[R-sig-phylo] tests of correlated evolution
Emmanuel.Paradis at mpl.ird.fr
Mon Apr 14 10:35:18 CEST 2008
Katie Wagner a écrit :
> Hi Emmanuel,
> Thanks for these thoughts -- this is a really interesting issue.
>> In R, when you print the results of gls() with summary() you have both
>> the AIC and log-likelihood printed at once. If you use the corMartins
>> function from ape to specify an OU type correlation structure, then
>> gls will estimate the 'alpha' parameter of the OU process and adjust
>> the number of estimated parameters accordingly.
> Yes, I used corMartins for the OU -- so that AIC should be ok.
>> How did you specify the correlation structure for the 'lambda BM'
>> model? We don't have one in ape at the moment.
> Perhaps this is incorrect, but I simply transformed my phylogeny
> according the the estimated lambda using lambdaTree() in geiger, and
> used this transformed tree and corBrownian() to generate a correlation
> structure to give to gls(). I think by using the transformed tree, I
> should be generating the appropriately modified BM correlation
> structure. But perhaps I'm misunderstanding how the lambda-modified
> correlation structure is generated. Please correct me if I am wrong
At least you should add 2 to the AIC returned by gls. But this won't
solve the problem pointed out by Simon.
I shall try to implement this correlation structure in ape. Does anybody
know where to find the formulae for this? They don't seem to be in
Pagel's paper in Nature (1999)...
> Katie Wagner
> Ph.D. Candidate
> Department of Ecology & Evolutionary Biology
> Corson Hall
> Cornell University
> Ithaca NY 14853
> Ph: (607)254-4240 email: cew35 at cornell.edu
> [[alternative HTML version deleted]]
> R-sig-phylo mailing list
> R-sig-phylo at r-project.org
Institut de Recherche pour le Développement
GAMET - BP 5095
361 rue Jean-François Breton
F-34196 Montpellier cédex 5
phone: +33 (0)4 67 16 64 47
fax: +33 (0)4 67 16 64 40
mailto: Emmanuel.Paradis at mpl.ird.fr
More information about the R-sig-phylo