problem with running Fossilized Birth Death dating analysis on BEAST2

[Damien Esquerre]

Hi all,

I have been trying to do a dating analysis with the FBD model on BEAST2 following a tutorial. However I have encountered a problem I can't solve.

Because all the fossils I am using are crown and not stem, I thought I should condition on the root and not the origin (according to the tutorial, although I am not sure what it means).

When filling the priors in BEAUTi, an error appears on the tree prior saying "input 'origin' must be specified". But then the field to fill the origin is gone, and looks like this: 

I tried ignoring it but when I run the xml in beast I get the same error message. I hope an easy edit of the xml file could be the solution!

Thanks so much for your time

[Oriane Loiseau]

Hello,

I faced the same problem when I tried to condition on root, following what Tracy Heath's tutorial on FBD recommends, just like Damien. So it seems that there may be a bug of BEAUTi here.

Because conditioning on root was impossible to choose, I ran an analysis using "condition on origin" even though I have no stem fossils. The result is on the overall similar to what I get using node dating with calibration densities in BEAST1, except for the root and a few basal nodes that are way too old and completely unrealistic for my group of organisms. Although it is not very clear to me how this prior impacts the birth-death process,  I guess it wouldn’t happen using the "condition on root" option.

Any insight on how to get around this problem would be much appreciated !

 

[Filipe Gudin]

Dear beast users,

I am facing a similar problem when setting the configuration on beauti and running a FBD dating analysis. I am trying to estimate the divergence times of a group of extant flies, using a few fossils available. First, following the tutorial provided by Heath, I conditioned the age of the tree on the time of origin, specifying an age interval for the group and a prior distribution. However, the ages estimated for the backbone of the tree were much older than the ones provided in previous publications (with a difference of more than 40 ma). I realized then that I have only crown fossils, so it should be more appropriate to condition the age of the tree on the root. Therefore, I used the same configuration in the xml file, only toggling on "Condition on Root". After one week of analysis, however, the ages are older than before and totally unrealistic, exhibiting a larger variance too (a figure of Tree height on tracer is attached, comparing both analysis within the period of one week on CIPRES. The analysis conditioning on root is the blue, around 200 ma, and the analysis conditioning on the origin is the gray one, around 100 ma).

Does anyone know what could be the cause of this huge difference?