-## drop.tip.R (2009-03-22)
+## drop.tip.R (2011-11-21)
## Remove Tips in a Phylogenetic Tree
-## Copyright 2003-2009 Emmanuel Paradis
+## Copyright 2003-2011 Emmanuel Paradis
## This file is part of the R-package `ape'.
## See the file ../COPYING for licensing issues.
-extract.clade <- function(phy, node, root.edge = 0)
+extract.clade <- function(phy, node, root.edge = 0, interactive = FALSE)
{
Ntip <- length(phy$tip.label)
ROOT <- Ntip + 1
Nedge <- dim(phy$edge)[1]
wbl <- !is.null(phy$edge.length)
- if (length(node) > 1) {
- node <- node[1]
- warning("only the first value of 'node' has been considered")
- }
- if (is.character(node)) {
- if (is.null(phy$node.label))
- stop("the tree has no node labels")
- node <- which(phy$node.label %in% node) + Ntip
+ if (interactive) node <- identify(phy)$nodes else {
+ if (length(node) > 1) {
+ node <- node[1]
+ warning("only the first value of 'node' has been considered")
+ }
+ if (is.character(node)) {
+ if (is.null(phy$node.label))
+ stop("the tree has no node labels")
+ node <- which(phy$node.label %in% node) + Ntip
+ }
+ if (node <= Ntip)
+ stop("node number must be greater than the number of tips")
}
- if (node <= Ntip)
- stop("node number must be greater than the number of tips")
if (node == ROOT) return(phy)
phy <- reorder(phy) # insure it is in cladewise order
root.node <- which(phy$edge[, 2] == node)
start <- root.node + 1 # start of the clade looked for
anc <- phy$edge[root.node, 1] # the ancestor of 'node'
- next.anc <- which(phy$edge[-(1:start), 1] == anc) # find the next occurence of 'anc'
+ next.anc <- which(phy$edge[-(1:start), 1] <= anc) # find the next occurence of 'anc' or an 'older' node
keep <- if (length(next.anc)) start + 0:(next.anc[1] - 1) else start:Nedge
if (wbl) phy$edge.length <- phy$edge.length[keep]
TIPS <- phy$edge[, 2] <= Ntip
tip <- phy$edge[TIPS, 2]
- phy$tip.label <- phy$tip.label[tip]
+ ## Fix by Ludovic Mallet and Mahendra Mariadassou (2011-11-21):
+ name <- vector("character", length(tip))
+ name[order(tip)] <- phy$tip.label[tip]
+ phy$tip.label <- name
+ ## End of fix
## keep the ordering so no need to reorder tip.label:
phy$edge[TIPS, 2] <- order(tip)
if (!is.null(phy$node.label))
}
drop.tip <-
- function(phy, tip, trim.internal = TRUE, subtree = FALSE, root.edge = 0)
+ function(phy, tip, trim.internal = TRUE, subtree = FALSE,
+ root.edge = 0, rooted = is.rooted(phy), interactive = FALSE)
{
- if (class(phy) != "phylo")
+ if (!inherits(phy, "phylo"))
stop('object "phy" is not of class "phylo"')
- phy <- reorder(phy)
+ if (!length(tip)) return(phy)
+
Ntip <- length(phy$tip.label)
+ ## find the tips to drop:
+ if (interactive) {
+ cat("Left-click close to the tips you want to drop; right-click when finished...\n")
+ xy <- locator()
+ nToDrop <- length(xy$x)
+ tip <- integer(nToDrop)
+ lastPP <- get("last_plot.phylo", envir = .PlotPhyloEnv)
+ for (i in 1:nToDrop) {
+ d <- sqrt((xy$x[i] - lastPP$xx)^2 + (xy$y[i] - lastPP$yy)^2)
+ tip[i] <- which.min(d)
+ }
+ } else {
+ if (is.character(tip))
+ tip <- which(phy$tip.label %in% tip)
+ }
+ if (any(tip > Ntip))
+ warning("some tip numbers were higher than the number of tips")
+
+ if (!rooted && subtree) {
+ phy <- root(phy, (1:Ntip)[-tip][1])
+ root.edge <- 0
+ }
+
+ phy <- reorder(phy)
NEWROOT <- ROOT <- Ntip + 1
Nnode <- phy$Nnode
Nedge <- dim(phy$edge)[1]
edge1 <- phy$edge[, 1] # local copies
edge2 <- phy$edge[, 2] #
keep <- !logical(Nedge)
- ## find the tips to drop:
- if (is.character(tip))
- tip <- which(phy$tip.label %in% tip)
- trms <- edge2 <= Ntip
+
## delete the terminal edges given by `tip':
keep[match(tip, edge2)] <- FALSE
if (trim.internal) {
- ## delete the internal edges that do not have descendants
- ## anymore (ie, they are in the 2nd column of `edge' but
+ ints <- edge2 > Ntip
+ ## delete the internal edges that do not have anymore
+ ## descendants (ie, they are in the 2nd col of `edge' but
## not in the 1st one)
repeat {
- sel <- !(edge2 %in% edge1[keep]) & !trms & keep
+ sel <- !(edge2 %in% edge1[keep]) & ints & keep
if (!sum(sel)) break
keep[sel] <- FALSE
}
if (subtree) {
## keep the subtending edge(s):
subt <- edge1 %in% edge1[keep] & edge1 %in% edge1[!keep]
- ## <FIXME> 'if (... ' needed below?
