obophenotype
Incorrect GCI relations generating cycles
The following GCI relations seem incorrect, and generate cycles:
[Term]
id: UBERON:0000074
name: renal glomerulus
is_a: UBERON:0005325 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! mesonephric glomerulus
[Term]
id: UBERON:0000045
name: ganglion
is_a: UBERON:0003338 {gci_filler="NCBITaxon:8353", gci_relation="part_of"} ! ganglion of peripheral nervous system
is_a: UBERON:0003338 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! ganglion of peripheral nervous system
[Term]
id: UBERON:0000955
name: brain
is_a: UBERON:0008823 {gci_filler="NCBITaxon:10090", gci_relation="part_of"} ! neural tube derived brain
is_a: UBERON:0008823 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! neural tube derived brain
is_a: UBERON:0008823 {gci_filler="NCBITaxon:8353", gci_relation="part_of"} ! neural tube derived brain
[Term]
id: UBERON:0001232
name: collecting duct of renal tubule
is_a: UBERON:0005319 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! mesonephric collecting duct
[Term]
id: UBERON:0001285
name: nephron
is_a: UBERON:0005322 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! mesonephric nephron
[Term]
id: UBERON:0002329
name: somite
is_a: UBERON:0005598 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! trunk somite
[Term]
id: UBERON:0004198
name: comma-shaped body
is_a: UBERON:0005328 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! mesonephric comma-shaped body
[Term]
id: UBERON:0004199
name: S-shaped body
is_a: UBERON:0005332 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! mesonephric S-shaped body
[Term]
id: UBERON:0004212
name: glomerular capillary
is_a: UBERON:0006183 {gci_filler="NCBITaxon:7954", gci_relation="part_of"} ! mesonephric glomerular capillary
Note that these relations are not always present in the species-specific ontology, e.g. UBERON:0000074 is_a UBERON:0005325 is not present in zfa.
These are actually all intended, and used in cases where the levels of the hierarchy make distinctions across taxa but are meaningless within: all somites in fish are trunk somites.
I'm not sure of the validity of the renal ones - it looks like the intent was to say all nephrons in fish are mesonephric nephrons - but could they not also be pronephric?
Technically a cycle with subclass is not a problem, say a and b are taxon-specific subclasses of the uberon classes above, if a ⊆ b and b ⊆ a then b=a, which is the intent here. But I can see how this would be problematic for your workflow. If this is the sum total of problematic GCIs then we can just get rid of them at least for the main release
Hmm, interesting. Is this documented somewhere? And why is it not done using EquivalentClasses with IntersectionOf (UBERON:xxx and part_of NCBITaxon:xxx)?
Don't remove them, I think I can quite easily filter them. I'm trying to find a better use than simply removing them.
I note here the new cycles I identified related to this issue, to make sure they are correct, and for reference for future updates of Bgee (it was very helpful for me to retrieve this issue):
Cycle:
UBERON:0002437 "cerebral hemisphere white matter" -
UBERON:0011299 "white matter of telencephalon" -
UBERON:0019261 "white matter of forebrain" -
UBERON:0003544 "brain white matter" -
UBERON:0003544 "brain white matter" -
UBERON:0002316 "white matter" -
UBERON:0002437 "cerebral hemisphere white matter"
[Term]
id: UBERON:0002316
name: white matter
is_a: UBERON:0002437 {gci_relation="part_of", gci_filler="NCBITaxon:9443"} ! cerebral hemisphere white matter
Cycle :
UBERON:0002371 "bone marrow" -
UBERON:0012429 "hemopoietic tissue" -
UBERON:0002371 "bone marrow"
[Term]
id: UBERON:0012429
name: hemopoietic tissue
relationship: part_of UBERON:0002371 {gci_relation="part_of", gci_filler="NCBITaxon:9606", source="FMA"} ! bone marrow
Cycle:
UBERON:2005073 "atrioventricular ring" -
UBERON:2005072 "endocardial ring" -
UBERON:2005073 "atrioventricular ring"
[Term]
id: UBERON:2005072
name: endocardial ring
relationship: part_of UBERON:2005073 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! atrioventricular ring
On 30 Nov 2016, at 9:36, fbastian wrote:
I note here the new cycles I identified related to this issue, to make sure they are correct, and for reference for future updates of Bgee (it was very helpful for me to retrieve this issue):
Cycle: UBERON:0002437 "cerebral hemisphere white matter" - UBERON:0011299 "white matter of telencephalon" - UBERON:0019261 "white matter of forebrain" - UBERON:0003544 "brain white matter" - UBERON:0003544 "brain white matter" - UBERON:0002316 "white matter" - UBERON:0002437 "cerebral hemisphere white matter" [Term] id: UBERON:0002316 name: white matter is_a: UBERON:0002437 {gci_relation="part_of", gci_filler="NCBITaxon:9443"} ! cerebral hemisphere white matter
This is obviously wrong. It says all white matter in primates is found in the cerebral hemisphere.
