1 ## drop.tip.R (2010-11-24)
3 ## Remove Tips in a Phylogenetic Tree
5 ## Copyright 2003-2010 Emmanuel Paradis
7 ## This file is part of the R-package `ape'.
8 ## See the file ../COPYING for licensing issues.
10 extract.clade <- function(phy, node, root.edge = 0, interactive = FALSE)
12 Ntip <- length(phy$tip.label)
14 Nedge <- dim(phy$edge)[1]
15 wbl <- !is.null(phy$edge.length)
16 if (interactive) node <- identify(phy)$nodes else {
17 if (length(node) > 1) {
19 warning("only the first value of 'node' has been considered")
21 if (is.character(node)) {
22 if (is.null(phy$node.label))
23 stop("the tree has no node labels")
24 node <- which(phy$node.label %in% node) + Ntip
27 stop("node number must be greater than the number of tips")
29 if (node == ROOT) return(phy)
30 phy <- reorder(phy) # insure it is in cladewise order
31 root.node <- which(phy$edge[, 2] == node)
32 start <- root.node + 1 # start of the clade looked for
33 anc <- phy$edge[root.node, 1] # the ancestor of 'node'
34 next.anc <- which(phy$edge[-(1:start), 1] <= anc) # find the next occurence of 'anc' or an 'older' node
36 keep <- if (length(next.anc)) start + 0:(next.anc[1] - 1) else start:Nedge
39 NewRootEdge <- phy$edge.length[root.node]
40 root.edge <- root.edge - 1
42 if (anc == ROOT) break
43 i <- which(phy$edge[, 2] == anc)
44 NewRootEdge <- NewRootEdge + phy$edge.length[i]
45 root.edge <- root.edge - 1
48 if (root.edge && !is.null(phy$root.edge))
49 NewRootEdge <- NewRootEdge + phy$root.edge
50 phy$root.edge <- NewRootEdge
53 phy$edge <- phy$edge[keep, ]
54 if (wbl) phy$edge.length <- phy$edge.length[keep]
55 TIPS <- phy$edge[, 2] <= Ntip
56 tip <- phy$edge[TIPS, 2]
57 phy$tip.label <- phy$tip.label[sort(tip)] # <- added sort to avoid shuffling of tip labels (2010-07-21)
58 ## keep the ordering so no need to reorder tip.label:
59 phy$edge[TIPS, 2] <- order(tip)
60 if (!is.null(phy$node.label))
61 phy$node.label <- phy$node.label[sort(unique(phy$edge[, 1])) - Ntip]
62 Ntip <- length(phy$tip.label)
63 phy$Nnode <- dim(phy$edge)[1] - Ntip + 1L
64 ## The block below renumbers the nodes so that they conform
65 ## to the "phylo" format -- same as in root()
66 newNb <- integer(Ntip + phy$Nnode)
67 newNb[node] <- Ntip + 1L
68 sndcol <- phy$edge[, 2] > Ntip
69 ## executed from right to left, so newNb is modified before phy$edge:
70 phy$edge[sndcol, 2] <- newNb[phy$edge[sndcol, 2]] <-
71 (Ntip + 2):(Ntip + phy$Nnode)
72 phy$edge[, 1] <- newNb[phy$edge[, 1]]
77 function(phy, tip, trim.internal = TRUE, subtree = FALSE,
78 root.edge = 0, rooted = is.rooted(phy), interactive = FALSE)
80 if (!inherits(phy, "phylo"))
81 stop('object "phy" is not of class "phylo"')
83 Ntip <- length(phy$tip.label)
84 ## find the tips to drop:
86 cat("Left-click close to the tips you want to drop; right-click when finished...\n")
88 nToDrop <- length(xy$x)
89 tip <- integer(nToDrop)
90 lastPP <- get("last_plot.phylo", envir = .PlotPhyloEnv)
91 for (i in 1:nToDrop) {
92 d <- sqrt((xy$x[i] - lastPP$xx)^2 + (xy$y[i] - lastPP$yy)^2)
93 tip[i] <- which.min(d)
96 if (is.character(tip))
97 tip <- which(phy$tip.label %in% tip)
100 warning("some tip numbers were higher than the number of tips")
102 if (!rooted && subtree) {
103 phy <- root(phy, (1:Ntip)[-tip][1])
108 NEWROOT <- ROOT <- Ntip + 1
110 Nedge <- dim(phy$edge)[1]
