[Treebase-guts] [ treebase-Bugs-2970700 ] Analysis download scrambles taxon labels

[SourceForge.net <no...@so...>]

Bugs item #2970700, was opened at 2010-03-15 11:29
Message generated for change (Tracker Item Submitted) made by sfrgpiel
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=1126676&aid=2970700&group_id=248804

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: ui
Group: None
Status: Open
Priority: 9
Private: No
Submitted By: William Piel (sfrgpiel)
Assigned to: Nobody/Anonymous (nobody)
Summary: Analysis download scrambles taxon labels

Initial Comment:
Analysis download scrambles taxon labels

Analysis download creates a taxon block in which the order of the names does not match with the order of the names in the associated character block. This will cause software programs like PAUP and Mesquite to scramble the taxon labels with the characters that they are supposed to map to. 

For example, see here:

http://treebase-stage.nescent.org/treebase-web/search/downloadAnAnalysisStep.html?analysisid=5539&id=10232

By contrast, if I download the matrix alone, a taxon block with the correct order is used:

http://treebase-stage.nescent.org/treebase-web/phylows/matrix/TB2:M4847?format=nexus

And if I download the *whole* study as nexus, the correct order is used:

http://treebase-stage.nescent.org/treebase-web/phylows/study/TB2:S10232?format=nexus

So the problem is in the analysis download. It looks as though the analysis download is re-creating new taxon blocks on the fly, whereas the matrix download is fetching them from the "taxonlabelset" and "taxonlabelset_taxonlabel" tables. 

A quick fix for the analysis download is the following: create taxon blocks using the information in taxonlabelset and taxonlabelset_taxonlabel tables. If any resulting trees come from a tree block that shares the same taxonlabelset_id as a matrix, then the resulting tree should have a "LINK TAXA" command that points to the *same* taxon block as the matrix. i.e. the number of taxa blocks that are generated at the top of the nexus file are determined by the number of unique taxonlabelset_id values linked to all character blocks and all trees/treeblocks linked to the analysis. 

The long-term fix is more complicated, namely auto-generate a single taxon block for a particular analysis download, use "LINK TAXA" to have all character blocks and tree blocks refer back to this same taxon block, but within each character block insure that the rows are sorted in exactly the same way as in the taxon block. 

I'll let you guys decide what to do, but I'm guessing that the quick fix is easier to implement, and while not as ideal as the long-term fix, it solves the row scrambling problem prior to public release. 



----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=1126676&aid=2970700&group_id=248804