2
$\begingroup$

Using starBeast I ran a 50 locus tree, with 34 individuals, using evolutionary models indicated by AICc in jModel test. I ran this tree with a MCMC chain length of 500 million, storing every 10,000 generations. I used a relaxed clock log normal.

I get low ESS values for posterior, prior, birthRate.t:Species, YuleModel.t:Species, popMean, and TreeHeight.Species. ucldMean.c:Locus_name is low for each loci. The TreeHeight statisitcs also have low ESS values. Also rate.c:Locus_name.mean for each loci is low as well.

Is there a way to change my parameters in Beauti to help increase the ESS values of these statistics? Is there a parameter that would likely cause the ESS values of these particular statistics to be low? Would increasing the MCMC chain length increase the ESS values, given that I've already run half a billion?

Improve this question
$\endgroup$
1
  • 1
    $\begingroup$ It is not my field, but I don't understand many abbreviations you use. You might get (more) answers if you explain with simpler words what is the problem. As I understand you want to increase the value of one value by optimizing some parameters of a program, is this helpful? Is there something in the mailing list of the tool helpful? Did you try or read something to solve this? $\endgroup$
    – llrs
    Commented Apr 17, 2018 at 7:26

1 Answer 1

Reset to default
1
$\begingroup$

The official BEAST website suggests a number of solutions to improve the ESS of a parameter

It is not uncommon to have several billions of generations so I don't imagine increasing your number of generations would hurt.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.