A hack to show clades in an arc or fan-style tree using polygons

Just now messing around in R (as one does), I discovered a new hack for plotting clades as triangles (a la phylo.toBackbone) for fan or arc style trees.

Let me give a quick & lightly annotated demo.

First, I’ll simulate a 80 taxon tree. I’m going to set my computer seed first, just to ensure we all end up with a tree with the same node indices (for those following along at home).

## load packages
library(phytools)

## set seed
set.seed(99)

## simulate tree
tree<-pbtree(n=80)
tree

## 
## Phylogenetic tree with 80 tips and 79 internal nodes.
## 
## Tip labels:
##   t5, t73, t74, t38, t53, t54, ...
## 
## Rooted; includes branch length(s).

## graph our tree in arc style
plotTree(tree,type="arc",arc_height=1,fsize=0.6)
nodelabels(bg="white",cex=0.6)

Let’s say I want to replace the clade descended from node “125” with a triangle in my plot.

My “hack” to do so is to first identify all the tips descended from the node, which can be done as follows:

node<-125
tips<-getDescendants(tree,node)
tips<-tips[tips<=Ntip(tree)]
tips

##  [1] 44 42 43 45 46 47 48 49 50 51 52 55 53 54

Then we create a paint a “regime” on our tree for the clade descended from node “125”.

tree<-paintSubTree(tree,node=node,state="1",anc.state="0")

Then we re-plot our tree using a transparent color for the mapped “regime,” and add the polygon in a standard way. Here, I use chull only to sort my coordinates into a closed polygonal shape. All of them will appear in the polygon!

plot(tree,setNames(c("black","transparent"),0:1),type="arc",
  arc_height=1,fsize=0.6)
pp<-get("last_plot.phylo",envir=.PlotPhyloEnv)
dd<-c(node,tips)
mh<-chull(pp$xx[dd],pp$yy[dd])
polygon(pp$xx[dd][mh],y=pp$yy[dd][mh],lwd=1,col="red")

OK, it looks a bit more interesting than planned – I warned you it was a hack!

Let’s add the other major clades of the tree.

nodes<-c(85,93,95,108,118,125,138,146,152)
for(i in 1:length(nodes)) 
  tree<-paintSubTree(tree,node=nodes[i],state="1")
plot(tree,setNames(c("black","transparent"),0:1),type="arc",
  arc_height=1,ftype="off")
pp<-get("last_plot.phylo",envir=.PlotPhyloEnv)
for(i in 1:length(nodes)){
  tips<-getDescendants(tree,nodes[i])
  tips<-tips[tips<=Ntip(tree)]
  dd<-c(nodes[i],tips)
  mh<-chull(pp$xx[dd],pp$yy[dd])
  polygon(pp$xx[dd][mh],y=pp$yy[dd][mh],lwd=1,
    col=hcl.colors(n=length(nodes))[i])
}

That’s kind of fun.

We can do one more experiment in which we do the same thing, but we include all the nodes in our polygon – instead of just the MRCA and the tips.

par(bg="black")
nodes<-c(85,93,95,108,118,125,138,146,152)
for(i in 1:length(nodes)) 
  tree<-paintSubTree(tree,node=nodes[i],state="1")
plot(tree,setNames(c("white","transparent"),0:1),type="arc",
  arc_height=1,ftype="off")
pp<-get("last_plot.phylo",envir=.PlotPhyloEnv)
for(i in 1:length(nodes)){
  tips<-getDescendants(tree,nodes[i])
  dd<-c(nodes[i],tips)
  mh<-chull(pp$xx[dd],pp$yy[dd])
  polygon(pp$xx[dd][mh],y=pp$yy[dd][mh],lwd=1,
    col=hcl.colors(n=length(nodes))[i],border="white")
}

That’s too far.