112 trim.internal <- TRUE
113 tr <- reorder(phy, "pruningwise")
114 N <- .C("node_depth", as.integer(Ntip), as.integer(Nnode),
115 as.integer(tr$edge[, 1]), as.integer(tr$edge[, 2]),
116 as.integer(Nedge), double(Ntip + Nnode),
117 DUP = FALSE, PACKAGE = "ape")[[6]]
119 wbl <- !is.null(phy$edge.length)
120 edge1 <- phy$edge[, 1] # local copies
121 edge2 <- phy$edge[, 2] #
122 keep <- !logical(Nedge)
124 ## delete the terminal edges given by `tip':
125 keep[match(tip, edge2)] <- FALSE
129 ## delete the internal edges that do not have anymore
130 ## descendants (ie, they are in the 2nd col of `edge' but
131 ## not in the 1st one)
133 sel <- !(edge2 %in% edge1[keep]) & ints & keep
138 ## keep the subtending edge(s):
139 subt <- edge1 %in% edge1[keep] & edge1 %in% edge1[!keep]
142 if (root.edge && wbl) {
143 degree <- tabulate(edge1[keep])
144 if (degree[ROOT] == 1) {
145 j <- integer(0) # will store the indices of the edges below the new root
147 i <- which(edge1 == NEWROOT & keep)
150 degree <- tabulate(edge1[keep])
151 if (degree[NEWROOT] > 1) break
154 if (length(j) > root.edge) j <- 1:root.edge
155 NewRootEdge <- sum(phy$edge.length[j])
156 if (length(j) < root.edge && !is.null(phy$root.edge))
157 NewRootEdge <- NewRootEdge + phy$root.edge
158 phy$root.edge <- NewRootEdge
163 if (!root.edge) phy$root.edge <- NULL
166 phy$edge <- phy$edge[keep, ]
167 if (wbl) phy$edge.length <- phy$edge.length[keep]
169 ## find the new terminal edges (works whatever 'subtree' and 'trim.internal'):
170 TERMS <- !(phy$edge[, 2] %in% phy$edge[, 1])
172 ## get the old No. of the nodes and tips that become tips:
173 oldNo.ofNewTips <- phy$edge[TERMS, 2]
175 ## in case some tips are dropped but kept because of 'subtree = TRUE':
177 i <- which(tip %in% oldNo.ofNewTips)
179 phy$tip.label[tip[i]] <- "[1_tip]"
184 n <- length(oldNo.ofNewTips) # the new number of tips in the tree
186 ## the tips may not be sorted in increasing order in the
187 ## 2nd col of edge, so no need to reorder $tip.label
188 phy$edge[TERMS, 2] <- rank(phy$edge[TERMS, 2])
189 phy$tip.label <- phy$tip.label[-tip]
191 ## make new tip labels if necessary:
192 if (subtree || !trim.internal) {
193 ## get the numbers of the nodes that become tips:
194 node2tip <- oldNo.ofNewTips[oldNo.ofNewTips > Ntip]
195 new.tip.label <- if (subtree) {
196 paste("[", N[node2tip], "_tips]", sep = "")
198 if (is.null(phy$node.label)) rep("NA", length(node2tip))
199 else phy$node.label[node2tip - Ntip]
201 if (!is.null(phy$node.label))
202 phy$node.label <- phy$node.label[-(node2tip - Ntip)]
203 phy$tip.label <- c(phy$tip.label, new.tip.label)
206 ## update node.label if needed:
207 if (!is.null(phy$node.label))
208 phy$node.label <- phy$node.label[sort(unique(phy$edge[, 1])) - Ntip]
210 phy$Nnode <- dim(phy$edge)[1] - n + 1L # update phy$Nnode
212 ## The block below renumbers the nodes so that they conform
213 ## to the "phylo" format -- same as in root()
214 newNb <- integer(n + phy$Nnode)
215 newNb[NEWROOT] <- n + 1L
216 sndcol <- phy$edge[, 2] > n
217 ## executed from right to left, so newNb is modified before phy$edge:
218 phy$edge[sndcol, 2] <- newNb[phy$edge[sndcol, 2]] <-
219 (n + 2):(n + phy$Nnode)
220 phy$edge[, 1] <- newNb[phy$edge[, 1]]
221 storage.mode(phy$edge) <- "integer"
222 collapse.singles(phy)