1 ## dist.topo.R (2008-07-18)
3 ## Topological Distances, Tree Bipartitions,
4 ## Consensus Trees, and Bootstrapping Phylogenies
6 ## Copyright 2005-2008 Emmanuel Paradis
8 ## This file is part of the R-package `ape'.
9 ## See the file ../COPYING for licensing issues.
11 dist.topo <- function(x, y, method = "PH85")
13 if (method == "BHV01" && (is.null(x$edge.length) || is.null(y$edge.length)))
14 stop("trees must have branch lengths for Billera et al.'s distance.")
15 n <- length(x$tip.label)
16 bp1 <- .Call("bipartition", x$edge, n, x$Nnode, PACKAGE = "ape")
17 bp1 <- lapply(bp1, function(xx) sort(x$tip.label[xx]))
18 bp2 <- .Call("bipartition", y$edge, n, y$Nnode, PACKAGE = "ape")
19 bp2 <- lapply(bp2, function(xx) sort(y$tip.label[xx]))
22 if (method == "PH85") {
26 if (identical(all.equal(bp1[[i]], bp2[[j]]), TRUE)) {
32 dT <- if (q1 == q2) 2*(q1 - p) else 2*(min(q1, q2) - p) + abs(q1 - q2)
34 if (method == "BHV01") {
41 if (identical(bp1[[i]], bp2[[j]])) {
42 dT <- dT + abs(x$edge.length[which(x$edge[, 2] == n + i)] -
43 y$edge.length[which(y$edge[, 2] == n + j)])
44 found1 <- found2[j] <- TRUE
48 if (found1) found1 <- FALSE
49 else dT <- dT + x$edge.length[which(x$edge[, 2] == n + i)]
52 dT <- dT + sum(y$edge.length[y$edge[, 2] %in% (n + which(!found2))])
57 .compressTipLabel <- function(x)
59 ## 'x' is a list of objects of class "phylo" possibly with no class
60 if (!is.null(attr(x, "TipLabel"))) return(x)
61 ref <- x[[1]]$tip.label
62 if (any(table(ref) != 1))
63 stop("some tip labels are duplicated in tree no. 1")
65 for (i in 2:length(x)) {
66 if (identical(x[[i]]$tip.label, ref)) next
67 ilab <- match(x[[i]]$tip.label, ref)
68 ## can use tabulate here because 'ilab' contains integers
69 if (any(tabulate(ilab) > 1))
70 stop(paste("some tip labels are duplicated in tree no.", i))
72 stop(paste("tree no.", i, "has different tip labels"))
73 ie <- match(1:n, x[[i]]$edge[, 2])
74 x[[i]]$edge[ie, 2] <- ilab
76 for (i in 1:length(x)) x[[i]]$tip.label <- NULL
77 attr(x, "TipLabel") <- ref
81 prop.part <- function(..., check.labels = TRUE)
84 if (length(obj) == 1 && class(obj[[1]]) != "phylo")
87 ## class(obj) <- NULL # needed?
90 if (ntree == 1) check.labels <- FALSE
91 if (check.labels) obj <- .compressTipLabel(obj)
92 for (i in 1:ntree) storage.mode(obj[[i]]$Nnode) <- "integer"
94 ## The 1st must have tip labels
95 ## Maybe simply pass the number of tips to the C code??
96 if (!is.null(attr(obj, "TipLabel")))
97 for (i in 1:ntree) obj[[i]]$tip.label <- attr(obj, "TipLabel")
99 clades <- .Call("prop_part", obj, ntree, TRUE, PACKAGE = "ape")
100 attr(clades, "number") <- attr(clades, "number")[1:length(clades)]
101 attr(clades, "labels") <- obj[[1]]$tip.label
102 class(clades) <- "prop.part"
106 print.prop.part <- function(x, ...)
108 if (is.null(attr(x, "labels"))) {
109 for (i in 1:length(x)) {
110 cat("==>", attr(x, "number")[i], "time(s):")
111 print(x[[i]], quote = FALSE)
114 for (i in 1:length(attr(x, "labels")))
115 cat(i, ": ", attr(x, "labels")[i], "\n", sep = "")
117 for (i in 1:length(x)) {
118 cat("==>", attr(x, "number")[i], "time(s):")
119 print(x[[i]], quote = FALSE)
124 summary.prop.part <- function(object, ...) attr(object, "number")
126 plot.prop.part <- function(x, barcol = "blue", leftmar = 4, ...)
