3 Aug 2021 PAINT Conference Call
Present
Agenda
Taxon constraint family review
The validation program removed certain IBDs due to TCV. Each of us was assigned to review a few of these families (Huaiyu's email on June 7th, 2021).
Huaiyu's review - The IBD removals due to TCV seem to be correct in the cases reviewed. However, most of TCs should have "only_in" relations to the ancestors, meaning they would gain the function in a more recent species. Therefore, there should be a mechanism in the PAINT tool to check it. For example, in PTHR10161, the GO term "ossification (GO:0001503)" was annotated to PTN000017834 (Deuterostomia) with IBD. It is correct to remove it. However, the terms should be correct to a child node of Euteleostomi. We have many cases like this.
A file has been generated for each book in http://data.pantherdb.org/Paint/taxon_check/ with possible node that can be annotated for given term with IBD code.
PAINT software updates
Disabled functionality that prevents creation of IBD and TCV annotations due to taxonomy violation
Added functionality via drop down menu to view taxonomy violations for existing annotations in book
Future update - Upgrade Java version to 11
PAINT IBA generation change
- GH ticket: https://github.com/pantherdb/fullgo_paint_update/issues/52
- Switchover test comparison: https://docs.google.com/spreadsheets/d/1fRnyKQ798NSjnzekSJCNOZ9GCYpgkIcM-4Cc7kENEzo/edit#gid=1679685245
In order to align IBA propagation logic between the PAINT tool and our published IBA GAFs, we're proposing switching to a new IBA GAF generation method that exports IBA data directly from the PAINT tool. Testing results (linked above) look like there are some huge shifts in IBA annotation counts for certain files (e.g. SGD, ecocyc, human), though I'm still at an early stage of confirming these shifts are accurate (i.e. not due to a stupid bug). I'll dig up some example IBAs that changed using the new method before the meeting.
Example human IBAs that are not propagated by the PAINT tool:
- UniProtKB:A0AVT1 (PTN002475160) GO:0005737, from IBD PANTHER:PTN000102040 PTHR10953:AN0
- Due to UniProtKB:A0AVT1 having an IDA to GO:0005737. PAINT tool blocks IBA if IDA to same term.
- UniProtKB:Q9Y6F1 (PTN002470862) GO:0006302, from IBD PANTHER:PTN000396992 PTHR10459:AN0
- Same
- UniProtKB:O43614 (PTN002517462) GO:0016499, from IBD PANTHER:PTN002797346 PTHR24241:AN212
- Same
These IBAs will be dropped when we switch IBA GAF generation methods.
ACTION ITEM: Anushya will update PAINT propagator tool code to export these IBAs in this "IDA to same term" scenario and then regenerate XML. Dustin will then regenerate new GAFs and check the differences again.