5 \alias{print.prop.part}
6 \alias{summary.prop.part}
8 \title{Tree Bipartition and Bootstrapping Phylogenies}
10 boot.phylo(phy, x, FUN, B = 100, block = 1,
11 trees = FALSE, quiet = FALSE, rooted = FALSE)
12 prop.part(..., check.labels = TRUE)
13 prop.clades(phy, ..., part = NULL, rooted = FALSE)
14 \method{print}{prop.part}(x, ...)
15 \method{summary}{prop.part}(object, ...)
16 \method{plot}{prop.part}(x, barcol = "blue", leftmar = 4, ...)
19 \item{phy}{an object of class \code{"phylo"}.}
20 \item{x}{in the case of \code{boot.phylo}: a taxa (rows) by characters
21 (columns) matrix; this may be presented as a list; in the case of
22 \code{print} and \code{plot}: an object of class \code{"prop.part"}.}
23 \item{FUN}{the function used to estimate \code{phy} (see details).}
24 \item{B}{the number of bootstrap replicates.}
25 \item{block}{the number of columns in \code{x} that will be resampled
26 together (see details).}
27 \item{trees}{a logical specifying whether to return the bootstraped
28 trees (\code{FALSE} by default).}
29 \item{quiet}{a logical: a progress bar is displayed by default.}
30 \item{rooted}{a logical specifying whether the trees should be treated
31 as rooted or not (the default).}
32 \item{\dots}{either (i) a single object of class \code{"phylo"}, (ii) a
33 series of such objects separated by commas, or (iii) a list
34 containing such objects. In the case of \code{plot} further
35 arguments for the plot (see details).}
36 \item{check.labels}{a logical specifying whether to check the labels
37 of each tree. If \code{FALSE}, it is assumed that all trees have the
38 same tip labels, and that they are in the same order (see details).}
39 \item{part}{a list of partitions as returned by \code{prop.part}; if
40 this is used then \code{\dots} is ignored.}
41 \item{object}{an object of class \code{"prop.part"}.}
42 \item{barcol}{the colour used for the bars displaying the number of
43 partitions in the upper panel.}
44 \item{leftmar}{the size of the margin on the left to display the tip
48 These functions analyse bipartitions found in a series of trees.
50 \code{prop.part} counts the number of bipartitions found in a series
51 of trees given as \code{\dots}. If a single tree is passed, the
52 returned object is a list of vectors with the tips descending from
53 each node (i.e., clade compositions indexed by node number).
55 \code{prop.clades} counts the number of times the bipartitions present
56 in \code{phy} are present in a series of trees given as \code{\dots} or
57 in the list previously computed and given with \code{part}.
59 \code{boot.phylo} performs a bootstrap analysis.
62 The argument \code{FUN} in \code{boot.phylo} must be the function used
63 to estimate the tree from the original data matrix. Thus, if the tree
64 was estimated with neighbor-joining (see \code{nj}), one maybe wants
65 something like \code{FUN = function(xx) nj(dist.dna(xx))}.
67 \code{block} in \code{boot.phylo} specifies the number of columns to
68 be resampled altogether. For instance, if one wants to resample at the
69 codon-level, then \code{block = 3} must be used.
71 Using \code{check.labels = FALSE} in \code{prop.part} decreases
72 computing times. This requires that (i) all trees have the same tip
73 labels, \emph{and} (ii) these labels are ordered similarly in all
74 trees (in other words, the element \code{tip.label} are identical in
77 The plot function represents a contingency table of the different
78 partitions (on the \emph{x}-axis) in the lower panel, and their observed
79 numbers in the upper panel. Any further arguments (\dots) are used to
80 change the aspects of the points in the lower panel: these may be
81 \code{pch}, \code{col}, \code{bg}, \code{cex}, etc. This function
82 works only if there is an attribute \code{labels} in the object.
84 The print method displays the partitions and their numbers. The
85 summary method extracts the numbers only.
88 \code{prop.clades} calls internally \code{prop.part} with the option
89 \code{check.labels = TRUE}, which may be very slow. If the trees
90 passed as \code{\dots} fulfills conditions (i) and (ii) above, then it
91 might be faster to first call, e.g., \code{pp <- prop.part(...)}, then
92 use the option \code{part}: \code{prop.clades(phy, part = pp)}.
95 \code{prop.part} returns an object of class \code{"prop.part"} which
96 is a list with an attribute \code{"number"}. The elements of this list
97 are the observed clades, and the attribute their respective
98 numbers. If the default \code{check.labels = FALSE} is used, an
99 attribute \code{"labels"} is added, and the vectors of the returned
100 object contains the indices of these labels instead of the labels
103 \code{prop.clades} and \code{boot.phylo} return a numeric vector
104 which \emph{i}th element is the number associated to the \emph{i}th
105 node of \code{phy}. If \code{trees = TRUE}, \code{boot.phylo} returns
106 a list whose first element (named \code{"BP"}) is like before, and the
107 second element (\code{"trees"}) is a list with the bootstraped
110 \code{summary} returns a numeric vector.
113 Efron, B., Halloran, E. and Holmes, S. (1996) Bootstrap confidence
114 levels for phylogenetic trees. \emph{Proceedings of the National
115 Academy of Sciences USA}, \bold{93}, 13429--13434.
117 Felsenstein, J. (1985) Confidence limits on phylogenies: an approach
118 using the bootstrap. \emph{Evolution}, \bold{39}, 783--791.
120 \author{Emmanuel Paradis}
122 \code{\link{dist.topo}}, \code{\link{consensus}}, \code{\link{nodelabels}}
126 f <- function(x) nj(dist.dna(x))
128 ### Are bootstrap values stable?
130 print(boot.phylo(tr, woodmouse, f, quiet = TRUE))
131 ### How many partitions in 100 random trees of 10 labels?...
132 TR <- replicate(100, rtree(10), FALSE)
133 pp10 <- prop.part(TR)
135 ### ... and in 100 random trees of 20 labels?
136 TR <- replicate(100, rtree(20), FALSE)
137 pp20 <- prop.part(TR)
139 plot(pp10, pch = "x", col = 2)
140 plot(pp20, pch = "x", col = 2)