"directly activates" and "directly inhibits"

[mah11]

OWL GAF checks flagged these, so I've removed them from the GAF that's gone to GO central today. Same issue as in #1720 - they simply don't allow catalytic MFs with IPI evidence. (In these cases I also think the info in the extensions is only really going to work well in GO-CAMs.)

I haven't changed the annotations in Canto.

GO_AR:0000007 IPI should not be used with catalytic activity molecular function terms Warning count:  2
GO_AR:0000007   Warning IPI should not be used with catalytic activity molecular function terms PomBase SPAC17H9.09c    ras1            GO:0003924      PMID:7923372    IPI     PomBase:SPAC16E8.09     F       GTPase Ras1     ste5    protein taxon:4896      20180429        PomBase directly_positively_regulates(PomBase:SPAC16E8.09),part_of(GO:0071963)
GO_AR:0000007   Warning IPI should not be used with catalytic activity molecular function terms PomBase SPBC1D7.05      byr2            GO:0004709      PMID:9315645    IPI     PomBase:SPAC1D4.13      F       MAP kinase kinase kinase Byr2   SPBC2F12.01|ste8        protein taxon:4896      20180429        PomBase directly_positively_regulates(PomBase:SPAC1D4.13),part_of(GO:0071507)

https://curation.pombase.org/pombe/curs/7537db502e936856/ https://curation.pombase.org/pombe/curs/c3f91b7bec8a34a1/

[Antonialock]

Ah, cheers, I've sorted those out.

[ValWood]

In these cases I also think the info in the extensions is only really going to work well in GO-CAMs

Which info in which extensions?

[Antonialock]

I changed them to EXP's as a stopgap until we have a better solution. I don't think the annotations are wrong.

[ValWood]

We know there is a problem with the relationship because it refers to the gene here and we want to refer to the activity, so we need a solution for that but there are tickets about this. Is there another issue that we don't know about?

[Antonialock]

I don't understand what the problem is? (aside from that I used IPI with catalytic MF?)

[ValWood]

We are using "positively directly regulates" and "negatively directly regulates" to mean regulates the activity of x, but what we are actually saying is "regulates the gene x" so, when we submit to GO we need to swicth these to has_substrate.

I hope this can be fixed with an additional relationship, but I can't get anyone to approve or add and I'm not even sure who needs to do what to where, so I kind of 'gave up' for a while. This is why we don't want to mention this relationship explicitly in the NAR manuscript.

Our users will know what it means within PomBase because where we have used it there is only one activity to regulate. We would need to specify if a gene product was multifunctional. As we mainly use this for single-function signalling pathway components where the same activity can alternately activate, or inhibit specific substrates.

If there is an additional issue here I'd like to know because we would need to deal with that too.

I think for the next curator meeting we should concentrate on "signalling pathway stuff"

I'll make a list ...

(use pom1 as an example)

[ValWood]

I can't find any open (or closed) tickets on the curation tracker about the "directly regulates" issue although I do have an e-mail folder about it.

We need to resolve this if we don't want to have to do it in Noctua....

[mah11]

1. The part_of(GO:BP) extensions pose no problem at all.

We are using "positively directly regulates" and "negatively directly regulates" to mean regulates the activity of x, but what we are actually saying is "regulates the gene x" so, when we submit to GO we need to switch these to has_substrate.

2. For the byr2 GO:0004709 annotation, this is basically it, although there is one aspect that Val's comment doesn't explicitly cover: computers can't do everything that humans can. As I understand it, reasoning won't infer a couple of things that we (and our users) get intuitively:

   • that the first gp regulates the second gp by modifying it (in other words, reasoning doesn't infer the connection that gp1-modifies-gp2 is the mechanism by which gp1 regulates gp2's activity)
   • which activity of the substrate is being regulated, even when the substrate has only one activity annotated

For the first point, a new extension relation for "regulates activity of by modifying" should do the trick. For the second, I'm not even 100% sure we get all the way there with a new relation; it needs the "nesting" that can only be done in GO-CAMs.

There's some discussion in https://github.com/geneontology/annotation_extensions/issues/63, but the comment thread there is lo-o-o-ng and the stuff about what we want to do is really buried. Good luck. If we want to renew efforts with GO to try to get new relations I can try to draft a new ticket.

3. There's another issue with the ras1 GO:0003924 annotation, in addition to the stuff noted above. For that one we can't even use has_substrate as our placeholder for GO, because the substrate of any GTPase activity is, wait for it, GTP. It would be utter nonsense to put a gp identifier in as a GTPase substrate, and reasoning could do nothing with it. This also means the proposed new regulates-by-modifying relation would be irrelevant and unhelpful.

Reasoning aside, I also find it a lot less intuitive to figure out what's going on in this case. How does hydrolyzing GTP regulate any activity of Scd1?

[ValWood]

Yep I think behind the scenes we will need to do something nested to connect these, but we don't want to plonk all of that part on the gene pages.

Let's leave this part until the curator meeting. My head is full right now. One of the exercises I wanted to try is to put a very small model into noctua (ie just 2 connected genes)...... and see what comes out....I haven't summoned the will to do this yet...

I don't know about the connection between ras1 and scd2, Antonia will need to answer that one

[ValWood]

gpad-from-cdc25-cdc2-model.txt

[ValWood]

fix https://curation.pombase.org/pombe/curs/05ebdc773cdd0a9d/feature/gene/view/7

[ValWood]

out of date