MrBayes: Calls MrBayes

Description

A slight improvement of 'ips::mrbayes()'

Usage

MrBayes(x, file="", nst=6, rates="invgamma", ngammacat=4, nruns=2, ngen=1e+06,
 printfreq=100, samplefreq=10, nchains=4, savebrlens="yes", temp=0.2, burnin=10,
 contype="allcompat", run=FALSE, simple=TRUE, exec="mb-mpi", method="dna")

Arguments

x: The object to process (must be 'DNAbin' class)
file: A character string, giving the name of the MrBayes input file
nst: An integer giving the number of rates in the model of sequence evolution
rates: A character string; allowed are "equal", "gamma", "propinv", "invgamma", and "adgamma"; the default is "equal"
ngammacat: An integer; the number rate categories for the discretized Gamma distribution; the default is '4'
nruns: An integer; the number of runs
ngen: An integer; the number of states of the MCMC
printfreq: An integer; the interval between states of the MCMC to be printed on the screen
samplefreq: An integer; the interval between states of the MCMC to be sampled
nchains: An integer; number of Metropolis coupled MCMCs in each run
savebrlens: Logical; shall branch lengths be saved
temp: Heating parameter
burnin: An integer; the number of samples from the MCMC to be discarded prior to further analysis
contype: A character string; the type of consensus tree calculated from the posterior distribution of trees

either "halfcompat" (majority-rule consensus tree) or "allcombat" (strict consensus tree)

run: Logical; 'run = FALSE' will only print the NEXUS file, 'run = TRUE' will also start the MCMC runs, if the 'path' argument is correctly specified
simple: New option: if TRUE (default), then outputs tree in the format readable by functions from 'ape' package
exec: New option: name of UNIX executable (to allow multi-threaded version)
method: New option: either "dna", or "mixed" to handle mixed or purely morphologic data (see below)

Author

Alexey Shipunov

Details

MrBayes() is an improvement of ips::mrbayes() and ips::mrbayes.mixed(). Please see its documentation for clarity and other options.

Comparing with 'ips' sources, MrBayes() has some code alterations and three more options. It also both views and saves output (works only on UNIX).

If 'method="mixed"', the function requires character matrix as input where missing data are labeled with "N", morphological columns encoded as 0/1 and placed after nucleotide columns (which might be absent).

Examples

Run this code

require(ips)
data(ips.cox1)
x <- ips.cox1[, 100:140]

if (FALSE) {
## requires MrBayes program installation
MrBayes(x, file="cox1", ngen=100, run=TRUE)

str(plantago)
plantago[is.na(plantago)] <- "N"
row.names(plantago) <- gsub(" ", "_", row.names(plantago))
## requires MrBayes program installation
tr <- MrBayes(plantago, file="plantago", method="mixed", burnin=5000, run=TRUE) # makes many files
tr <- tr[[1]]
tr <- root(tr, outgroup="Plantago_maritima", resolve.root=TRUE)
tr$node.label <- suppressWarnings(round(as.numeric(tr$node.label)*100)) # warning is OK
tr$node.label[tr$node.label == "NA"] <- ""
plot(tr)
nodelabels(tr$node.label, frame="none", bg="transparent", adj=-0.1)
add.scale.bar()
}