Wednesday, May 6, 2015

How to get the colors of the tips in a plotted "contMap" object

Today I received an interesting question about the phytools function contMap. contMap, remember, is a function that maps the observed & reconstructed value of a continuous character on to the tree. The question is as follows:

“How can I recover the color codes that are used to paint the tips (or the last bit) of a phylogeny after a reconstruction with contMap? I have a character mapped on my phylogeny and I want to use the same tip colors in a scatter plot (first two components of a PCA in this case)… each species in the plot should have the same color as in the tree. I tried going through the code but I get lost at some point and can't figure out how the last bit of the terminal branches are indexed and how the color assigment is done…”

An object created by contMap is actually just the same object that is read & plotted by plotSimmap, which plot.contMap uses internally. The colors are encoded in the "contMap" element cols, and the corresponding indices for the colors mapped along each edge, in the object element tree$maps.

To demo how we can pull the colors of all mapped species from this object, see the following demo:

library(phytools)
## simulate tree & data
tree<-pbtree(n=26,tip.label=LETTERS)
X<-fastBM(tree,nsim=2)
colnames(X)<-c("trait 1","trait 2")
## contMap for the first character only
obj<-contMap(tree,X[,1])

plot of chunk unnamed-chunk-1

## function to pull out the color of a given tip
foo<-function(obj,tip){
    jj<-which(obj$tree$tip.label==tip)
    kk<-which(obj$tree$edge[,2]==jj)
    setNames(obj$cols[names(obj$tree$maps[[kk]])[length(obj$tree$maps[[kk]])]],
        NULL)
}
colors<-sapply(obj$tree$tip.label,foo,obj=obj)
colors
##           A           B           C           D           E           F 
## "#004BFFFF" "#0023FFFF" "#00FF5FFF" "#0018FFFF" "#0EFF00FF" "#10FF00FF" 
##           G           H           I           J           K           L 
## "#3000FFFF" "#0062FFFF" "#0F00FFFF" "#00A1FFFF" "#00FF64FF" "#00FF4DFF" 
##           M           N           O           P           Q           R 
## "#00FF33FF" "#00FF49FF" "#00FF79FF" "#FF7F00FF" "#FF2400FF" "#FF0300FF" 
##           S           T           U           V           W           X 
## "#00FF03FF" "#3EFF00FF" "#D9FF00FF" "#FFAE00FF" "#FF4A00FF" "#FFF500FF" 
##           Y           Z 
## "#FFA400FF" "#FF7800FF"

OK, now how about using these to plot in a scatterplot? We can do that in our two dimensional simulated space:

plot(X,col=colors,pch=19,cex=1.5)
## just to position the labels a bit more nicely
ii<-X[,1]>(max(X[,1])+min(X[,1]))/2
text(X[ii,],rownames(X)[ii],pos=2)
text(X[!ii,],rownames(X)[!ii],pos=4)

plot of chunk unnamed-chunk-2

That's all there is to it!

No comments:

Post a Comment