Some relations in the Noctua GPAD are not being converted from labels to identifiers correctly

[ukemi]

As part of our QC, @LiNiMGI and I noticed that there are annotations not passing because they have 'invalid' relations in annotation extensions. Here is the list:

Invalid Relation in GO-Property GOREL:0000032 exists_during.
GOREL:0001019 results_in_division_of.
GOREL:0098790 regulates_translation_of. RO:0002232 has_end_location
RO:0002449 directly negatively regulates activity of.

In the Noctua GPAD output, relations are represented by their labels, but in the 'new' GPAD, they are represented by identifiers. I think most of them have been updated (or the policy for their usage has been updated) and the following changes need to be made: GOREL:0000032 exists_during. ---> RO:0002491 (existence starts and ends during) GOREL:0001019 results_in_division_of. ---> RO:0002233 (has_input-new policy and should be changed in the models? Although I can't find any documentation)
GOREL:0098790 regulates_translation_of. ---> deprecated (2 annotations from SynGO. see below) RO:0002232 has_end_location ---> This one is correct, but not in MGI. These extensions are the result of an inference that isn't very useful, but we could add it.
RO:0002449 directly negatively regulates activity of. ----> RO:0002630 (I think this is an incorrect translation of 'directly negatively regulates' See below).

[Typedef] id: regulates_translation_of name: obsolete regulates translation of def: "OBSOLETE. A relationship that holds between a process and a protein coding gene whose translation is regulated by that process." [] comment: Regulation of translation includes processes that regulate the levels of mature mRNA available for translation, as well as direct regulation of the rate of translation. comment: Usage of this relation has been deprecated in favor of 'has direct input'. subset: display_for_curators subset: valid_for_annotation_extension xref: GOREL:0098790 property_value: local_domain BFO:0000015 xsd:string property_value: local_range "PR:000000001 SO:0000673 SO:0000704" xsd:string property_value: usage "Relates a regulatory process or function to the gene whose translation it regulates." xsd:string {xref="GOC:dos"} is_obsolete: true

Annotation in the original Noctua file: MGI MGI:1345138 enables GO:0140678 PMID:15962011 ECO:0000315 20230504 MGI part_of(GO:0045742),directly_negatively_regulates(GO:0060090) contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:6446bfcb00001000|model-state=production

Annotation in Lori’s import file: Invalid Relation in GO-Property (11:Annotation_Extensions,12:Annotation_Properties): cannot find RO:,BFO: id: RO:0002449 MGI:MGI:1345138 RO:0002327 GO:0140678 PMID:15962011 ECO:0000315 2023-05-04 MGI BFO:0000050(GO:0045742),RO:0002449(GO:0060090) contributor=https://orcid.org/0000-0001-7476-6306|noctua-model-id=gomodel:6446bfcb00001000|model-state=production

[sierra-moxon]

from managers call: perhaps a question for @balhoff - is there a stale mapping somewhere in the ontology build? - @ukemi volunteers to discuss with @balhoff check ontobio for these IDs (maybe there is a mapping file somewhere that don't have the ids specifically)

MGI awesome curators found this, but its not MGI specific :)

[sierra-moxon]

sierra: follow up with a small group call: David, Li, Sierra, Jim, Kimberly, Seth to figure out the relations mapping

[balhoff]

@sierra-moxon I'm not sure this is exactly the issue or not, but GPADs out of Noctua use only these relations for extensions: https://github.com/geneontology/minerva/blob/df132bc5f1846c99735936f38058ee49fc339c84/minerva-converter/src/main/resources/org/geneontology/minerva/legacy/sparql/gpad-extensions.rq#L17

That list has been evolving based on feedback ever since it was created, so if there are some that need to be added there, it would be no problem.

[pgaudet]

@sierra-moxon

Li and I are looking into this again. For example for the first pair: GOREL:0000032 exists_during. ---> RO:0002491 (existence starts and ends during)

I don't find GOREL:0000032 in the 'conversion' file @balhoff pasted in the previous comment. We do see RO:0002491.

Do you know where these obsolete relations are coming from ? So we could update whatever file is injecting bad data. (see first comment for full list of incorrect relations coming out in the Noctua GPAD export).

Thanks, Pascale

[ukemi]

From the GOA mouse gaf: UniProtKB P46935 Nedd4 located_in GO:0043197 PMID:25505317 IDA C E3 ubiquitin-protein ligase NEDD4 Nedd4|Kiaa0093|Nedd-4|Nedd4-1|Nedd4a|Rpf1 protein taxon:10090 20160412 SynGO part_of(CL:0002608),exists_during(GO:0035235)

[pgaudet]

I see - this is a SynGO issue. Are all SynGO ? do you know?

[ukemi]

I don't know if they are all SynGO.

[LiNiMGI]

many of the MGI annotations are not passing, mostly due to RO:0002449 and RO:0002232. See David's example above for RO:0002449 -incorrect translation of 'directly negatively regulates' ...

[ukemi]

Note that we could add those to MGI, but I thought the plan was to replace or obsolete them. That would make adding them a bad idea on the MGI end. @vanaukenk

[ukemi]

We went ahead and added RO:0002232 has_end_location into MGI, since it seems to be still inferred. The fate is the hands of the GOC.