I’m interested in using the OUCH package to estimate BM and OU parameters for a 
specific trait among many different trees.
My goal is to determined which model is the most suitable for each tree, 
applying likelihood ratio test.

As I’m a new user in R when it comes to phylogenetic analysis, I started by 
running the documentation example (Hansen, documentation page 11) and was 
surprised to see that the loglikelihood was a positive number

brown(data = otd[c("tarsusL", "beakD")], tree = ot)

           [,1]       [,2]
[1,] 0.02878091 0.08897504
[2,] 0.08897504 0.43711838

[1] 3.020419

[1] 1.826695

[1] 9.90115

As this number is crucial for further analysis - Is this a transformation of 
the resulting log likelihood? (e.g. -2 * log(L) as described in the paper) or 
am I missing something here..?

In addition I have another issue, I have a tree constructed of ~400 viral 
genomes and their corresponding trait values. When I’m running the 
documentation script with my own data (in the same format)
I get the following error:
Error in solve.default(v, e) : 
  system is computationally singular: reciprocal condition number = 1.59061e-17

I guess it says that my variance covariance matrix is not inversable, hence I 
manually tried to adjust the retol parameter in the Hansen function in order to 
make it work (however I’ll need to second guess my results?), but I still get 
the same error.
Code example:
h1 <- hansen(
+     tree=ot,
+     data=otd[c("k5")],
+     regimes=otd["regimes"],
+     fit=TRUE,
+     sqrt.alpha=1,
+     sigma=1,
+     maxit=500000,
+     reltol=1e-20,
+     method="Nelder-Mead"
+ )

I’ll be thankful for any advice or answer,



R-sig-phylo mailing list - R-sig-phylo@r-project.org
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/

Reply via email to