Phylogenetically independent contrasts

Question

[sam]

Hello,

I'm trying to use Phylogenetically independent contrasts for a trait evolution model. I have a genotypic matrix of 500 samples by 5500 loci. Basically, I have a set of selected loci that I am using to quantify the trait. I would like to use R's PIC method to compute the traits (in the ape package). I have the genealogy that describes these selected loci. I know that PIC takes in a gene tree, however I'm little confused about how to make use of this method to get a quantitative phenotype for each sample. Any help would be greatly appreciated.

Answer 1

[Brice Sarver]

I think you are fundamentally confused about why you would use PICs in the first place.

Joe Felsenstein showed that data points from organisms are not independent due to their shared ancestry. You calculate PICs in order to make your data points IID. The theory behind this is beyond the scope of this post.

One assumption of PICs is that the traits evolve by Brownian motion, a simplistic process with nice mathematical properties. Your inputs, at least in the ape R package, are the tree relating your individuals and a list of the phenotypic traits (i.e., body size, brain size, etc.). These trait values are then 'corrected' using the phylogeny, and it is then possible to make statistical inferences from them.

TL; DR: you want to give pic() a vector of trait values, not the loci underpinning them.