Extended Bayesian Skyline Plot

[Petr]

Hi all,

I am very keen to use the Extended Bayesian Skyline Plot for our data.
With great interest, I read the paper by Joseph Heled and Alexei
Drummond published last year in BMC Evolutionary Biology and
describing this method. Unfortunately, I was not able to locate the
version of BEAST that implements EBSP (as mentioned in that paper),
using either the links provided in the paper or by browsing or
searching the BEAST website. Would anybody be so kind and give me some
hints on how to obtain a program implementing this new method, please.

Thank you.

Best wishes,
Petr Kotlik

[Jillian]

Hi Petr,

You can download the Beast development jar file that includes EBSP at
http://code.google.com/p/beast-mcmc/downloads/list - it's the first
one on the list.

Best,
Jillian

[Petr]

Hi Jillian,

Thanks for the reply!

Ok, I have found the way how to execute an appropriate xml file using
that jar file (beast-ebsp-devl.jar). However, the current BEAUti
version (1.4.8) does not seem to include ESPB as an option for
creation of an xml file. I appreciate that I could modify my xml file
manually, e.g. using the example file associated with the paper on
EBSP. But I did not find a way how to analyse the log file - the
current version of Tracer does not include the possibility to use
demographic indicators that are logged during EBSP analyses instead of
group size parameters logged by the standard BSP.

Have you actually used the beast-ebsp-devl.jar file to perform the
EBSP analysis? If you did, I would really appreciate if you could give
me some hints how to overcome these apparently missing elements in the
current implementation of EBSP (especially that of appropriate Tracer
version).

Thank you.

All the best,
Petr

[Jillian]

Hi Petr,

You can still open the log file in Tracer to check the ESS and and
convergence, but you cannot make the EBSP in Tracer. Instead, you use
the .csv output file, which includes the skyline plot data - I made it
a .txt file and opened it in Excel as delimited data so I could graph
it myself. This worked just fine for me.

The current Beauti doesn't include options for setting up EBSP. I've
been using the example file to modify the xml by hand and I've been
successful so far - haven't tried changing the substitution models and
such yet, but I think that should be fairly straightforward.

Hope that helps!

Best,
Jillian

[Petr]

Dear Jillian,

That was very helpful! I did not realize that the csv file can be used
to plot the EBSP. Thanks so much for pointing me in that direction!

I think that it is a shame that there are no instructions available on
the BEAST web how to us this EBSP implementation (or are they?).

Thanks again, Petr