 33 UBERON:0006476
>
> Missing restriction: Part of restriction with a range of cingulate cortex

Fixed, thanks.

> distal tarsal bone pre-cartilage condensation UBERON:0015101
>
> There appear to be missing subclasses. I'm surprised there are no associated descendants such as distal tarsal bone 4 pre-cartilage condensation (this can also be stated as the missing subclasses are missing this class as a superclass).

Thank you. This can be traced back to UBERON:0015111 'distal tarsal bone 4 endochondral element', which was lacking specific parentage to UBERON:0015099 ! distal tarsal endochondral element. This has been fixed. The reasoner will now give the correct inferences for all members of this design pattern.

> femur endochondral element UBERON:0015052
>
> Incorrect superclass: Limb endochoindral element is theoretically a redundant superclass and should be the superclass of hindlimb endochondral element, but it is not.

The redundancy was caused by a incorrect logical definition of 'hindlimb endochondral element', this has been fixed.

> hypothenar muscle UBERON:0014377
>
> The three individual muscles involved are not in Uberon and not children of this concept. (Only) Extensor digiti minimi exists.

Thank you. These have been added, as well as  palmaris brevis, which FMA considers hypothenar

> lamina VII of gray matter of spinal cord UBERON:0016578
>
> Missing part of restriction with range ventral horn of spinal cord. See UBERON:0016579 for similarity.

Fixed, thank you.

> lamina VIII of gray matter of spinal cord UBERON:0016579
>
> Missing part of restriction with range ventral commissural nucleus of spinal cord.

Fixed, thank you.

> muscularis mucosa UBERON:0006676
>
> Incorrect restrictions: remove part_of digestive tract and part_of mucosa and replace with part_of gastrointestinal system mucosa

Replaced mucosa with GI system mucosa.

There is still work to do here. Formally, DT is more specific than GI system. However, I don't believe MM lines as part of the GI system that is not part of the DT?

There is also the issue of this class:

 [UBERON:0006912](http://purl.obolibrary.org/obo/UBERON_0006912) ! urinary bladder muscularis mucosa
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment