TVM+G+I

[joe]

Hi,

Modeltest indicates that the best fit model under AIC is TVM+G+I, and
secondly the GTR+G+I model for codon-aligned viral base sequence data
that I'm assessing. Can (How can) Beauti parameters and/or the xml be
modified to use the TVM+G+I values. Other models I'm selecting in
Beauti are SRD06 and all parameters unlinked. I'll compare Strict vs
Relaxed uncorrelated log normal and start with the Yule prior (or
would another prior be better suited for the TVM+I+G assumptions) ?

Many thanks,
Joe

[Gerald Schneeweiss]

Hi Joe,

you cannot get the TVM model directly in Beauti, but you choose the
GTR-model and then change the rate parameters in the xml-file in the
gtrModel element in a way, that there is only one transition rate. e.g.,
by restricting gtr.ag to be the same as gtr.ct (<rateAG> parameter
idref="gtr.ct"/> </rateAG>). You will also have to remove the swapoperator
for (in this case) gtr.ag and remove it also from the log-section. A full
description of how to get simpler models from the GR-model is given (for
the Tamura-Nei-model) in the Tutoral section of BEAST at
http://beast.bio.ed.ac.uk/Nucleotide_Substitution_Models

Cheers, Gerald