rank.partitionedRAD: Bin trees into 'suppported' or 'disfavored' by locus.

Description

This function takes output from match.lnL.to.trees and scores trees as favored or disfavored for all loci that satisfy the criteria specified

Usage

rank.partitionedRAD(radMat, minTrees = 10,
	                  min.overall.diff.lnL = 5,
		threshold.lnL = 2,
                    discardDoubleCounts = TRUE)

Arguments

radMat

matrix output from match.lnL.to.trees

minTrees

integer, minimum number of unique trees required per locus

min.overall.diff.lnL

minimum log-likelihood difference required between the most poorly- supported tree and the best supported tree for each locus

threshold.lnL

the log-likelihood window for binning a tree into favored or disfavored

discardDoubleCounts

determines whether that place any tree within both the favored and disfavored bins should be counted (TRUE) or not (FALSE)

Value

A list composed of 7 matrices and one numeric vector. The matrices all have loci as the rows and trees as the columns:

bestMat

boolean; which trees are supported by which loci

worstMat

boolean; which trees are disfavored by which loci

doubleCountMat

boolean; which trees are double-counted by which loci

radMat

the final likelihood matrix corresponding with bestMat and worstMat

radMat.preLnLDiff

the likelihood matrix before filtering by lnL range

radMat.preMinTrees

the likelihood matrix before filtering by minimum number of trees

radMat.preDoubleCounts

the likelihood matrix before filtering by double counts

params

the analysis parameters, in this order: minTrees, min.overall.diff.lnL, threshold.lnL, discardDoubleCounts

Details

This function works by first filtering loci by the minimum number of trees and the overall range in log-likelihood, then binning trees into supported if they are within threshold.lnL of the best-supported tree for each locus, and disfavored if they are within threshold.lnL of the least supported tree for each locus. Overlapping trees can occur when the range in log-likelihood is set to less than twice the threshold log-likelihood or quite near it.

References