1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
|
\name{node.dating}
\alias{node.dating}
\alias{estimate.mu}
\alias{estimate.dates}
\title{node.dating}
\description{
Estimate the dates of a rooted phylogenetic tree from the tip dates.
}
\usage{
estimate.mu(t, node.dates, p.tol = 0.05)
estimate.dates(t, node.dates, mu = estimate.mu(t, node.dates),
min.date = -.Machine$double.xmax, show.steps = 0,
opt.tol = 1e-8, nsteps = 1000,
lik.tol = 0, is.binary = is.binary.phylo(t))
}
\arguments{
\item{t}{an object of class "phylo"}
\item{node.dates}{a numeric vector of dates for the tips, in the same
order as 't$tip.label' or a vector of dates for all of the nodes.}
\item{p.tol}{p-value cutoff for failed regression.}
\item{mu}{mutation rate.}
\item{min.date}{the minimum bound on the dates of nodes}
\item{show.steps}{print the log likelihood every show.steps. If 0 will
supress output.}
\item{opt.tol}{tolerance for optimization precision.}
\item{lik.tol}{tolerance for likelihood comparison.}
\item{nsteps}{the maximum number of steps to run.}
\item{is.binary}{if TRUE, will run a faster optimization method that
only works if the tree is binary; otherwise will use optimize() as
the optimization method.}
}
\value{
The estimated mutation rate as a numeric vector of length one for estimate.mu.
The estimated dates of all of the nodes of the tree as a numeric vector with
length equal to the number of nodes in the tree.
}
\details{
This code duplicates the functionality of the program Tip.Dates (see references).
The dates of the internal nodes of 't' are estimated using a maximum likelihood
approach.
't' must be rooted and have branch lengths in units of expected substitutions per
site.
'node.dates' can be either a numeric vector of dates for the tips or a numeric
vector for all of the nodes of 't'. 'estimate.mu' will use all of the values
given in 'node.dates' to estimate the mutation rate. Dates can be censored with
NA. 'node.dates' must contain all of the tip dates when it is a parameter of
'estimate.dates'. If only tip dates are given, then 'estimate.dates' will run an
initial step to estimate the dates of the internal nodes. If 'node.dates'
contains dates for some of the nodes, 'estimate.dates' will use those dates as
priors in the inital step. If all of the dates for nodes are given, then
'estimate.dates' will not run the inital step.
If 'is.binary' is set to FALSE, 'estimate.dates' uses the "optimize" function as
the optimization method. By default, R's "optimize" function uses a precision
of ".Machine$double.eps^0.25", which is about 0.0001 on a 64-bit system. This
should be set to a smaller value if the branch lengths of 't' are very short. If
'is.binary' is set to TRUE, estimate dates uses calculus to deterimine the maximum
likelihood at each step, which is faster. The bounds of permissible values are
reduced by 'opt.tol'.
'estimate.dates' has several criteria to decide how many steps it will run. If
'lik.tol' and 'nsteps' are both 0, then 'estimate.dates' will only run the initial
step. If 'lik.tol' is greater than 0 and 'nsteps' is 0, then 'estimate.dates'
will run until the difference between successive steps is less than 'lik.tol'. If
'lik.tol' is 0 and 'nsteps' is greater than 0, then 'estimate.dates' will run the
inital step and then 'nsteps' steps. If 'lik.tol' and 'nsteps' are both greater
than 0, then 'estimate.dates' will run the inital step and then either 'nsteps'
steps or until the difference between successive steps is less than 'lik.tol'.
}
\note{
This model assumes that the tree follows a molecular clock. It only performs a
rudimentary statistical test of the molecular clock hypothesis.
}
\author{Bradley R. Jones <email: brj1@sfu.ca>}
\references{
Felsenstein, J. (1981) Evolutionary trees from DNA sequences: a maximum likelihood
approach. \emph{Journal of Molecular Evolution}, \bold{17}, 368--376.
Rambaut, A. (2000) Estimating the rate of molecular evolution:
incorporating non-contemporaneous sequences into maximum likelihood
phylogenies. \emph{Bioinformatics}, \bold{16}, 395--399.
Jones, Bradley R., and Poon, Art F. Y. (2016)
node.dating: dating ancestors in phylogenetic trees in R
\emph{Bioinformatics}, \bold{33}, 932--934.
}
\seealso{
\code{\link[stats]{optimize}, \link{rtt}},
\code{\link{plotTreeTime}}
}
\examples{
t <- rtree(100)
tip.date <- rnorm(t$tip.label, mean = node.depth.edgelength(t)[1:Ntip(t)])^2
t <- rtt(t, tip.date)
mu <- estimate.mu(t, tip.date)
## Run for 100 steps
node.date <- estimate.dates(t, tip.date, mu, nsteps = 100)
## Run until the difference between successive log likelihoods is
## less than $10^{-4}$ starting with the 100th step's results
node.date <- estimate.dates(t, node.date, mu, nsteps = 0, lik.tol = 1e-4)
## To rescale the tree over time
t$edge.length <- node.date[t$edge[, 2]] - node.date[t$edge[, 1]]
}
\keyword{model}
|
