search

pombase / mitochondrion

0 stars 0 forks source link

term matrix for mitochondrial processes (overannotation) #9

Open ValWood opened 2 weeks ago

ValWood commented 2 weeks ago

So far I have only looked at processes that are entirely mitochondrial:

pombe

Screenshot 2024-11-06 at 22 11 37

All species

Screenshot 2024-11-06 at 22 25 27

Most of the issues appear to be from causally upstream, and their propagation. We know exactly what these gene processes are involved in, causally upstream or within shouldn't be necessary?

@pgaudet @vanaukenk @hattrill

ValWood commented 2 weeks ago

current term list GO:0005739 GO:0032543 GO:0000963 GO:0006390 GO:0061668 GO:0006119 GO:0045239 GO:0016226 GO:0006744 GO:0097250

ValWood commented 2 weeks ago

Example issues https://www.uniprot.org/uniprotkb/Q96QE5/entry oxidative phosphorylation (challenged in protein2GO) disputed in protein2GO

https://uniprot-lume.netlify.app/uniprotkb/Q388R1/entry (I am not allowed to dispute these, oxidative phosphorylation should be removed, translation should be part_of, not causally upstream)

hattrill commented 2 weeks ago

@ValWood I think that some group went with the "causally upstream or within" as a default for the switch to gp2term rels as it is more likely to be "correct". We went with "involved in" as it is either right or wrong and I don't let "causally upstream or within" into FB because I wanted to prevent our curators from using it - just makes a bunch of trouble to fix at a later date.

In disputes, I just treat the "causally upstream or within" ones as if they were "involved in" as I think we should all work towards precision.

Perhaps we need a clearer policy on "causally upstream or within" - I am thinking that going forward, curators should not make new annotations with this gp2term rel and for groups to aim to eventually update to involved in or upstream of or remove annotations with this relation - just a thought (as it is no work for me, so easy for me to say!)

ValWood commented 2 weeks ago

Do you agree that this should be a policy @pgaudet @vanaukenk

A good first step would be to stop the transfer of causally_upstream altogether. If an annotation needs to be transferred, it should be changed to part_of.

There seem to be 2 schools of thought around causality_upstream. Initially, they were considered a 'placeholder' until the annotations could be reviewed and changed to part of', but in some cases, there is a reluctance to remove 'phenotypes' even when they are clearly indirect.

IMHO we shouldn't need to make, or keep "causally upstream or within" annotations to cellular processes if we know the process that the gene is part_of