8 Feb 2022 PAINT Conference Call
Propagation of IBA with conflict NOT qualifier
IBAs are propagated to the leaf nodes even if there is experimental annotation with a conflict NOT qualifier to the same gene. Anushya created a list of nodes with conflict.
- Here is the file with nodes to be curated: https://drive.google.com/file/d/1btEkCOzoTERf-NBgN-7VBKUB86_y1unh/view?usp=sharing
- Here are the ones that have been curated with notes: https://drive.google.com/file/d/1fDfbSOXGsi_pAiXUj633nrTYpS9WiFpG/view?usp=sharing
- Here are some examples where IBAs are allowed to stay.
Differences in processing NOT qualifier in PAINT and literature curation.
Minutes A few topics were brought up and discussed.
- Most of the experimental annotations with NOT qualifier in FlyBase, Pombe and SGD seem to be correct, while many of them in MGD and human seem to be specific to isoforms. The discussion is whether to block the annotation from isoforms. The isoform information can be found in column 17 of the gaf file.
- It is also discussed whether only the NOT isoform annotation should be blocked or positive annotations should also be blocked.
- The evidence by isoform alone probably should be used as evidence in IBD propagation.
- In literature curation, a NOT annotation seems to mean the NOT to the term and also to the entire hierarchy. The correct way to do it is to put a NOT annotation to the most general GO term in the hierarchy. Probably not high priority. We need to deal with it case by case.
- Should IBA with conflict NOT qualifier be propagated and then recurated? It seems to be a lot of effort to review those conflicts.
Here are the decisions.
- IBA with conflict NOT qualifier should be allowed in PAINT.
- Curators can submit tickets if they don't agree, and we will review.
- We will look into how to use domain and active site information to automatically apply STOP propagation and avoid conflicts.
Prioritize families with human genes
Huaiyu created a google sheet with all families that are labeled "not curated" and have at least one human genes and one experimental annotation. Marc went through about 15 of them and found:
- 2 of them have no IBD in the tree and have experimental annotations.
- 4 of them have no IBD but only have binding MF which are not shown in the PAINT tool.
- The remaining all have IBDs already in the tree, probably from merge or split.
Marc asked if there is a way to prioritize them.
- The families that already have IBDs are from split or merge. Although they already have IBDs, they still need to be reviewed. However, they can be lower priority. These families often have PTHR IDs higher than 45xxx.
- The priority should be to curate families lower than PTHR45xxx first. We will continue to curate in PANTHER15.0 until these families are all curated. Then we will switch the PAINT to PANTHER 17.