The source is the Allen primate brain atlas:
/ PBA:3999 ! brain po PBA:128011350 ! telencephalon po PBA:128011354 ! cerebral cortex po PBA:294021746 ! neocortex po PBA:294022044 ! white matter *** po PBA:10007 ! white matter of rostral cingulate cortex po PBA:10016 ! white matter of medial orbitofrontal cortex po PBA:10025 ! white matter of caudal orbitofrontal cortex po PBA:10038 ! white matter of V1 po PBA:10051 ! white matter of V2
It seems that the label for PBA:294022044 is deceptively broad, and it actually means neocortex white matter.
I will fix this, but the larger issue is whether to continue with this strategy for using external AOs for phylogenetic GCIs. We can revisit the strategy of doing a one-off dump and then curating.
Cycle : UBERON:0002371 "bone marrow" - UBERON:0012429 "hemopoietic tissue" - UBERON:0002371 "bone marrow"
[Term] id: UBERON:0012429 name: hemopoietic tissue relationship: part_of UBERON:0002371 {gci_relation="part_of", gci_filler="NCBITaxon:9606", source="FMA"} ! bone marrow
Here we are saying that in humans all hemopoietic tissue is found in the bone marrow. I think this is correct.
The problem is that bone marrow is itself classified at HT. I think HT has too broad an OWL def:
intersection_of: UBERON:0000479 ! tissue
intersection_of: contains_process GO:0030097 ! hemopoiesis
Cycle: UBERON:2005073 "atrioventricular ring" - UBERON:2005072 "endocardial ring" - UBERON:2005073 "atrioventricular ring"
[Term] id: UBERON:2005072 name: endocardial ring relationship: part_of UBERON:2005073 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! atrioventricular ring
will look at this one later
New ones:
[Term]
id: UBERON:0001083
name: myocardium of ventricle
is_a: UBERON:0018649 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! cardiac muscle tissue of ventricle
[Term]
id: UBERON:0002302
name: myocardium of atrium
is_a: UBERON:0004490 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! cardiac muscle tissue of atrium
[Term]
id: UBERON:0002343
name: abdomen musculature
is_a: UBERON:0002378 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! muscle of abdomen
[Term]
id: UBERON:0011362
name: cranial blood vasculature
is_a: UBERON:0003496 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! head blood vessel
[Term]
id: CL:0000103
name: bipolar neuron
is_a: CL:0000748 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! retinal bipolar neuron
[Term]
id: CL:0000287
name: eye photoreceptor cell
is_a: CL:0010009 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! camera-type eye photoreceptor cell
[Term]
id: CL:0000653
name: glomerular visceral epithelial cell
is_a: CL:0002523 {gci_relation="part_of", gci_filler="NCBITaxon:7954"} ! mesonephric glomerular visceral epithelial cell
1+2: I think these are OK?
3+4: there is a general issue here that uberon follows fma in making precise divisions between elements + stuff (e.g. vessels/muscles vs vasculature/musculature). other ontologies are a bit looser. This results in oddities like this. I'm afraid I don't have an immediate fix. It's not clear we want to keep the shadows in uberon long term; we may prefer to just keep the elements and add a special synonym for the mass form.
