mcmc.target.seq: Find MCMC Target Sequence

Description

mcmc.target.seq uses MCMC to find a configuration of DNA positions to get as close as possible to a given tree.

boot.samples.idxs bootstraps over indices into a DNA matrix.

lookup.samples goes from an index representation of a configuration of DNA to the actual DNAbin format.

convert.table.to.idx converts a table of counts for positions 1..n into a list of indices corresponding to positions (i.e. goes from the tabled form to a vector whose tabling matches the input).

Usage

mcmc.target.seq(data, x, F, n)
boot.samples.idxs(data, B = 100, block = 1)
lookup.samples(data, idxs)
convert.table.to.idx(T)

Arguments

data

A DNA matrix in DNAbin format.

A tree of class 'phylo' to estimate.

A tree estimation function, accepting a DNA matrix in DNAbin format and returning a tree of class 'phylo.'

The number of MCMC iterations to perform.

The number of bootstrap replicates.

block

The block size to use during bootstrapping.

idxs

A list of numeric vectors of indices to use for lookup.

A table or table-like vector to convert.

Value

mcmc.target.seq returns a list of 4 elements: a numeric vector of counts of each position in the original matrix, the best estimated tree, the best distance from the estimated tree to the target tree, and a numeric vector of the distances for every iteration of the simulation.

boot.samples.idxs returns a numeric vector representing the bootstrapped idices.

lookup.samples returns a list of objects of class DNAbin corresponding to the DNA sequences generated from indices into the original DNA matrix.

convert.table.to.idx returns a numeric vector of indices based on the table counts.

Details

mcmc.target.seq performs an MCMC with simulated annealing to locate a configuration of DNA positions from the original matrix that gets as close as possible to a target tree. Propositions for the MCMC replacing one character with another uniformly at random.

The remaining functions are intended to be used as support functions.

References

Chakerian, J. and Holmes, S. P. Computational Tools for Evaluating Phylogenetic and Heirarchical Clustering Trees. arXiv:1006.1015v1.

Examples

Run this code

# NOT RUN {
## This example has been excluded from checks:
## copy/paste the code to try it

data(woodmouse)
otree <- root(fastme.ols(dist.dna(woodmouse)), "No305", resolve.root=TRUE)
breps <- 200

trees <- boot.phylo(otree, woodmouse, B=breps, function(x)
        root(fastme.ols(dist.dna(x)), "No305", resolve.root=TRUE),
        trees = TRUE)

combined.trees <- c(list(otree), trees$trees)

binning <- bin.multiPhylo(combined.trees)

tree.a <- combined.trees[[match(1, binning)]]
i <- 2
max.bin <- max(binning)
tree.b <- combined.trees[[match(2, binning)]]

while(length(distinct.edges(tree.a,tree.b)) > 1 && i < max.bin)
{
    i = i + 1
    tree.b = combined.trees[[match(i, binning)]]
}

bdy.tree <- orthant.boundary.tree(tree.a, tree.b)

f.est <- function(x) root(nj(dist.dna(x)), "No305", resolve.root=TRUE)

res <- mcmc.target.seq(woodmouse, bdy.tree, f.est, 1000)

par(mfrow=c(2,1))
plot(res$tree)
plot(res$vals)
# }

Run the code above in your browser using DataCamp Workspace