- if (any(subt)) keep[which(subt)] <- TRUE
+ keep[subt] <- TRUE
}
if (root.edge && wbl) {
degree <- tabulate(edge1[keep])
if (!root.edge) phy$root.edge <- NULL
- ## upate the tree; 1) drop the edges and tip labels
+ ## drop the edges
phy$edge <- phy$edge[keep, ]
if (wbl) phy$edge.length <- phy$edge.length[keep]
+
+ ## find the new terminal edges (works whatever 'subtree' and 'trim.internal'):
+ TERMS <- !(phy$edge[, 2] %in% phy$edge[, 1])
+
+ ## get the old No. of the nodes and tips that become tips:
+ oldNo.ofNewTips <- phy$edge[TERMS, 2]
+
+ ## in case some tips are dropped but kept because of 'subtree = TRUE':
+ if (subtree) {
+ i <- which(tip %in% oldNo.ofNewTips)
+ if (length(i)) {
+ phy$tip.label[tip[i]] <- "[1_tip]"
+ tip <- tip[-i]
+ }
+ }
+
+ n <- length(oldNo.ofNewTips) # the new number of tips in the tree
+
+ ## the tips may not be sorted in increasing order in the
+ ## 2nd col of edge, so no need to reorder $tip.label
+ phy$edge[TERMS, 2] <- rank(phy$edge[TERMS, 2])
phy$tip.label <- phy$tip.label[-tip]
- ## 2) renumber the remaining tips now
- TIPS <- phy$edge[, 2] <= Ntip
- ## keep the ordering so no need to reorder tip.label:
- phy$edge[TIPS, 2] <- rank(phy$edge[TIPS, 2])
- ## 3) update node.label if needed
- if (!is.null(phy$node.label))
- phy$node.label <- phy$node.label[sort(unique(phy$edge[, 1])) - Ntip]
- Ntip <- length(phy$tip.label) # 4) update Ntip
- ## make new tip labels if necessary
+ ## make new tip labels if necessary:
if (subtree || !trim.internal) {
- new.trms <- !(phy$edge[, 2] %in% phy$edge[, 1]) & phy$edge[, 2] > Ntip
- node2tip <- phy$edge[new.trms, 2]
- if (subtree)
- new.lab <- paste("[", N[node2tip], "_tips]", sep = "")
- else {
- new.lab <-
- if (is.null(phy$node.label)) rep("NA", length(node2tip))
- else phy$node.label[node2tip - Ntip]
+ ## get the numbers of the nodes that become tips:
+ node2tip <- oldNo.ofNewTips[oldNo.ofNewTips > Ntip]
+ new.tip.label <- if (subtree) {
+ paste("[", N[node2tip], "_tips]", sep = "")
+ } else {
+ if (is.null(phy$node.label)) rep("NA", length(node2tip))
+ else phy$node.label[node2tip - Ntip]
}
- ## change the #'s in the edge matrix
- new.tip <- Ntip + 1:length(node2tip)
- phy$edge[new.trms, 2] <- new.tip
- phy$tip.label[new.tip] <- new.lab
- Ntip <- length(phy$tip.label)
if (!is.null(phy$node.label))
phy$node.label <- phy$node.label[-(node2tip - Ntip)]
+ phy$tip.label <- c(phy$tip.label, new.tip.label)
}
- phy$Nnode <- dim(phy$edge)[1] - Ntip + 1L # 3) update phy$Nnode
+
+ ## update node.label if needed:
+ if (!is.null(phy$node.label))
+ phy$node.label <- phy$node.label[sort(unique(phy$edge[, 1])) - Ntip]
+
+ phy$Nnode <- dim(phy$edge)[1] - n + 1L # update phy$Nnode
## The block below renumbers the nodes so that they conform
## to the "phylo" format -- same as in root()
- newNb <- integer(Ntip + phy$Nnode)
- newNb[NEWROOT] <- Ntip + 1L
- sndcol <- phy$edge[, 2] > Ntip
+ newNb <- integer(n + phy$Nnode)
+ newNb[NEWROOT] <- n + 1L
+ sndcol <- phy$edge[, 2] > n
## executed from right to left, so newNb is modified before phy$edge:
phy$edge[sndcol, 2] <- newNb[phy$edge[sndcol, 2]] <-
- (Ntip + 2):(Ntip + phy$Nnode)
+ (n + 2):(n + phy$Nnode)
phy$edge[, 1] <- newNb[phy$edge[, 1]]
storage.mode(phy$edge) <- "integer"
collapse.singles(phy)
}
+