Hi Katharine.

You can also do ancestral state estimation for the model "ARD" using the packages diversitree. There is a worked example on my blog here: http://blog.phytools.org/2016/09/possible-bug-in-acemodelard-and-other.html (also highlighting a possible bug in ace for model="ARD" & other non-reversible models). Let us know if this helps.


All the best, Liam

Liam J. Revell, Associate Professor of Biology
University of Massachusetts Boston
web: http://faculty.umb.edu/liam.revell/
email: liam.rev...@umb.edu
blog: http://blog.phytools.org

On 12/5/2016 10:07 AM, Katharine Walter wrote:
Hi,

I am interested in estimating discrete ancestral states in ace using the
calls:
fitER<-ace(tipRegion,tree,model="ER",type="discrete")
fitARD<-ace(tipRegion,tree,model="ARD",type="discrete", use.expm=TRUE)

When I fit the ARD model (with either use.expm=TRUE or use.eigen=TRUE), I
get the following warning:
 1: In nlminb(rep(ip, length.out = np), function(p) dev(p), lower = rep(0,
 : imaginary parts discarded in coercion

Additionally, the two rates which I expect to be the highest are estimated
as 0. There are 3 tip states for a tree of 146 tips and 145 internal nodes.
A likelihood ratio test supports the ARD model, but I cannot trust model
results. I am wondering:
1) is there an obvious error that is creating this warning and unexpected
output?
2) should I use a simpler model (i.e. SYM)? Or should I reclassify tip
states so that I examine transitions between only 2 states?

Thank you in advance for your help! This is a fantastic package!
Best,

        [[alternative HTML version deleted]]

_______________________________________________
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/


_______________________________________________
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/

Reply via email to