missing ISO annotatins in noctua annotation preview

[LiNiMGI]

For MGI gene model Zfp750 (MGI:MGI:2442210)(model ID: gomodel:653b0ce600001157 ), I noticed that:

annotation preview missing the ISO annotations:
Zap750
promoter-specific chromatin binding GO:1990841 ISO PMID:37115925 UniProtKB:Q32MQ0
Part of:
regulation of transcription by RNA polymerase II GO:0006357 ISO PMID:37115925 UniProtKB:Q32MQ0

Though, these ISO annotations were included in the export GAPD(GAP).

[balhoff]

@kltm doesn't the annotation preview load the GPAD export and then prettify it? Would this point to a problem in the annotation preview workbench?

• model: http://noctua.geneontology.org/editor/graph/gomodel:653b0ce600001157
• GPAD export (6 annotations): http://noctua.geneontology.org/download/gomodel:653b0ce600001157/gpad
• annotation preview (4 annotations): http://noctua.geneontology.org/workbench/annpreview/?model_id=gomodel:653b0ce600001157

[kltm]

@balhoff could be. Looking at the wire, two things are requested of m3Batch: the GPAD output (6 lines) and an id/label map.

Around this section of the table building (v/trivial), there is an explicit filter set: https://github.com/geneontology/noctua/blob/master/workbenches/annpreview/AnnPreview.js#L241, limiting lines to 12 cols.
I'd have to play a little to see if this is what is actually reducing lines (I'm having a little trouble simulating locally), but it seems like a likely candidate. That said, it is explicitly in there for the preview; @vanaukenk I don't suppose you have any recollection of this? Git blame and my notes are vague geneontology/noctua#437 .

[kltm]

Reading through that little bit of history again, I'm wondering if we really need this workbench still? The only reason, really, we don't just go with the GPAD is the labels, and the table view should have that now, right?

[kltm]

@balhoff Okay, I think there might be an issue in the GPAD output we're parsing from:

!gpa-version: 1.1
MGI	MGI:2442210	involved_in	GO:0006357	PMID:37115925	ECO:0000266	UniProtKB:Q32MQ0
		20231108	MGI		noctua-model-id=gomodel:653b0ce600001157|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693
MGI	MGI:2442210	acts_upstream_of_or_within	GO:0044091	PMID:37115925	ECO:0000315	MGI:MGI:7470773		20231108	MGI	occurs_in(GO:0001533)	noctua-model-id=gomodel:653b0ce600001157|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693
MGI	MGI:2442210	acts_upstream_of_or_within	GO:2000304	PMID:37115925	ECO:0000315	MGI:MGI:7470773		20231108	MGI	occurs_in(GO:0001533)	noctua-model-id=gomodel:653b0ce600001157|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693
MGI	MGI:2442210	enables	GO:1990841	PMID:37115925	ECO:0000266	UniProtKB:Q32MQ0
		20231108	MGI	part_of(GO:0006357),has_input(MGI:MGI:1915050),has_input(MGI:MGI:1917309),has_input(MGI:MGI:1921809),has_input(MGI:MGI:1927578),has_input(MGI:MGI:2156528)	noctua-model-id=gomodel:653b0ce600001157|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693
MGI	MGI:2442210	acts_upstream_of_or_within	GO:0061436	PMID:37115925	ECO:0000315	MGI:MGI:7470773		20231108	MGI		noctua-model-id=gomodel:653b0ce600001157|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693
MGI	MGI:2442210	acts_upstream_of_or_within	GO:0010628	PMID:37115925	ECO:0000315	MGI:MGI:7470773		20231108	MGI		noctua-model-id=gomodel:653b0ce600001157|model-state=production|contributor=https://orcid.org/0000-0002-9796-7693

Dropping into whitespace mode, there seems to be an erroneous newline after UniProtKB:Q32MQ0 and UniProtKB:Q32MQ0; dropping those two bad lines could account for the filtering.

[ukemi]

Hi @kltm. I think that curators use the workbench as a human-friendly way of viewing the annotations that are coming from their models, at least I do. I have also told MGI curators to do a sanity check if they are making 'true' GO-CAMs. Although I still think there are issues with the annotations that are being generated by causal models, the curators should still check. It might be nice to see if other groups use the annotation preview. Maybe check on an annotation call?

[balhoff]

Dropping into whitespace mode, there seems to be an erroneous newline after UniProtKB:Q32MQ0 and UniProtKB:Q32MQ0; dropping those two bad lines could account for the filtering.

@kltm I see what you mean—weird! I will try to figure out where those newlines are coming from.

[balhoff]

@kltm if you look in the OWL export you can see that those newlines were somehow put into the string literals: http://noctua.geneontology.org/download/gomodel:653b0ce600001157/owl

[balhoff]

@ukemi or @LiNiMGI could you edit those two with values to remove the newlines? I think you have to add a whole new evidence; it didn't seem directly editable to me. Or else would it be better for me to make an edit directly in the OWL during this week's Noctua downtime?

[ukemi]

@balhoff I'll do it.

[ukemi]

Done.

[vanaukenk]

Looks like that fixed it?

[kltm]

@vanaukenk Likely fixed, but maybe we should open a companion issue that no form action sends leading or trailing whitespace?