128 if (is.null(attr(x, "labels")))
129 stop("cannot plot this partition object; see ?prop.part for details.")
131 n <- length(attr(x, "labels"))
132 layout(matrix(1:2, 2, 1), heights = c(1, 3))
133 par(mar = c(0.1, leftmar, 0.1, 0.1))
134 plot(1:L, attr(x, "number"), type = "h", col = barcol, xlim = c(1, L),
135 xlab = "", ylab = "Frequency", xaxt = "n", bty = "n")
136 plot(0, type = "n", xlim = c(1, L), ylim = c(1, n),
137 xlab = "", ylab = "", xaxt = "n", yaxt = "n")
138 for (i in 1:L) points(rep(i, length(x[[i]])), x[[i]], ...)
139 mtext(attr(x, "labels"), side = 2, at = 1:n, las = 1)
142 prop.clades <- function(phy, ..., part = NULL)
146 if (length(obj) == 1 && class(obj[[1]]) != "phylo")
147 obj <- unlist(obj, recursive = FALSE)
148 part <- prop.part(obj, check.labels = TRUE)
150 bp <- .Call("bipartition", phy$edge, length(phy$tip.label),
151 phy$Nnode, PACKAGE = "ape")
152 if (!is.null(attr(part, "labels")))
153 for (i in 1:length(part))
154 part[[i]] <- sort(attr(part, "labels")[part[[i]]])
155 bp <- lapply(bp, function(xx) sort(phy$tip.label[xx]))
156 n <- numeric(phy$Nnode)
157 for (i in 1:phy$Nnode) {
158 for (j in 1:length(part)) {
159 if (identical(all.equal(bp[[i]], part[[j]]), TRUE)) {
160 n[i] <- attr(part, "number")[j]
169 boot.phylo <- function(phy, x, FUN, B = 100, block = 1, trees = FALSE)
172 if (class(x) == "DNAbin") x <- as.matrix(x)
178 x <- matrix(x, n, nL/n, byrow = TRUE)
182 boot.tree <- vector("list", B)
185 y <- seq(block, ncol(x), block)
186 boot.i <- sample(y, replace = TRUE)
187 boot.samp <- numeric(ncol(x))
188 boot.samp[y] <- boot.i
189 for (j in 1:(block - 1))
190 boot.samp[y - j] <- boot.i - j
191 } else boot.samp <- sample(ncol(x), replace = TRUE)
192 boot.tree[[i]] <- FUN(x[, boot.samp])
194 for (i in 1:B) storage.mode(boot.tree[[i]]$Nnode) <- "integer"
195 storage.mode(phy$Nnode) <- "integer"
196 ans <- attr(.Call("prop_part", c(list(phy), boot.tree),
197 B + 1, FALSE, PACKAGE = "ape"), "number") - 1
198 if (trees) ans <- list(BP = ans, trees = boot.tree)
202 consensus <- function(..., p = 1, check.labels = TRUE)
204 foo <- function(ic, node) {
206 ## node: node number in the final tree
209 for (j in (ic + 1):m) {
210 wh <- match(pp[[j]], pool)
211 if (!any(is.na(wh))) {
212 edge[pos, 1] <<- node
214 edge[pos, 2] <<- nextnode <<- nextnode + 1L
222 ind <- pos:(pos + size - 1)
223 edge[ind, 1] <<- node
224 edge[ind, 2] <<- pool
229 if (length(obj) == 1) {
230 ## better than unlist(obj, recursive = FALSE)
231 ## because "[[" keeps the class of 'obj':
233 if (class(obj) == "phylo") return(obj)
235 if (!is.null(attr(obj, "TipLabel")))
236 labels <- attr(obj, "TipLabel")
238 labels <- obj[[1]]$tip.label
239 if (check.labels) obj <- .compressTipLabel(obj)
242 ## Get all observed partitions and their frequencies:
243 pp <- prop.part(obj, check.labels = FALSE)
244 ## Drop the partitions whose frequency is less than 'p':
245 pp <- pp[attr(pp, "number") >= p * ntree]
246 ## Get the order of the remaining partitions by decreasing size:
247 ind <- sort(unlist(lapply(pp, length)), decreasing = TRUE,
248 index.return = TRUE)$ix
252 edge <- matrix(0L, n + m - 1, 2)
261 structure(list(edge = edge, tip.label = labels,
262 Nnode = m), class = "phylo")