quartetTableParallel: Produce table of counts of quartets displayed on trees, in parallel for large data sets

Description

Compiles table of quartet count concordance factors (qcCFs) for topological quartets displayed on a collection of trees. Gives the same output as quartetTable, but operates in parallel.

Usage

quartetTableParallel(trees, taxonnames = NULL, epsilon = 0, numCores)

Value

an (n choose 4)x(n+4) matrix (or (random)x(n+4) matrix) encoding 4 taxon subsets of taxonnames and counts of each of the quartets 12|34, 13|24, 14|23, 1234 across the trees

Arguments

trees: multiphylo object containing un/rooted metric/topological trees
taxonnames: vector of n names of taxa of interest; if NULL then taken from taxa on trees[[1]]
epsilon: minimum for branch lengths to be treated as non-zero
numCores: number of cores to use for parallel calls

Details

The number of available cores can be determined by parallel::detectCores(). With overhead, tabulating quartets for a large data set (many taxa and/or many gene trees) on a 4-core computer using numCores=4 may require less than half the elapsed time of the sequential quartetTable.

The names in taxonnames may be any subset of those on the trees. Branch lengths of non-negative size less than or equal to epsilon are treated as zero, giving polytomies.

In the returned table, columns are labeled by taxon names and quartet names ("12|34", etc.). 1s and 0s in taxon columns indicate the taxa in a quartet. Quartet 12|34 means the first and second indicated taxa form a cherry, 13|24 means the first and third form a cherry, 14|23 means the first and fourth form a cherry, and 1234 means the quartet is unresolved.

An error occurs if any branch length is negative. Warnings are given if some of taxonnames are missing on some trees, or if some 4-taxon set is not on any tree.

If random>0, then for efficiency random should be much smaller then the number of possible 4 taxon subsets.

If the quartet counts are to be used for NANUQ, or any other routines requiring resolved quartet counts, quartetTableResolved must be run following quartetTableParallel. See example below.

Examples

Run this code

# \donttest{
gtrees=read.tree(file=system.file("extdata","dataHeliconiusMartin",package="MSCquartets"))
QT=quartetTableParallel(gtrees,numCores=2)
RQT=quartetTableResolved(QT)
pTable=NANUQ(RQT,alpha=1e-40, beta=1e-30, outfile = file.path(tempdir(), "NANUQdist"))# }