bug for stage merging in last composite-metazoan

[ANiknejad]

Hi Chris,

even if xref are present for stages in Uberon, no merge applied.
example for both xref and original class:

Line 140100: id: HsapDv:0000001
Line 154855: xref: HsapDv:0000001

Thank you

Anne

[fbastian]

And more importantly, FBdv terms that have an xref in Uberon are not merged at all, e.g., FBdv:00000000, FBdv:00005317, etc.

[ANiknejad]

the problem is still present in the composite 28 August:

[Term]
id: RnorDv:0000011
intersection_of: UBERON:0007220 ! late embryonic stage
intersection_of: part_of NCBITaxon:10116 ! Rattus norvegicus
relationship: only_in_taxon NCBITaxon:10116 ! Rattus norvegicus

[ANiknejad]

Hi Chris,

below another problem that also sounds to be a bug in composite-metazoan:
the class below is displayed as a class AND as a xref

Thank you for taking care of that as well!
All the best,

Anne

---

[Term]
id: MmusDv:0000007
name: blastula stage (mouse)
namespace: mouse_developmental_stage
def: "Embryonic stage that is an early stage of embryonic development in animals and is produced by cleavage of a fertilized ovum, with formation of a central fluid-filled cavity called the blastocoel." [UBERON:0000108]
comment: The blastula follows the morula and precedes the gastrula in the developmental sequence.
is_a: UBERON:0000067 ! embryo stage part
is_a: UBERON:0000105 ! life cycle stage
relationship: only_in_taxon NCBITaxon:10090 ! Mus musculus
relationship: part_of MmusDv:0000002 ! embryonic mouse stage (mouse)

[fbastian]

More generally, I have a feeling that there are two concurrent imports of HsapDv and MmusDv, one that existed before the developmental-stage repository and that is now outdated, and one that is based on the current version of the repository, using the merged -uberon ontologies.

Am I wrong?

[ANiknejad]

and also (sorry if I am mixing bugs!):
this term has two part_of relationships in the composite-metazoan,
while HsapDv:0000093 is correctly xref of UBERON:0007222 in in the same composite-metazoan file...

id: HsapDv:0000094
..
relationship: part_of HsapDv:0000093 ! human aged stage (human)
relationship: part_of UBERON:0007222 ! late adult stage

[cmungall]

Good news: latest release fixes HsapDv:0000001

Bad news: some classes are losing their axioms in composite-metazoan.obo, for example, 'late adult stage' loses it's part_of axiom (it retains a taxon GCI part_of, but not the generic part_of to post-juvenile).

This appears to be in the owl2obo conversion. The owl is fine.

[cmungall]

See the above commit to the Makefile for what was going wrong - after translating from owl2obo, we were running the oboedit obo2obo to roundtrip the file. Not exactly sure why. I changed this so that oe obo2obo is only used as a validation step (e.g. to see that no artefacts OE cannot handle escape).

[cmungall]

This is also now fixed, all should be good in the composite metazoan world now:

[Term]
id: HsapDv:0000093
name: human aged stage (human)
namespace: human_developmental_stage
def: "Late adulthood that refers to an adult who is over 65." [Bgee:curator]
synonym: "65+ years" EXACT []
synonym: "elderly" RELATED []
is_a: UBERON:0000105 ! life cycle stage
relationship: immediately_preceded_by HsapDv:0000092 ! human middle aged stage (human)
relationship: only_in_taxon NCBITaxon:9606 ! Homo sapiens
relationship: part_of HsapDv:0000091 ! human late adulthood stage (human)
property_value: start_ypb "65.0" xsd:float

[ANiknejad]

Hi Chris,

I am sorry to still comment this ticket about bug. I checked the last version (2014-09-19) both for
http://purl.obolibrary.org/obo/uberon/composite-metazoan.obo
and
http://purl.obolibrary.org/obo/uberon/composite-metazoan.owl

and in both
RnorDv:0000005 gastrula stage
and
RnorDv:0000007 neurula stage

are completely missing ??????

[cmungall]

When do you need this by? Is this a blocker for your release? Can you just use the bridge files such as rnordv-bridge-to-uberon (imstead of using xrefs)?

It looks like I made a mistake in speccing out the development stage ontology derived files for Frederic. There isn't a way to get what you want using the files as is. You want RnorDv:0000005 'gastrula stage' to be present as an xref for the generic uberon gastrula stage. To do this, we need yet another derived file, SPCDv-xref.obo with entries like

id: UBERON:0000109 ! gastrula stage
xref: RnorDv:0000005

I can then merge this in.

Should not be hard to do. Just more plumbing... can I do this next week?

I would like to move away from reliance on xrefs. And also, is there a new for a class like RnorDv:0000005 rat gastrula stage to exist? Do you annotate your data to these species-specific time-generic IDs? Can we just use Uberon IDs directly throughout here?

[fbastian]

• This is not a blocker, we have derived a custom life-stages ontology that we can currently use. We are fixing things for the next release.
• I have no problem with not relying on xrefs anymore. I could use the ECA to IntersectionOf(Uberon_stage, po some species). Xrefs are just convenient.

is there a new for a class like RnorDv:0000005 rat gastrula stage to exist?

Well, the idea was to develop the species-specific stage ontologies independently, and then to merge them with Uberon using your scripts. This is how we did it. We could change this for the future.

[cmungall]

OK, I think we're good. Latest release will be public midnight. xrefs should be present in all cases, everything should look good in general. Some issues we already know about e.g. Dmel, dev stage vs life cycle stage. Also Medaka is lacking grouping

[ANiknejad]

hi Chris,

we found that in release uberon/releases/2014-09-23
yes...... in .obo!

[Term]
id: EHDAA2:0004523
name: heart associated veins (human)
is_a: HsapDv:0000027 ! CS20 (human)
is_a: UBERON:0000480 ! anatomical group
is_a: UBERON:0002050 ! embryonic structure
relationship: existence_starts_during_or_after HsapDv:0000016 ! CS09 (human)
relationship: part_of EHDAA2:0004144 ! heart-associated blood vessels (human)
relationship: part_of EHDAA2:0004543 ! body veins (human)

!!!!! is_a: HsapDv:0000027 ! CS20 (human)

[cmungall]

Already reported here: obophenotype/human-developmental-anatomy-ontology#15

[cmungall]

Apologies but I just realized the heart associated veins issue is back in the 2015-01-01 release. I will check this out...