Annotation Conf. Call 2018-08-14
From GO Wiki
Montreal Meeting, October Wednesday, 17th - Friday, 19th
GO Conference Calls
- Still will be Tuesdays at 8am PDT
- Proposed meeting schedule:
- 1st Tuesday: Alliance Biological Function
- 2nd Tuesday: GO Consortium
- 3rd Tuesday: Alliance Biological Function/GO-CAM Working Group
- 4th Tuesday: GO-CAM Working Group
- 5th Tuesday: ad hoc, as needed
- One Zoom URL for all - https://stanford.zoom.us/j/976175422
- Automatic deletion of ND-evidenced annotations
- ND annotations to a root node term are intended to signify that the curator looked at all available evidence and the MF, BP, or CC of a gene/gene product is not known
- GOA proposed to remove ND annotations automatically from their curation database because this was not happening manually
- Note that curators can only update or remove annotations from their respective groups in Protein2GO
- Annotation from only a select set of evidence codes would result in automatic removal of ND annotations
ECO:0000269 (EXP) and its descendants
ECO:0006056 (HTP) and its descendants
ECO:0000250 (ISS) and its descendants
ECO:0000317 (IGC) and its descendants
- Generally agreed that when there is experimental or sequence-based evidence for a more granular annotation, the ND annotation should be removed
- However, 'protein binding' MF annotations are thought of differently by different groups and we need to reach a consensus about how these MF annotations should be considered wrt automatically removing existing MF ND annotations
- Proposal: as part of QC pipeline, alert groups to genes/gene products that have an ND annotation and also an annotation to a more granular term (any evidence code?)
- Refresher on how PAINT annotations are created
- Doubled up IBA+EXP annotations (from Karen Christie)
- PAINT:circular transfer of annotations
- 84 open annotation review tickets
- Reminder to check go-annotation tracker tickets where your group has been assigned and finish up where you can
- Questions? Please add questions/comments to the individual tickets
- We can discuss questions on future calls, if needed
- On call: David, Edith, George, Harold, Helen, Karen, Kimberly, Laurent-Philippe, Li, Liz, Marie-Claire, Michele, Midori, Pascale, Paul T., Sabrina, Shur-Jen, Seth, Stacia, Suzi A, Suzi L, Tanya, Tom, Stacia, Val
- We discussed the proposal to automatically remove ND annotations from the GOA database when experimental annotations from another source are available.
- Generally, removing ND annotations once a suitable, more granular term can be assigned to a gene/gene product is standard practice.
- However, not all ND annotations in Protein2GO can be edited by all groups and having an automated mechanism for dealing with ND annotations that are no longer applicable is desirable.
- One sticking point, however, has been what constitutes an appropriate Molecular Function annotation for removing NDs to the root MF node, specifically wrt child terms in the 'binding' (GO:0005488) hierarchy of MF.
- Some groups do not consider annotations to 'protein binding' (GO:0005515) sufficient to otherwise remove an ND annotation to root MF; other groups do.
- Annotations to child terms of 'protein binding' (GO:0005515) are viewed differently, however, by some groups who feel that the more granular protein binding terms provide useful information.
- Some groups also view annotations to terms like 'ATP binding' (GO:0005524) suitable for supplanting an ND annotation to root MF.
- We discussed various options around this issue, e.g. handling the potentially contradictory information at the level of display, allowing binding annotations to supplant ND annotations, moving binding out from under MF and having it as a separate ontology.
- For now, the automatic removal of ND annotations in Protein2GO is on hold.
- ACTION ITEM: Kimberly will come up with more concrete proposals for the different options and we will discuss them and pick the best option.
- We also discussed whether it is okay to include IBA annotations derived from PAINT curation when an experimental annotation that was used to create the IBA annotation exists for that gene product.
- Are these annotations circular or do they provide an additional level of QC that should also be captured?
- Note that during PAINT curation, propagation of an annotation to a given node does not require that there is experimental evidence for more than one leaf node, i.e. annotations may be propagated based on one experimentally supported annotation and a review of the overall conservation of family members.
- If we consider propagation from one experimental annotation plus family review a good form of QC, then do other forms of QC warrant inclusion in our annotation set and if so, what would the evidence code be?
- Should we just not allow propagation of annotations using IBA when there already exists an experimental annotation?
- We didn't reach a resolution on this yet, but further discussion is on Doubled up IBA+EXP annotations (from Karen Christie)