Starting trees for gene trees in starBeast2

[Ben Stone]

Hello all,

I'm having trouble figuring out the format for inputting starting gene trees for a starBeast2 run. The goal of my analysis is to constrain the species tree topology and estimate only branch lengths on the species tree (which I think I have figured out). However, as it's currently set up, the MCMC is failing to converge, and I think it's likely due to difficulties with the gene trees (gene tree likelihoods and some rate parameters are failing to converge). My thought is that providing starting MLE gene trees will help with the convergence issues.

I've looked at this blog post (https://www.beast2.org/2014/07/28/all-about-starting-trees) for help with setup, but it seems a bit outdated, and I wasn't able to really follow the directions. I have attached my XML file, which includes my (failed) changes to specify starting gene trees. I don't get an error message, but the starting gene trees are not the trees I have put into the XML file (starting at line 8737).

Any help with where I'm going wrong is greatly appreciated.

-Ben

Ben Stone

PhD Candidate

The Ohio State University

Dept. of Evolution, Ecology & Organismal Biology

[Remco Bouckaert]

Hi Ben,

Attached some XML where I changed the following:

1. copy element with id=SBI to before the run element

2. for the species tree, add an init element with the Newick for species starting tree. The Newick had taxon labels for genes, not species, so I replaced them with species tree taxa. You might want to verify whether that was done correctly.

3. for every gene tree, add an init element with a Newick tree for the gene trees.

This still won’t start because the starting gene trees refer to taxa that are not included in the alignments. For example, vizcainensis is in the tree for gene 19, but vizcainensis is not in the alignment gene19. Also, the gene trees do not fit inside the species tree, since gene trees are much shorter than the species tree. If you edit the Newick gene trees and species tree so that the gene trees taxa match with those in the alignment and fit inside the species tree, the analysis should start.

You mention that you thought the description on the blog post was outdated — can you point to something specific so I can fix this?

Cheers,

Remco

[Ben Stone]

Hi Remco,

Thanks for the response. With respect to the blog post, it seems I thought it was outdated because I misunderstood what was meant by the element with id=SBI. It is now clear to me what is meant by copying that element to before the run element, and then for the gene trees, subsequently adding the init elements to the SBI element after the run element.

-Ben