Annotation Call September 9, 2014
Present: Rama, Rachael, Judy, Suzi, Li, Kimberly, Petra, Paola, Aleks, Jane, Susan T, becky, Moni, Val, Diane
Upcoming GOC meeting
We are collecting agenda items for the next meeting. Please send us any items that you would like discussed.
Recent SF Jambnoree, Jenkins Reports
- FYI- Ontology editors went through all the SF items and resolved most of them.
Please check the reports for your GAF file and make efforts to fix them. Please pay attention to the do_no_annotate tag when you are annotating.
new feature in Jenkins report (Val)
In the recent past we talked about reporting loss of annotation specificity when parentages are changed/removed. I realise that not all databases have the resources to reannotate in these situations, but we would like to retain the originally intended annotation specificity in PomBase.
A recent (ish) example is the term "protein insertion into membrane raft". http://www.ebi.ac.uk/QuickGO/GTerm?id=GO:0071210 Although I remember this change happening, I now can't find any related SF ticket with a search on "raft"? but basically I believe it was because the membrane raft turned out to be located at non plasma membrane locations, (although I cant figure out what these other locations were from the existing terms and defs?) The result was that "GO:0090002 establishment of protein localization to plasma membrane" parentage, and hence the association with this term and "plasma membrane organization" were lost. However, all of our existing annotations referred to "plasma membrane insertion".
There are probably not a huge number of ontology fixes of this type, but enough that over the years this can add up to a gradual attrition of intended annotation specificity, especially in areas subjected to large annotation overhauls.
It would be a huge benefit to keeping the curation current if we could have a Jenkins report for the list of affected terms affected by parentage changes which resulted in an existing term (with annotations) becoming less specific. Individual groups could then choose whether or not to reannotate, but annotation fixes would be relatively quick and easy (we would only need to transfer annotations in question to a more specific term). Would it be possible to add this at some point soon?
"Binding" term (GO:0005488)
The term "binding" GO:0005488 has been put on the do_not_manually_annotate list, therefore all annotations to this term should be reviewed.
There are 27 annotations direct to this term and all of them have ChEBI IDs in the with field, therefore in most cases a more granular term should be used or created.
Some of the CHEBI IDs are not found and some seem unsuitable for GO (e.g. strychnine).
These are the unique ChEBI IDs used and in the right hand column are potential GO terms that could be used for some of them;
|CHEBI ID||name||Possible GO term|
|CHEBI:15367||all-trans-retinoic acid||GO:retinoic acid binding|
|CHEBI:33217||3,3',5,5'-tetrabromobisphenol A||GO:organic cyclic compound binding|
|CHEBI:49809||2,2',5,5'-tetrachloro-4,4'-bis(methylsulfonyl)biphenyl||GO:organic cyclic compound binding|
|CHEBI:8925||ryanodine||GO:organic cyclic compound binding?|
The distribution of annotations is;
- Rama will check with Heiko/Chris about empty GAF files. Also document how to run those checks from command line.
Please move the binding term to an appropriate granular term as shown in the table above. Some chemicals are not physiologically relevant (e.g.28973) and hence are suitable to move to granular term.
Ub/Ub-like functions (David H)
1) We need annotator feedback on the fate of 'ubiquitin thiolesterase activity' (GO:0004221). This term can either be obsoleted or merged into 'ubiquitinyl hydrolase activity' (GO:0036459). There has been an observation that many of the annotations to (GO:0004221) could actually be transferred to 'ubiquitin-specific protease activity' (GO:0004843). Annotators will need to reexamine the annotations to GO:0004221 and reannotate to either GO:0036459 or GO:0004843. See SF items:#11094 and #10519
2) Moving along with the E1,E2, E3 terms, GO:0019787 small conjugating protein ligase activity suffers from the same issues as the original ubiquitin ligase terms. We need to revamp the MF ontology for the other small proteins in the same way that we did for the ubiquitin activating, ligase and conjugating enzymes. Looking at annotations to GO:0019787, it appears that all three types of enzyme activity have been annotated using this term. We need to go back and reexamine those annotations once this part of the ontology has been cleaned up. See SF item #11026