Fixed Topology Only, Estimate Branch Lengths

[Sierra Gillis]

Hi there,

I am new to RevBayes, and currently studying phylogenetic graphical modelling (grad school) so I'm fairly new to that as well.

I have a LOT of discrete morphological traits, but they are all encoded the same way (values of 1, 2, or 3). I also have a fixed topology. I'm trying to estimate the branch lengths and ancestral states. 

I have tried to read in a BranchLengthTree, and then follow in the discrete morphology tutorial (https://revbayes.github.io/tutorials/morph_tree/V2.html) and setting the value of the tree after creating the stochastic variable. This gets stuck in a loop of trying to initialize a starting state for m_morph[1].

T <- readBranchLengthTrees("data/mytree.nwk")[1]

#set up the moves for branch lengths, no moves on topology

phy ~ dnUniformTopologyBranchLength(<...>)

phy.setValue(T)

then like in the tutorial

m_morph[idx] ~ dnPhyloCTMC(tree=phy, ...)

I tried following the BDP tutorial (https://revbayes.github.io/tutorials/divrate/simple.html) with rho=1 and setting the value of the tree to the timeTree, but the same loop occurs, and in general I can't estimate branch lengths.

T <- readTrees("data/mytree.nex")[1]

phy ~ dnBDP(<...>)

phy.setValue(T)

Is there a way to set the stochastic tree variable with a fixed topology but still estimate the branch lengths and the m_morph (and eventually ancestral states) using discrete morphological data? Is there a gap in my understanding of how to set up the Mk model for this analysis?

Any help is greatly appreciated,

Sierra

[Sierra Gillis]

I should note that the output of the "loop" I'm referring to is:

Drawing new initial states ...

Could not compute lnProb for node m_morph[1].

Standard character matrix with 7 taxa and 14056 chara