divergence time estimation with only stem age constraint / phylogeny estimate with binary traits

[David H.]

Hello,

Apologies for the rather long list of questions here. I'm new to BEAST X and have run into some challenges (some apparently with BEAST X itself and others of a more general nature). 

On a more general topic, I'm conducting an analysis of ~200 samples of one species with interests in determining source location of disjunct subpopulations, including an estimate of the timing of founding of subpopulations. 

In the literature, there is a single estimate of the age of the MRCA with its sister species. This stem age of the focal species is around 21 million years old, but the founding events of subpopulations are expected to be on the time scale of hundreds of years old. So, this is several orders of magnitude more recent than the stem calibration estimate. 

Given that the age of the crown group (i.e., age of the MRCA of the 200 or so samples) is likely to be much more recent than 21 MYA, and we don't have samples from the sister species, what is the best approach to constrain the stem age? 

On a more BEAST-specific issue, I am having trouble setting hyperparameters for lognormal priors for coalescent tree priors in beauti (Version v10.5.0, 2002-2025 on Ubuntu). The default is saved to the beast xml file after manually adjusting hyperparameters in beauti.  

I can edit the XML definitions by hand, but I am concerned there are other bugs that I may not be aware of.  For example, whether or not I click the 'stem' checkbox for taxon sets in beauti, the same output seems to appear in the beast xml file. I couldn't find enough documentation about the stem checkbox to determine if this is intended.  

On a separate topic, is there a way to conduct an analysis of phylogeny based on binary traits in Beast X? I apologize if there are already tutorials on this topic, but I could find none for BEAST X.  Additionally, beauti does not appear to allow one to define taxon sets when only discrete traits are loaded... I get a large java stack trace when I try to do this. 

Thank you for your help.

[Omar Mejía]

Dear David,

For me, it seems that the nature of your work is more phylogeographic
than phylogenetic. First of all, my suggestion is to found the number
of genetic clusters through baps, Structure or similar, once you
recovered the composition of the genetic clusters you will be able to
estimate the divergence times among populations in BEAST using a
strict molecular clock and a substitution rate for the gene that you
are using. You can use the substitution rate for the same gene in a
related taxon, otherwise, you can be able to obtain the substitution
rate from a phylogenetic analysis, may be you will be able to include
some additional species or genera, use fossil record or vicariant
events in a wide phylogenetic frame, and estimate the substitution
rate for your lineage using a relaxed molecular clock .

On the other hand, the Mk model could be used for a Bayesian
phylogenetic inference of binary data (Lewis 2001) and a calibrated
tree could be obtained in BEAST using the “Tip-Dating” section (Turner
et al. 2017). An alternative is to use the Paleotree R library
(https://pseudoplocephalus.com/2017/02/15/so-you-want-to-make-a-time-calibrated-phylogenetic-tree/).
Using either approach, you are able to recover a calibrated
phylogenetic tree from binary data.

Best regards

> --
> You received this message because you are subscribed to the Google Groups "beast-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to beast-users...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/beast-users/b3811933-164a-4d60-95b0-11d9471d1ca8n%40googlegroups.com.



--
Omar Mejía G
Laboratorio de Variación Biológica y Evolución
Departamento de Zoología
Escuela Nacional de Ciencias Biológicas-IPN