Closed pgaudet closed 5 years ago
Impacted groups:
Contributor (Assigned by) | Number of annotations |
---|---|
BHF-UCL | 6 |
CAFA | 1 |
ComplexPortal | 6 |
FlyBase | 7 |
MGI | 20 |
ParkinsonsUK-UCL | 2 |
PINC | 5 |
RGD | 14 |
UniProt | 82 |
ZFIN | 15 |
@RLovering OK @ggeorghiou @hattrill OK @slaulederkind @sabrinatoro @krchristie @bmeldal OK
Let me know if I can go ahead with the merge or if you'd like more time.
Thanks, Pascale
please give us a week or more to capture the information in the AE field before the merge
Thanks
Ruth
Sure no worries. (In any case everything is saved in the google doc). Pascale
Hi I just noticed that GO:0004887 thyroid hormone receptor activity has the parent GO:0038023 signaling receptor activity. I think this parent should be added to the GO:0004879 nuclear receptor activity term. What do you think?
Thanks
Ruth
OK all UCL annotations updated, ChEBI IDs added to AE field and also added new annotation using GO:0038023 signaling receptor activity + ChEBI in AE field if this wasn't implied by the ontology. But others might want to update their annotations before the merge? Ruth
I just noticed that GO:0004887 thyroid hormone receptor activity has the parent GO:0038023 signaling receptor activity. I think this parent should be added to the GO:0004879 nuclear receptor activity term. What do you think?
Yes, I added that back.
@ggeorghiou @hattrill @slaulederkind @sabrinatoro @krchristie @bmeldal
Do you need more time ? I'll merge that on Tuesday if I dont hear anything.
Pascale
@pgaudet Can you hold off until I have a chance to look? Should get to it soon. Thanks.
Doing them now. I have 23 complexes to fix in total (incl the historic ones that don't go in the GPAD). I will change them in our DB, but we don't use extensions.
done
Question: what relationship should I use in Noctua? Chebi# --directly activates--> nuclear receptor activity ? @pgaudet
@sabrinatoro Yes that seems right
@pgaudet It doesn't seem like "directly activates" is the correct relationship (I get an "inconsistent" error)
I'm done.
@hdrabkin Can you have a look for MGI ? @ggeorghiou @sylvainpoux Are you good with just a merge or do you want to add extensions ?
@sabrinatoro I'd rather have @vanaukenk's input on this one. I can do 'has input' but 'activated by' would be more precise, but it's not in RO, just in GO_REL (I think).
Sorry this is not very helpful!
Pascale
Regulation relations can only be used between processes, so directly_activates should fail the reasoning check. Unless there was some progress on sub-classifying participation at the recent RO meeting, I think 'has input' is the best we can do right now.
What we needed was to say 'x activity' activated_by CHEBI:xxxx
as described here: https://github.com/geneontology/annotation_extensions/blob/master/doc/activated_by.md
Yes, right now 'has input' is the best relation to use between the MF of the transcription factor and the chemical entity.
As @ukemi points out, the 'directly activates' relation describes a regulatory interaction between two processes so will give a reasoning error when used between a ChEBI term and a GO MF.
It would be helpful to get an update on any progress towards sub-classifying participants as discussed at the RO meeting. Perhaps @cmungall could report back from the meeting on the 11/5 ontology editors call?
A report from the RO meeting is on the agenda.
Great; thanks @ukemi !
Hi please could you provide a link to the agenda Thanks Ruth
To GO editors: Can I merge this ? Or do we need to wait for clarification on relations?
@sabrinatoro Please use 'has input' then :)
@pgaudet : will do! :-) I think it is ok if you continue with this merge (I'll get to it asap).
@pgaudet. @ukemi @vanaukenk Can we have "has input" + Chebi#? (I am still having some inconsistency error... Thanks!
Hmmm. What is your model?
oops, forgot to add this... http://noctua.geneontology.org/editor/graph/gomodel:5b91dbd100001025 (I also put this on Gitter for discussion)
@ukemi it is an issue with ChEBI roles being used as material entities.
Wow. The roles come back to bite us again. Good catch Jim! @sabrinatoro, can you replace thyroid hormone and androgen with bonafide chemicals?
@ukemi : It took some digging, but I was able to find the bonafide chemicals. That being said, this is a solution which is different from the one suggested in this case (using "androgen" and "thyroid hormone" in annotation extensions)... I don't know if this will have consequences in the future and/or whether it is something to worry about.
We really need to work out how to align annotation extension and noctua models. I have used the activated by 'ChEBI ID' and the noctua models will be using has_input. My concern about the use of 'has_input' is that I would have thought you will be using 'has_input' to specify the gene regulated by the nuclear receptor. How will the regulated gene and the activating protein be distinguised in cases where the annotated protein is transcription factor which is 'activated' by a protein rather than a chemical?
The complexity of the various ChEBI Ids could also cause more alignment issues. But this is an issue for all ChEBI IDs included in GO.
On yesterday's ontology call, @ukemi proposed that we annotate nuclear receptors like this: http://noctua.geneontology.org/editor/graph/gomodel:5b91dbd100001177?barista_token=v9idfze5v6yaqeork9wg#
(right @ukemi @vanaukenk ? )
@RLovering Aligning with protein2GO is certainly a concern - we just need to first agree on how to annotate...
Pascale
Yes. But of course we need to check how this works with property chains etc. It would be nice to have a general pattern for all receptors that works. These are essentially receptors where the linked activity is a transcription factor rather than a kinase for example. The problem with 'activated by' and many other column 16 relations is that they haven't been rigorously defined, particularly with how they fit with other relations.
@vanaukenk and @sabrinatoro and I played around with this a lot this afternoon. In the end we decided it was least confusing to just use has_input, but I think that we really want a short-cut relation to describe this type of participation.
BUT we need to define the short-cut relation very precisely with respect to how it relates to existing relations.
@ukemi So we don't use binding anymore ?
Hello,
The following terms describing specific nuclear receptors (transcription factors) will be merged into GO:0000981 DNA-binding transcription factor activity, RNA polymerase II-specific:
There is no action needed, unless groups want to capture the ligand using an annotation extension or a Noctua model. I have added the replacements here: https://docs.google.com/spreadsheets/d/19KvlqWegkVPXeWG1kvekH5HeDatimW68bGju-jski_0/edit#gid=0
Thanks, Pascale