last 3: these are all formally correct, but it would be more straightforward if we simply merged these in the composite file
No problem, I report them all here for our pipeline, we can't accept cycles in Bgee for now. Other ones:
[Term]
id: UBERON:0000474
name: female reproductive system
is_a: UBERON:0000992 {gci_relation="part_of", gci_filler="NCBITaxon:6237"} ! ovary
[Term]
id: UBERON:0000079
name: male reproductive system
is_a: UBERON:0000473 {gci_relation="part_of", gci_filler="NCBITaxon:6237"} ! testis
[Term]
id: UBERON:0006337
name: distal early tubule
is_a: UBERON:0007308 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! pronephric distal early tubule
[Term]
id: UBERON:0001705
name: nail
is_a: UBERON:0009567 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! nail of pedal digit
[Term]
id: UBERON:0002481
name: bone tissue
is_a: UBERON:4000115 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! mineralized bone tissue
[Term]
id: UBERON:0002099
name: cardiac septum
is_a: UBERON:0004142 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! outflow tract septum
Also, we have
[Term]
id: UBERON:0001135
name: smooth muscle tissue
is_a: WBbt:0005780 {gci_relation="part_of", gci_filler="NCBITaxon:6237"} ! non-striated muscle (C elegans)
While I think WBbt:0005780 should be is_a UBERON:0001135. The current relation seems incorrect:
[Term]
id: WBbt:0005780
name: non-striated muscle (C elegans)
is_a: CL:0000187 ! muscle cell
Also the following relation seems incorrect:
[Term]
id: UBERON:0001135
name: smooth muscle tissue
is_a: WBbt:0007810 {gci_relation="part_of", gci_filler="NCBITaxon:6237"} ! body muscle cell (C elegans)
On Sat, Mar 27, 2021 at 7:26 AM fbastian @.***> wrote:
No problem, I report them all here for our pipeline, we can't accept cycles in Bgee for now. Other ones:
[Term] id: UBERON:0000474 name: female reproductive system is_a: UBERON:0000992 {gci_relation="part_of", gci_filler="NCBITaxon:6237"} ! ovary
[Term] id: UBERON:0000079 name: male reproductive system is_a: UBERON:0000473 {gci_relation="part_of", gci_filler="NCBITaxon:6237"} ! testis
WBbt mappings will fix
[Term] id: UBERON:0006337 name: distal early tubule is_a: UBERON:0007308 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! pronephric distal early tubule
formally correct, these are equivalent in frogs (which don't progress beyond pronephros, AFAICR)
[Term] id: UBERON:0001705 name: nail is_a: UBERON:0009567 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! nail of pedal digit
will fix
[Term] id: UBERON:0002481 name: bone tissue is_a: UBERON:4000115 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! mineralized bone tissue
I think we need to merge the ext edit file in first then fix this, Nico is on it
it's not clear why this subclass was added for ext in the first place
[Term] id: UBERON:0002099 name: cardiac septum is_a: UBERON:0004142 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! outflow tract septum
fixed in PR
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/obophenotype/uberon/issues/651#issuecomment-808741193, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAGZRCGGRSYMFXMZ24EBDYDTFXTH5ANCNFSM4A43Y5MQ .
[Term] id: UBERON:0002481 name: bone tissue is_a: UBERON:4000115 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! mineralized bone tissue
I think we need to merge the ext edit file in first then fix this, Nico is on it
it's not clear why this subclass was added for ext in the first place
Tech call: Phenoscape external were merged into uberon-edit a long time ago, don't see anymore external axioms to merge into file, what other file needs to be added to edit file?
Chris here. Sorry, I’m lost. Why are we talking about merging the ext file? This was done a long time ago.
I think this axiom is an example of something that is biologically correct, yet encoded in a way that is too clever and too confusing, with no real use case. The pattern that should have been followed is a taxon pato characteristic gci. We should forbid isa gcis. But just delete the axiom
On Mon, Sep 5, 2022 at 6:58 AM Shawn Tan @.***> wrote:
[Term] id: UBERON:0002481 name: bone tissue is_a: UBERON:4000115 {gci_relation="part_of", gci_filler="NCBITaxon:8353"} ! mineralized bone tissue
I think we need to merge the ext edit file in first then fix this, Nico is on it
it's not clear why this subclass was added for ext in the first place
Tech call: Phenoscape external were merged into uberon-edit a long time ago, don't see anymore external axioms to merge into file, what other file needs to be added to edit file?
— Reply to this email directly, view it on GitHub https://github.com/obophenotype/uberon/issues/651#issuecomment-1237079101, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAGZRCDX3Z7BN4UZGHZ34SLV4X4BTANCNFSM4A43Y5MQ . You are receiving this because you commented.Message ID: @.***>
Hi @cmungall - I think the tech team was a bit confused as to what needs to be done here? The GCI doesn't seem to be in uberon-edit file so if nothing else needs to be done we can close it? Thanks