Using RAxML bootstraps for phylogenetic post-analyses

[Klaas]

Hi All,
I want to use the RAxML output (best tree and bootstrap trees with branch length)
to conduct divergence time and diversification analyses. As I have a rather large data set,
I have to use r8s for estimating divergence times and cannot rely on BEAST.

I now  wonder, if it makes sense to use the bootstrap trees of RAxML to estimate some
kind of confidence intervals for node ages in r8s. I expect these trees being much worse
than the best tree ML and I am not sure, if this will result in erroneously large confidence
intervals, thus overestimating uncertainties in node ages.

In a next step, I plan to use these r8s trees to estimate diversification in BiSSE. Again, I
am afraid that the branch length of the underlying Bootstrap trees, as well as the derived
node ages will again produce to much noise and bias my diversification analyses.

Thanks in advance

Klaas

[Alexandros Stamatakis]

the BS trees will most probably have a different tree topology.

what you can do is the following (see raxml manual for individual commands):

1. generate BS replicate alignments with RAxML

2. Optimize br-lens and other ML parameters for these BS rep. alignments
on the best-scoring ML tree you found.

3. Use this collection of trees with r8s

This allows you to obtain a variance for the br-lens, but uisng the same
underlying tree topology ...

Cheers,

Alexis

> --
> You received this message because you are subscribed to the Google
> Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to raxml+un...@googlegroups.com
> <mailto:raxml+un...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology
Adjunct Professor, Dept. of Ecology and Evolutionary Biology, University
of Arizona at Tucson

www.exelixis-lab.org

[Klaas]

Dear Alexis,
thanks for your answer! This makes imho sense, I will give it a try.

Klaas