Closed pgaudet closed 7 months ago
Are these all MF or BP+MF?
Anyway, we decided on our group call this morning that we will keep these relations locally (we have many more display labels that are even more specific).
Instead we will fix our GO submission file and map up to the has_input term.
There are multiple reasons for this decision, but mainly:
We like the more natural language these relationships present when we provide these annotations on our gene pages are much more human readable. We don't these all to be has_input https://www.pombase.org/gene/SPAC6G10.12c https://www.pombase.org/gene/SPBC29B5.01 https://www.pombase.org/gene/SPBC216.05
Originally I thought we could have a display label "has_direct_input" for "has_input". However, if we simultaneously map has_regulation_target to has_input we can no longer do this without configuring at multiple ontology subsets to disentangle the different meanings.
A gene is not really a "direct input" for a transcription factor (the transcription factor regulates the RNA polymerase which regulates expression of the gene through specific promoters). For this we would eventually like to display "regulates the expression of".
Although this might not be critical information for GO, apart from looking better on the gene page, and being more precise, we believe we will make use of this information when we are constructing networks, so we don't want to lose what we have arduously captured to date.
Anyway, we can easily make sure GO has the requested info by mapping our submission up to the less informative relations.
What about coactivator and corepressors? To distinguish between the partner and the regulated gene, we will have to use a different relation for the partner. I suspect this would work as an independent activity that would be regulated and also have the same input.
Exactly. The direct_input of the coactivator/repressor is the transcription factor. The genes are regulated_by, they aren't "direct input". We want to keep this specificity at PomBase. Of course, you can figure it out based on the term but since we want to display the more specific labels anyway, for us it would be more work to remove them, and then add them back in another way.
For coactivators and corepressors it was decided we should use has_input.
@mah11 will map Pombase ones to has_input for GO
MGI done.
dictyBase done
The UCL annotations are in Protein2GO, so I'll remove my assignment here. Thanks!
I agree with Barbara, The UCL annotations are in Protein2GO, so I'll remove my assignment here. and I have added Alex to this ticket
@alexsign Please change any remaining annotations in P2GO using 'has_regulation_target' to 'has input', and remove 'has_regulation_target' from the options in P2GO.
Thanks, Pascale
@pgaudet just a bit of stats before I start making changes. In case of external sources I cannot do much. Ones 'has_regulation_target' removed from go-plus.json it should disappear from the P2GO automatically.
UniProt (EBI Curators) internal 1349 MGI external 1095 AgBase internal 600 AlzheimersResearchUK-UCL internal 586 BHF-UCL internal 575 NTNU internal 539 ParkinsonsUK-UCL internal 504 DictyBase internal 172 SGD internal 122 Automated Function Prediction SIG CAFA Experiment (UniProt Curators) internal 107 FlyBase internal 43 GO_Central internal 10 WormBase internal 6 GOC OWL Inferences external 3 HGNC internal 3 Withheld Manual Annotation internal 2 Alzheimers Project at University of Toronto internal 2 UniProt (SIB Curators) internal 1
Thanks @alexsign So, only MGI needs to update their annotations. @ukemi Is this under way ?
Thanks, Pascale
I'm massively confused. I thought @dustine32 was doing this as part of the import.
I propose we re-visit this issue after the meeting.
We need to make sure we're all on the same page about what is supposed to happen when and where.
Hi Alex please change all the UCL annotation relations 'has_regulation_target' to 'has_input' Thanks Ruth @alexsign
@alexsign We'll want to do this for every source (this also applies to the other relations)
HI Alex, For us they have creeped in again. Please change dictyBase annotation extension relations also from 'has_regulation_target' to 'has_input' Many thanks! @alexsign
@pfey03 @RLovering @pgaudet Updated in Protein2GO, available from QuickGO next week.
Brilliant thanks Ruth
Thanks so much Alex!
@pgaudet Oh shoot. I just went to remove myself from this ticket and removed all assignees. IS there a way to add back the people who still need to do?
@ValWood this happened to me as well on another ticket a while back. Something strange with the git settings here.
@vanaukenk Can we go ahead and close this - and remove the 'has regulation target' from GOREL?
Not yet completed - there are still 40 annotations using 'has_regulation_target'; both MF and BP. I will ask Alex to do an automatic replacement to 'has input'
Note that the target needs to be a gene product.
Annotations are here: https://docs.google.com/spreadsheets/d/1jw635dKG2g7D71SftEOAR1HyKi7-gYZaA8dPBIMj_p4/edit#gid=0
dictyBase | 19 FlyBase | 3 UniProt | 18
@pgaudet for MF as well? P2Go has fixed all but not for MF when it's regulator protein? Not sure if I should change that?
Yes, we used 'has input' for these as well.
@pgaudet I'm done with changing all 20 Dicty remaining extension annotations
Hello,
To simplify the relations used for annotation extensions, we propose to use 'has_input' instead of 'has_regulation_target'.
Annotations are here: https://docs.google.com/spreadsheets/d/1MwJaCUrPwrhgINBWRtA4XDODaOINN8aerFI_LwozT_g/edit#gid=0
If you agree, and if your annotations are in Protein2GO, no action is needed: we will ask for an automatic mapping.
Thanks, Pascale