Identify clusters of contiguous factors within a dendrogram

dendrogram-contigs.r

# Identify clusters of contiguous factors within a dendrogram
# Aaron Wolen
#
# See http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0038422
# for an example situation in which this would be useful
#
# x: dendrogram
# f: a named vector containing the factor that defines the grouping of leaves,
#    must be named using the same values that define x's leaf labels

contigs <- function(x, f) {
  UseMethod("contigs", x)
}

contigs.dendrogram <- function(x, f) {
  
  if(is.null(names(f))) stop("f must be a named vector.")
  
  # obtain ordered leaf labels
  leaves <- unlist(dendrapply(x, attr, which = "label"))
  groups <- f[leaves]
  
  # identify clusters
  cluster <- rle(groups)
  clustered <- cluster$values[cluster$lengths > 1]
  clustered <- rep(cluster$length, cluster$length) > 1
  groups[!clustered] <- NA
  
  return(groups)
}