resolveCycle: Resolve a node on a tree of blobs as a cycle
In MSCquartets: Analyzing Gene Tree Quartets under the Multi-Species Coalescent

resolveCycle

R Documentation

Resolve a node on a tree of blobs as a cycle

Description

Given a Tree of Blobs and quartet Concordance Factor data, resolve a specific polytomy to a cycle. Resolution is performed by finding a least-squares best-fit of an empirical distance to an expected distance related to the cycle, as described in \insertCiteABRW24;textualMSCquartets.

Usage

resolveCycle(
  ToB,
  node,
  pTable,
  test = "T3",
  alpha,
  beta,
  distance = "NANUQ",
  hdegree = 10,
  plot = TRUE,
  delta = 10^-6
)

Arguments

`ToB`	an unrooted tree of blobs (phylo object), as determined by TINNIK or another method, with multifurcations labelled by `labelIntNodes`
`node`	number of an internal node to be resolved
`pTable`	a table of qcCFs, with columns p_star and p_test
`test`	either "T3" or "cut", indicating the test to use for determining what qcCFs indicate hybridization
`alpha`	test level for p_test
`beta`	test level for p_star
`distance`	cycle resolution distance to be used; one of "NANUQ" or "modNANUQ"
`hdegree`	resolve a multifurcation of this degree or larger by a heuristic method; must be at least 5
`plot`	if FALSE (0), no plots; if TRUE (>0), make plots of resolved cycle(s) considered best and histogram of measure of fit for all hybrid/orders considered
`delta`	cutoff for relative difference in squared residuals and smallest, (RSS-minRSS)/minRSS, for determining near ties as "best" fit resolutions

Details

Possible distances to use are the NANUQ distance and a modified NANUQ distance that weights quartet trees differently, but from which the cycle structure is still identifiable.

For multifucations of degree less than a designated cutoff, all possible circular orders and choices of hybrid nodes are considered in choosing the best. Above that cutoff, a heuristic method built on the modified NANUQ distance is used to obtain a small number of orders likely to be good fits, with the least-squares fitting applied only to those.

Value

a list of resolution information, given as a list of:

$node node number,
$cycleRes list [[1]]-[[k]] of best resolutions,
$RSSs RSSs from all cycle resolutions considered in choosing best.

Each resolution is itself a 5-element list with entries:

$cycleNet Newick network with 1 cycle (with all edge lengths 1)
$cycleRSS RSS for cycle,
$taxonGroups taxon groups for cycle,
$order order of groups around cycle,
$nonRootEdges logical vector indicating edges of ToB where root cannot be.

(Items $taxonGroups,$order,$nonRootEdges are needed to combine resolutions to form networks with multiple cycles using combineCycleResolutions, and otherwise may not be of interest to users).

References

\insertRef

ABRW24MSCquartets

Examples

data(pTableYeastRokas)
out=TINNIK(pTableYeastRokas, alpha=.01, beta=.05)
ToB=labelIntNodes(out$ToB)
resolveCycle(ToB, node=9, pTable=out$pTable, alpha=.01, beta=.05, distance="NANUQ")

MSCquartets documentation built on Oct. 31, 2024, 1:08 a.m.