search

phylogeography / SpreaD3

Spatial Phylogenetic Reconstruction of Evolutionary Dynamics 3
GNU Lesser General Public License v3.0
13 stars 7 forks source link

no HPD_modality attribute causes a continuous trait analysis of MCC tree to fail #27

Closed SimonGreenhill closed 7 years ago

SimonGreenhill commented 7 years ago

If there is no _%HPD_modality attribute on the tree, then the SpreaD3 gui raises an error:

setSelectedIndex: 0 out of bounds

STDOUT shows the real error:

locationsgeo_%HPD_modality attribute could not be found on the internal node. Resulting visualisation may be incomplete!

... and yes, the visualisation is incomplete (no HPD ranges).

The Spherical Diffusion phylogeography package for BEAST2 doesn't seem to log this attribute, and the HPD info is in locationsgeo{1,2}_95%_HPD={..}.

Perhaps the GUI should display a more helpful error message (like the one in STDOUT).

Does the BEAST2 package need to log the modality for SpreaD3 to work?

GuyBaele commented 7 years ago

@SimonGreenhill Can you send me the file you were trying to load into SpreaD3 so that I can try to reproduce the problem and debug? Thanks!

GuyBaele commented 7 years ago

This is the same issue as posted on the BEAST user list: https://groups.google.com/forum/#!topic/beast-users/sl98sz4mVRU

I'm not getting the 'setSelectedIndex: 0 out of bounds' error message though when I load the MCC file that Kalene shared on the user list, as SpreaD3 tells me that there are 'No matching HPD_modality attributes found for selected coordinate attributes', which accurately reflects the problem.

SimonGreenhill commented 7 years ago

That looks and sounds exactly like it. Are you happy with the tree from that mailing list post, or want me to send my one too?

GuyBaele commented 7 years ago

This has been fixed in the source code but no new jar file had been released yet. I'm putting a new build onto our lab's website, but note that it has not yet been thoroughly tested: https://rega.kuleuven.be/cev/ecv/software/SpreaD3

SimonGreenhill commented 7 years ago

Just confirming, that this RC now generates the correct warning.

The underlying problem seems to come from treeannotator in the beast2 package not including HPD_modality annotations, I will follow up with beast2 developers.

GuyBaele commented 7 years ago

Is it possible to run a BEAST2 .trees file through TreeAnnotator of the BEAST 1 package?

SimonGreenhill commented 7 years ago

Yes, that works.