Question

I would like to know if there is a function that applies an identifier to all nodes in a given dendrogram (binary tree).

So I want a function that after it on a given tree would do the following:

attr(tr,"ID")             ## should give 1   or  1 

attr(tr[[1]],"ID")        ## should give 2  or 10

attr(tr[[2]],"ID")        ## should give 3  or 11

attr(tr[[c(1,1)]],"ID")   ##  should give 4  or 100

etc…

And, if given to start with binaryID 110 (ID of Head Node)

it’s 1st child ID should be 1100
it’s 2nd child ID should be 1101

Note:
– dendrapply() applies a function to all nodes in a tree

Package using= “stats”

 D=rbind(
+ c(1,1,1,1,1),
+ c(1,2,1,1,1),
+ c(2,2,2,2,2),
+ c(2,2,2,2,1),
+ c(3,3,3,3,3),
+ c(3,3,3,3,2))

Ddend=as.dendrogram(hclust.vector(D))

funID<-function(tr,StartID){
 .....
 attr(n,"ID")= ID  # for all the nodes in tr
 }

What would funID be?

Answer 1

This is code taken from stats:::reorder.dendrogram and modified to suit the purpose of labeling the root and each leaf with an increasing integer. It might not exactly fit your specs but see if it does …

label.leaves <- 
 function (x, wts) 
 { N=1
     if (!inherits(x, "dendrogram")) 
         stop("we require a dendrogram")
     oV <- function(x, wts) {
         if (is.leaf(x)) {
             attr(x, "ID") <- N; N <<- N+1
             return(x)
         }
         k <- length(x)
         if (k == 0L) 
             stop("invalid (length 0) node in dendrogram")
         vals <- numeric(k)
         for (j in 1L:k) { N <- N+1
             b <- oV(x[[j]], wts)
             x[[j]] <- b
             vals[j] <- N; N <- N+1
         }
       x
     }
     stats:::midcache.dendrogram(oV(x, wts))
 }

Testing:

> Ddend.L <- label.leaves(Ddend)
> rapply(Ddend.L, function(x) return( attr(x, "ID") ))
[1] 1 2 3 4 5 6