questions about tip pruning in synth pipeline

[Mark Holder]

Hi all,

I'm doing some work on the synth pipeline.

There is a step at which we prune off leaves of a tree if the OTT Id occurs multiple times in the tree or is nested inside another OTT ID that is mapped to a tip.

We have a flag (^ot:isTaxonExemplar) for indicating whether a particular node should be used as the exemplar of the taxon in the event that multiple tips share that same OTT ID.

There are a few cases in which it is not clear what the preferred behavior is:

  1. multiple nodes with the same OTT ID have this flag. I think that I'll just go to my next ranking criterion (lexicographic sorting of the node Ids) to resolve this.

  2. a tip mapped to a higher taxon and another tip mapped to a descendant of this taxon. I'm tempted to just prune off the higher taxon as ambiguous regardless of the exemplar flags.  If I don't do that, then:

    3. a higher taxon is flagged as the exemplar, but so is its descendant that is also in the tree. I think I'll delete the higher taxon in this case.

    4. a higher taxon is flagged as the exemplar, and at least one of it descendants is also flagged as an exemplar, but another one is not flagged. I think I'd lean toward including the descendants. For example, if a tree has a flagged tip mapped to Pan paniscus, an unflagged tip mapped to Pan trog., and a third tip mapped to Pan.

It is not clear that saying we can/should solve this by concluding: "these situations should be caught and fixed in the curator app before we get to synthesis". These situations can arise after mapping because of a subsequent change to OTT.

Thoughts welcome.

Mark

--

Mark Holder

mtho...@gmail.com

mtho...@ku.edu

http://phylo.bio.ku.edu/mark-holder

==============================================

Department of Ecology and Evolutionary Biology

University of Kansas

6031 Haworth Hall

1200 Sunnyside Avenue

Lawrence, Kansas 66045

lab phone:  785.864.5789

fax (shared): 785.864.5860

==============================================

[Jonathan A Rees]

Your proposed solutions favor lower (more specific) taxa over higher ones, when there is competition between them.  This is definitely a good policy