31 Aug 2021 PAINT Conference Call

From GO Wiki
Jump to navigation Jump to search

Present

Marc, Dustin, Anushya, Huaiyu, Peifen, Pascale, Paul

Agenda

PAINT software update (Anushya)

Updates on PAINT release (PAINT version 2021_08_23_1) 1. IBA is propagated to leaf nodes if they already have positive experimental annotations (e.g., IDA). 2. IBA is propagated to leaf nodes when they have conflict NOT annotations and the matrix will show a pink color to indicate the conflict. The paint curator can insert a block (IKR or IRD) to stop the propagation. Documented in https://github.com/pantherdb/db-PAINT section 'Rules for creating IBD annotations'.

The updated propagation code will be used to generate IBA files for PAINT release to GO.

Minutes There are three major concerns wrt #2 above:

  1. This will add a lot of extra work to the curators to make sure that IBA is not propagated to all leaf nodes with correct NOT annotations, especially when the tree is very large. As a result, some incorrectly IBAs maybe generated.
  2. Since the previous implementation automatically block the IBA to a conflict leaf node, they are not reviewed by the curator. With this new implementation will lead to error to those curated families. Retrofit will be necessary before releasing the data.
  3. There will be community push back on this when a large number of conflict annotation appear in GO.

Actions:

  1. Continue the current implementation of IBA propagation in the PAINT tool. To ensure that all possible conflict leaf annotations are reviewed before propagation, the following improvement is suggested. When an IBD is propagated, and if there are conflict leaf experimental annotations, they are listed in a dialog box. Curator can check if IBA should be propagated to any of them.
  2. We will gather all the curated families with potential conflict IBA propagation, and assess the amount of curation effort needed. This also helps to prioritize our review process.
  3. We will continue to use the current workflow to generate IBA (the perl script) until the above two are satisfactory.

Monthly IBA release update (Dustin)

  • After the PAINT tool propagator change to output IBAs even if an exp annotation to that leaf already exists, I regenerated the IBA GAFs and stats. These stats look much more like the traditional IBA generation process (perl):
  • Are these results acceptable for release?