Could multiple runs from EBSP combined in LogCombiner generate a bias in the results? (BEAST 2.X) Strange results

[nate...@gmail.com]

Dear Beast users
I run 3 different EBSP analyses in BEAST 2.7. I combined the log files in logcombiner and then used  R script following the Extended Bayesian Skyline Plot tutorial for BEAST 2 from Joseph Heled updated by Tim Vaughan.
But something got wrong.
When I combined the log file, the coalescent time was super estimated, as in the following figs:

Fig A – One run:

Fig B- The 3 runs combined:

 
I know I am missing something, but I can´t figure it out. 
I would really appreciate some help.
Cheers
Natácia

[Alexei Drummond]

Did you get a similar result from each individual run before combining? Perhaps there is a problem with the prior on the first population size? What is your mutation rate fixed to, or estimated as?

On 19/03/2023, at 3:18 PM, nate...@gmail.com <nate...@gmail.com> wrote:

Dear Beast users
I run 3 different EBSP analyses in BEAST 2.7. I combined the log files in logcombiner and then used  R script following the Extended Bayesian Skyline Plot tutorial for BEAST 2 from Joseph Heled updated by Tim Vaughan.
But something got wrong.
When I combined the log file, the coalescent time was super estimated, as in the following figs:

Fig A – One run:

<SingleRun.png>
Fig B- The 3 runs combined:

 <Combined logfiles.png>
I know I am missing something, but I can´t figure it out. 
I would really appreciate some help.
Cheers
Natácia

--
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 on the web visit https://groups.google.com/d/msgid/beast-users/14f6bd87-cf0c-44f3-b5b0-e5157f636302n%40googlegroups.com.
<Combined logfiles.png><SingleRun.png>