Fixed Tree Analysis Fixed topology estimate branch lengths error

[Antonio Gomez]

Dear BEAST nation,

I am trying to estimate divergence times for a fixed tree topology with ~500 tips. I'm using BEAST 2.7.7 and the FixedTreeAnalysis template. I can read in my topology (fully bifurcating, rooted & ultrametric) without issue and can succesfully set up my 13 MRCA priors.

However, when I go to run the .xml, I get the following error

"Warning: at least one operator required to run the MCMC properly, but none found.
Warning: state contains a node Tree.t:tree for which there is no operator.r"

Can you please help?

Thank you!

[higg...@gmail.com]

Hi Antonio,

The fixed tree analysis is the wrong template for what you want to do: it fixes the tree to a given tree and assumes it will not change (not only the topology is fixed but the internal node heights as well).

To perform an analysis on a fixed topology, allowing the node heights to change, you could do the following:

1. Create a dummy alignment (fasta or nexus file) with all taxa you want to include in the tree.

2. Use the standard template to import the alignment.

3. Set the starting tree by selecting menu "View=>Show starting tree panel", select "Newick tree" from the drop down box, and copy/paste the tree in Newick format in the entry for "Newick"

4. Disable operators that change the topology by selecting the menu "View=>Show operators" and set the weight of the "Subtree slide", "Narrow exchange", "Wide exchange" and "Wilson Balding" to zero.

5.Specify the priors in the priors panel.

Hope this helps,

Remco

[Antonio Gomez]

Dear Remco,

Fantastic! Yes, this is very helpful, thank you! 

With dummy data do you also set the clock operators to zero, as well? Otherwise, the MCMC will attempt to estimate these parameters from bogus data, correct? 

I appreciate it! :)

Toño

-- 
You received this message because you are subscribed to a topic in the Google Groups "beast-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/beast-users/O4r60W202Eo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to beast-users...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/beast-users/9a1f8567-f1e8-4a0c-bc8e-a66512463dd8n%40googlegroups.com.

[higg...@gmail.com]

Yes, that makes sense -- otherwise the clock rate will just sample from the prior, which is probably harmless, but wastes a bit of computation.

Cheers, Remco