Annotation Call December 9, 2014
- 1 Agenda
- 2 Discussion
Making positive annotation when there is a Not annotation (rama)
If a gene product has only Not annotations, there should still be a positive annotation (either to a different term or to the root node). Having just NOT annotations will filter out those gene from any analysis.
small conjugating enzyme ontology development (DavidH and Val)
1) The ontology edits are complete for all of the ubiquitin and other small conjugating enzymes. There are three parents: activating enzyme activity (E1), conjugating enzyme activity (E2) and ligase activity (E3)
2) Now all of the annotations for these need to be checked. We need a strategy for this re-annotation.
Checkpoint terms guidelines
Annotation to checkpoint terms is based partly on a priori knowledge that certain problems encountered by the cell lead to cell cycle arrest that is, at least in principle, reversible (provided that the problem encountered can be fixed). A checkpoint pathway comprises sensors which detect the problem, and signalling and effector molecules which function to negatively regulate a specific cell cycle transition.
Gene products should not be annotated to checkpoint terms based on mutations that result in activation of the checkpoint (i.e, cause cell cycle arrest). Such mutations cause a problem which is detected by the checkpoint sensor, and cell cycle arrest indicates that the checkpoint is functioning normally. This does not represent positive regulation of the checkpoint. The mutated gene itself is not necessarily part of the checkpoint in a normal cell (although some gene products are involved in both a cell cycle process and in a related checkpoint). For example, DNA replication defects due to mutations can result in DNA damage or stalled and collapsed replication forks, which can be recognised by a checkpoint sensor, but most gene products involved in DNA replication are not involved in the checkpoint. [add note about exception as an example?]
Gene products can be annotated to checkpoint terms when a mutation results in inactivation of the checkpoint. Such gene products may be the sensor (the first component of a checkpoint) or any component of the signalling pathway to the effector. Compromising the checkpoint means that mutations which would normally halt the cell cycle no longer do so.
Genes which are involved in correcting the problems detected by the checkpoint (e.g. DNA damage, spindle assembly defects) should not be annotated to the checkpoint terms, because they act downstream of the checkpoint itself. These genes should instead be annotated to "response to checkpoint x" terms, which are in a separate branch of the ontology.
Positive annotation with a NOT anotation
We reguest groups to check their NOT annotations and add a positive annotation to either the root node (after reviewing the literature) or some other positive term
small conjugating enzyme ontology development
- Please review annotations to GO:19787 and move them to one of its children terms
- GO:61650 maps to E2 and GO:61659 to E3. Please substrate details in col-16 for the E3 term.
- transferase and ligase are used by the community and GO project differently. this causes lot of problems when coming up with labels.
- GO:0019787 (small conjugating protein transferase activity)-Do we want to rename this 'ubiquitin-like protein transferase activity' even though one of its children is the ubiquitin enzyme?
- this is the term used in the literature, so yes, rename. Make sure the Process terms match with the MF term labels.
- GO:0008641 (small protein activating enzyme activity)- Do we want to rename this 'ubiquitin-like protein activating enzyme activity' as well?
- While reannotating it is useful to check the pfam domains. Pfam details from Val below.
E1’s All GO:0008641 small protein activating enzyme activity and children appear to belong to PF00899 (I didn’t come across any exceptions) This includes ubiquitin, NEDD8, Urm1, SUMO, Atg8 and Atg12 E2’s All GO:0061650 small conjugating protein conjugating enzyme activity and children appear to belong to PF00179 (this includes ubiquitin, Nedd8 and SUMO) OR PF03987 (Atg8/12) I was unable to establish the identity of the E2 for Urm1 (any ideas?). Note that there are E2’s where the catalytic site is mutated (e.g. Aktip), but they function in a heterodimer with an active conjugating enzyme and are required for the activity. Also, these family members sometimes appear to perform both the E2 and the E3 activity. E3’s There are quite a large number of (mostly related to HECT or RING finger) domains PF00632 HECT PF09814 HECT PF00097 C3HC4 ring-type zinc finger PF13639 Zinc finger, RING-type PF13923 Zinc finger, C3HC4 type (RING finger) PF13445 PF12906 RING-CH-type (March family) PF13920 Zinc finger, C3HC4 type (RING finger) PF12861 Anaphase-promoting complex subunit 11 PF12678 Zinc finger, RING-H2-type PF02207 Zinc finger, N-recognin This list isn’t exhaustive. This is the domains where a family member has been demonstrated to have E3 ligase activity in yeasts. There may be additional domains in higher eukaryotes, probably all of these Pfam Clan members: http://pfam.xfam.org/family/PF13920#tabview=tab2 are E3 type ligases.
- We decided to obsolete Ub-thiolesterease activity because that activity doesn't exist (EC # was deleted)
Jenkins GAF reports
- Midori and David mentioned that there is a problem with their GAF reports on Jenkins. They are mostly related to col-16 relationships. Chris and Heiko will look into it.
- All the rules are documented on the GOC website- http://geneontology.org/page/annotation-quality-control-checks
Godd bye Prudence
Prudence is moving on to ChEMBL project and this was her last GO call. Thank you for all your efforts and contributions Prudence. We will miss you, good luck.
Will be discussed next year.