Odd LnPr traces

[David Lohman]

Greetings, 

I just finished an ExaBayes analysis on 203 samples with up to 13 loci (11 partitions) using the configuration file parameters pasted below. When I open the parameters file in Tracer, I see that some of the ESS values are far too low. Moreover, some of the traces--for example, the LnPr shown in the attached screencap--look bizarre. Is this because the analysis is stuck on a local optimum?  I'd be grateful if someone could diagnose the problem and propose a solution.  Thanks!

Best wishes,

David Lohman

#NEXUS
begin RUN;
 numRuns  2        
 numGen 5000000  
 diagFreq 5000    
 samplingFreq 2000    
 tuneFreq 500        
 printFreq 5000  
 parsimonyStart true    
 heatedChainsUseSame false  
 checkPointInterval 10000  
 numCoupledChains 4  
 numSwapsPerGen 1      
 convergenceCriterion mean
 sdsfIgnoreFreq  0.1        
 sdsfConvergence 0.10      
 burninProportion 0.25      
end;

begin params;
 brlens    = (0-10)      
end; 

[Alexandros Stamatakis]

Hi David,

Could you also send me what the LnL trace looks like?

It could be that the MCMC is sampling two distinct local optima.

Could you maybe also compute the difficulty score of your dataset with
this easy to use tool here (it runs fast) and report it:

https://academic.oup.com/mbe/article/39/12/msac254/6832260

Alexis

> Screen Shot 2023-04-08 at 2.34.46 PM.jpg
>

> --
> You received this message because you are subscribed to the Google
> Groups "ExaBayes" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to exabayes+u...@googlegroups.com
> <mailto:exabayes+u...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/exabayes/601b4410-a2ca-404c-b92f-2de13add17ean%40googlegroups.com <https://groups.google.com/d/msgid/exabayes/601b4410-a2ca-404c-b92f-2de13add17ean%40googlegroups.com?utm_medium=email&utm_source=footer>.

--
Alexandros (Alexis) Stamatakis

ERA Chair, Institute of Computer Science, Foundation for Research and
Technology - Hellas
Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology

www.biocomp.gr (Crete lab)
www.exelixis-lab.org (Heidelberg lab)