[Mesquitelist] phylogenetically corrected PCA

[Alejandro Gonzalez V]

Hi Marcin,

You should have a look at the following paper:

Revell L. J. 2009 Size-correction and principal components for interspecific comparative studies. Evolution 63:3258-3268

Liam provides code in the paper to run Phylogenetic PCA on species data (not contrasts) in R and Matlab. It works well, I've used it myself.

Cheers,

Alejandro

 
__________________________________

Alejandro Gonzalez Voyer
Post-doc

NEW ADDRESS & NEW E-MAIL

Estación Biológica de Doñana (CSIC)
Avenida Américo Vespucio s/n
41092 Sevilla 
Spain

E-mail: alejandro.gonzalez at ebd.csic.es

Tel: +34- 954 466700, ext 1749

Website (From my previous position): http://www.iee.uu.se/zooekol/default.php?type=personalpage&lang=en&id=146








On Feb 8, 2010, at 12:43 PM, Marcin Czarnoleski wrote:

> Hi,
> I would be grateful for an advice how to solve my problem with 
> phyllogenetic analysis pased on PCA (possibly with Mesquite). I use data 
> on some traits of several species. I would like to use PCA to calculate 
> PC scores from these traits, and use these scores to correlate with 
> other characters of the species. But I would like to correct my analysis 
> on a phylogentic signal. So basically, I should calculate 
> phylogenetically corrected PC scores and then correlate them with 
> contrasts of other species characters. But I have a question, should I 
> first run PCA on raw data, and then calculate contrasts of the PC scores 
> obtained from this PCA, and finally use these contrasts in correlattions 
> with contrasts of other characters of species. Or alternativelly, I 
> should run PCA on contrasts of species traits, and then use scores of 
> these PCs to correlate them with contrasts of other charactes of 
> species? If the second approach is recommended then I see two problems. 
> First, to run PCA on contrasts it is adviced to use correlation matrix 
> from contrasts (but correlations are forced through 0 and positivised). 
> If I calculate such matrix and use it as an input file for PCA in an 
> ordinary statistical software, I am unable to calculate PC scores 
> because the information about species (cases) is missing (correlation 
> matrix provides PCA only with correlation coefficients). So withought 
> the scores I cannot proceed further and test correlations between 
> phylogenetically corrected PC scores and contrasts of other traist. 
> Second, even if I was able to obtain phylogenetically corrected PC 
> scores, how should I analyse their correlation with contrasts of other 
> traits. I understand that a normal correlation on contrasts requires 
> prior positivisation and forcing through 0, but I am not sure what 
> should be done in terms of this requirement in the case of scores vs 
> contrasts correlation. I should be grateful for any opinion on my 
> problem and I would be also grateful for the information whether 
> Mesquite can help solve my problem.
> Marcin Czarnoleski
> _______________________________________________
> Mesquitelist mailing list
> Mesquitelist at mesquiteproject.org
> http://mesquiteproject.org/mailman/listinfo/mesquitelist

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mesquiteproject.org/pipermail/mesquitelist/attachments/20100208/3fb63ba3/attachment.html