I think it's convenient to think about how fitContinuous actually works here,
or at least to think about the impact of the EB parameter on a transformed
tree's branch lengths. Using simulations, try this:
require(TreeSim);
require(geiger);
phy - sim.bd.taxa(50, 1, 0.1, 0.05, 1)[[1]][[1]]; ##
From: Nicolas Campione [nicolas.campi...@mail.utoronto.ca]
Sent: Monday, November 05, 2012 2:04 AM
To: Slater, Graham
Cc: David Bapst; Frank Burbrink; r-sig-phylo@r-project.org
Subject: Re: [R-sig-phylo] fitContinuous-Early Burst Model
Hello All,
Great
Hi all,
I think there might be 2 issues to consider here. The first concerns data
transformations and how we consider traits to evolve over phylogenies. Many
people would consider it most appropriate to use log transformed values of a
continuous trait for model fitting purposes; for example,
Hi Eliot and others,
You can also use the code contained in the fitContinuousMCMC package
available here -
https://www.eeb.ucla.edu/gradstudents/slater/Graham/code.html
And described in this paper -
http://onlinelibrary.wiley.com/doi/10./j.1558-5646.2012.01723.x/abstrac
t
To place
Dear list,
I'm writing to promote this month's issue of Methods in Ecology and Evolution,
which is now online and contains a special feature edited by Luke Harmon and I,
titled Unifying fossils and phylogenies for comparative analyses of
diversification and trait evolution. The feature
Hi Andrew, Simone, et al.,
It's probably worth adding the very minor distinction that the results of
Andrew's suggested test and of phytools' phylosig test have slightly
different interpretations. Assuming lambda 1, a significant LRT when
comparing the fit of a lambda model to a non-lambda, BM
Hi Jorge,
I find that this happens most often when I forget to set the row names of
data. Try rownames(data) and if they're not the tip labels, then that's
your problem.
Graham
---
--
Graham Slater
Peter Buck
Hi Mercedes,
The relaxed bm output summarizes the shifts sampled during the rj-mcmc, so it’s
not surprising that it will show some shifts in rate along some branches. The
important part of the output is the posterior probability of a shift at a given
node — if this is low, then there is little
Hi Louise,
There is a k for
all at the bottom of the output for each model fit of all traits so I
didn't think that it was probably the kappa needed for Kappa
You’re right - K here is the number of parameters in the model. BM has K= 2
(root state and rate) and other models add an additional
Hi Juanita
try:
lapply(simtrees, drop.random, n = 937)
when using lapply (or any of the apply family) you need to specify additional
arguments taken by the function youre applying by name, along with their
values.
graham
Graham
Hi Jon,
So there are are a few ways I can imagine doing this. One thing you should
probably not do though is rescale the branches of the tree directly, at least
in the case of OU, as this leads to incorrect covariances for pairs of taxa
where one or either does not survive to the present day
Hi Viviane,
If you believe that your traits are correlated - that is there is a
variance-covariance matrix that describes the evolutionary correlations among
these traits, then I dont think you don't want to fit the models to each trait
and transform the tree each time you simulate. Youd be
Dave,
I�ve had the same trouble with shuffling. However, all of this can be avoided
if you specify the simple format for your .con file in the mrBayes block.
sumt conformat = simple;
The resulting tree will correctly display posterior probabilities in a
phyloformat tree.
Graham
Hi Roger,
Branch lengths of trees in the t files are in expected changes per site. To get
these into time units, you just need to divide the branch lengths by the
corresponding clock rate in the p.file.
Cheers,
Graham
Graham Slater
Hi Daniel,
There’s a difference between a method being able to handle fossil data, that is
a dataset consisting of a non-ultrametric tree an data for all tips including
non contemporaneous ones, and a method allowing you to directly specify trait
values at nodes. Most trait evolution methods
Oops - sorry Daniel, yes that should have been addressed to Nathan...
Graham Slater
Peter Buck Post-Doctoral Fellow
Department of Paleobiology
National Museum of Natural History
The Smithsonian Institution [NHB, MRC 121]
P.O. Box 37012
Hi Milton,
The drift model gives a time-dependent change in the expected, or mean value of
the trait. If we denote the drift parameter as M, then the expected value of
the trait, E(x), at time t is a+Mt where a is the root state of the trait. So
if M is positive, your trait tends to get larger
Hi Dan et al,
Sorry I won�t be of much help analytically - my functions were written
specifically for the time slice problem. It wouldn�t be too difficult to write
a function to fit the model you describe, but it seems as though Julien has
this all covered in mvMorph already, so that does seem
Hi Solomon,
It’s because neither of the things you are estimating are Blomberg’s K and both
are different from one another. In Caper, you are estimating Pagel’s Kappa (a
branch length transformation that attempts to explain evolutionary change as
punctuational [1] vs gradual [=1] ), while in
19 matches
Mail list logo