geneontology / helpdesk

The Gene Ontology Helpdesk
http://help.geneontology.org
16 stars 6 forks source link

Cycle in the go-basic.obo graph #489

Closed GuyTeichman closed 2 months ago

GuyTeichman commented 3 months ago

Describe the issue/bug

In the most recent version of the basic Gene Ontology graph (go-basic.obo, release 2024-06-10), it appears that a cycle has been introduced into the graph. In particular, the GO term GO:0045281 (succinate dehydrogenase complex) is pointing at GO:0045257 (succinate dehydrogenase complex (ubiquinone) ) ("part_of" relationship), while GO:0045257 is pointing back at GO:0045281 ("is_a" relationship). I believe that the "part_of" relationship was supposed to be deleted in one of the recent updates to the ontology, but was not deleted from the go-basic.obo file.

Expected behavior

The go-basic.obo graph should be guaranteed to have no cycles in it - any programs that attempt to propagate annotations using this ontology graph would run into infinite loops.

To Reproduce

Access the ontology through the permenant link: https://current.geneontology.org/ontology/go-basic.obo format-version: 1.2 data-version: releases/2024-06-10

Screenshots

[Term]
id: GO:0045257
name: succinate dehydrogenase complex (ubiquinone)
namespace: cellular_component
def: "The enzyme that catalyzes the oxidation of succinate and ubiquinone to fumarate and ubiquinol; involved in aerobic respiration, repressed in anaerobic respiration." [GOC:kd, ISBN:0198547684]
is_a: GO:0045239 ! tricarboxylic acid cycle heteromeric enzyme complex
is_a: GO:0045281 ! succinate dehydrogenase complex
[Term]
id: GO:0045281
name: succinate dehydrogenase complex
namespace: cellular_component
def: "A multimeric complex which consists of flavoprotein (subunit A ; InterPro:IPR003952), iron-sulfur protein (subunit B) and membrane-bound cytochrome b560 (subunit C; InterPro:IPR000701). In some Archaea, the membrane-bound subunits (C or C and D) do not necessarily contain heme. Membrane-bound subunits can bind or react with quinones." [GOC:kd, InterPro:IPR000701]
is_a: GO:0098803 ! respiratory chain complex
is_a: GO:1990204 ! oxidoreductase complex
relationship: part_of GO:0045257 ! succinate dehydrogenase complex (ubiquinone)
relationship: part_of GO:0045273 ! respiratory chain complex II
cmungall commented 3 months ago

Also reported here:

davidbio commented 3 months ago

Just stumbled over the same problem.

image

pgaudet commented 3 months ago

Sorry about that! We are making a new release that should fix this, we'll post this here when it's finished.

kltm commented 3 months ago

@pgaudet Has this been cleared?

pgaudet commented 3 months ago

Both terms reported in the issue (https://github.com/geneontology/helpdesk/issues/489#issue-2358104177) have been obsoleted. Alex had reported a few more - @alexsign do you still get errors regarding this?

davidbio commented 3 months ago

So this has been "fixed" by obsoleting the terms? Can we consider the latest version as "correct"?

pgaudet commented 2 months ago

@balhoff Do you have a way of testing this? @davidbio Do you still see that error?

balhoff commented 2 months ago

@pgaudet we have a check now that should fail the pipeline if this happens again.

pgaudet commented 2 months ago

Awesome!

davidbio commented 2 months ago

@pgaudet Honestly, I was waiting for this to be marked as RESOLVED before trying again.