Weird display issue in AmiGO?

[pgaudet]

Hi,

GO:0140597 protein carrier activity has a is_a relationship with GO:0005344 oxygen carrier activity, but I dont see that in Protege nor in QuickGO - is this some bug?

AmiGO:

QuickGO:

Thanks, Pascale

[kltm]

Ref: http://amigo.geneontology.org/amigo/term/GO:0140597

[kltm]

@pgaudet Assuming the owltools loader isn't completely off, I think this is indeed what it's saying in it's own quirky way:
http://amigo.geneontology.org/visualize?format=png&term_data=GO%3A0140597&term_data_type=string&mode=amigo&inline=false
This collapsed-syle is from the perspective of the node in question "looking up" and asking how many steps things are away (tabbing) and the transitive relationship (relation). It's known that this is pretty weak for a lot of cases and there is discussion of better ontology viewers.

[cmungall]

I don't think this is about the display, its about the version of the ontology loaded in amigo. I assume either there is a version lag between quickgo and amigo or there is some axiom injected in the import chain The way to debug these is to open the exact same owl that is loaded in amigo in protege

…

On Mon, Aug 23, 2021, 18:32 kltm ***@***.***> wrote: @pgaudet <https://github.com/pgaudet> Assuming the owltools loader isn't completely off, I think this is indeed what it's saying in it's own quirky way: http://amigo.geneontology.org/visualize?format=png&term_data=GO%3A0140597&term_data_type=string&mode=amigo&inline=false This collapsed-syle is from the perspective of the node in question "looking up" and asking how many steps things are away (tabbing) and the transitive relationship (relation). It's known that this is pretty weak for a lot of cases and there is discussion of better ontology viewers. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#618 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAMMOIYRDAUIL3YGTLFKI3T6LZDVANCNFSM5CTORALA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .

[cmungall]

I'd hypothesize something like an older import of ro with an axiom such as transports o has_part -> transports causing the inference

…

On Mon, Aug 23, 2021 at 10:19 PM Chris Mungall ***@***.***> wrote: I don't think this is about the display, its about the version of the ontology loaded in amigo. I assume either there is a version lag between quickgo and amigo or there is some axiom injected in the import chain The way to debug these is to open the exact same owl that is loaded in amigo in protege On Mon, Aug 23, 2021, 18:32 kltm ***@***.***> wrote: > @pgaudet <https://github.com/pgaudet> Assuming the owltools loader isn't > completely off, I think this is indeed what it's saying in it's own quirky > way: > > http://amigo.geneontology.org/visualize?format=png&term_data=GO%3A0140597&term_data_type=string&mode=amigo&inline=false > This collapsed-syle is from the perspective of the node in question > "looking up" and asking how many steps things are away (tabbing) and the > transitive relationship (relation). It's known that this is pretty weak for > a lot of cases and there is discussion of better ontology viewers. > > — > You are receiving this because you are subscribed to this thread. > Reply to this email directly, view it on GitHub > <#618 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AAAMMOIYRDAUIL3YGTLFKI3T6LZDVANCNFSM5CTORALA> > . > Triage notifications on the go with GitHub Mobile for iOS > <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> > or Android > <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> > . >

[cmungall]

indeed curl -L -s http://purl.obolibrary.org/obo/go/extensions/go-gaf.owl > /tmp/go-gaf.owl robot reason -r elk -i /tmp/go-gaf.owl -o /tmp/go-gaf-reasoned.owl robot convert -i /tmp/go-gaf-reasoned.owl -o /tmp/go-gaf-reasoned.obo id: GO:0140597 name: protein carrier activity namespace: molecular_function def: "Binding to and carrying a protein between two different cellular locations by moving along with the target protein." [PMID:7628437] synonym: "protein carrier chaperone" EXACT [] synonym: "protein chaperone" BROAD [] is_a: GO:0005344 ! oxygen carrier activity intersection_of: GO:0140104 ! molecular carrier activity intersection_of: transports_or_maintains_localization_of PR:000000001 ! protein relationship: has_part GO:0005515 ! protein binding relationship: transports_or_maintains_localization_of PR:000000001 ! protein property_value: term_tracker_item geneontology/go-ontology#15917 xsd:anyURI

…

On Mon, Aug 23, 2021 at 10:26 PM Chris Mungall ***@***.***> wrote: I'd hypothesize something like an older import of ro with an axiom such as transports o has_part -> transports causing the inference On Mon, Aug 23, 2021 at 10:19 PM Chris Mungall ***@***.***> wrote: > I don't think this is about the display, its about the version of the > ontology loaded in amigo. I assume either there is a version lag between > quickgo and amigo or there is some axiom injected in the import chain > > The way to debug these is to open the exact same owl that is loaded in > amigo in protege > > On Mon, Aug 23, 2021, 18:32 kltm ***@***.***> wrote: > >> @pgaudet <https://github.com/pgaudet> Assuming the owltools loader >> isn't completely off, I think this is indeed what it's saying in it's own >> quirky way: >> >> http://amigo.geneontology.org/visualize?format=png&term_data=GO%3A0140597&term_data_type=string&mode=amigo&inline=false >> This collapsed-syle is from the perspective of the node in question >> "looking up" and asking how many steps things are away (tabbing) and the >> transitive relationship (relation). It's known that this is pretty weak for >> a lot of cases and there is discussion of better ontology viewers. >> >> — >> You are receiving this because you are subscribed to this thread. >> Reply to this email directly, view it on GitHub >> <#618 (comment)>, >> or unsubscribe >> <https://github.com/notifications/unsubscribe-auth/AAAMMOIYRDAUIL3YGTLFKI3T6LZDVANCNFSM5CTORALA> >> . >> Triage notifications on the go with GitHub Mobile for iOS >> <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> >> or Android >> <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> >> . >> >

[pgaudet]

Thanks @cmungall

We used to have that relation, but it was deleted back in March (see https://www.ebi.ac.uk/QuickGO/term/GO:0140597 - change log)

Why do we still have that relation?

Thanks, Pascale

[balhoff]

The relation is not in the current snapshot go-gaf. It is in the 2021-08-18 release of go-gaf, but not the 2021-08-18 release of go.owl. I think it must have lingered in go-gaf due to an import, but I can look more closely at it.

[balhoff]

I think it's two problems:

• go-gaf imports gorel, which is not rebuilt that often. It can have stale RO content.
• Even if gorel is updated (which it seems to have been in this case), there is a mistake in the catalog file. There is a location for imports/gorel.owl, instead of extensions/gorel.owl. Because of this, the go-gaf build will import gorel off the web instead of from the current source.
  • https://github.com/geneontology/go-ontology/blob/de92f261fc52efd8de5b0db80017202fe7f84535/src/ontology/catalog-v001.xml#L42