TreeDist (version 2.15.0)

NNIMaxStep: Largest clustering information distance reachable by one nearest neighbour interchange

Description

NNIMaxStep() returns the largest Clustering Information Distance that can separate an n-leaf tree from any tree that differs from it by a single nearest neighbour interchange (NNI) move.

Usage

NNIMaxStep(tree, normalize = FALSE)

Value

NNIMaxStep() returns a numeric vector, one entry per tree (or leaf count), giving the largest attainable distance, in bits when normalize = FALSE, or as a fraction in the range [0, 1] when normalize = TRUE. NA is returned where n < 4, as no NNI move exists.

The vector bears attributes "subtrees", giving the sizes of the four subtrees around the moved edge, and "splits", the sizes of the two splits that the move exchanges. Where more than one tree is supplied, each attribute is a list with one entry per tree.

Arguments

tree

Tree of class phylo, or list of trees of class list or multiPhylo, or an integer specifying the number of leaves in a tree.

normalize

Logical specifying whether to normalize the distance against the summed clustering information of the two trees.

Author

Martin R. Smith ()

See Also

The distance itself: ClusteringInfoDistance()

Diameter of the NNI metric: NNIDiameter()

Other tree distances: HierarchicalMutualInfo(), JaccardRobinsonFoulds(), KendallColijn(), MASTSize(), MatchingSplitDistance(), NNIDist(), NyeSimilarity(), PathDist(), Robinson-Foulds, SPRDist(), TransferDist(), TreeDistance()

Examples

# Largest clustering information distance from a single NNI move
NNIMaxStep(8)  # exactly two bits for any multiple of four
NNIMaxStep(6)  # a little less otherwise

# Read off the maximizing local topology
m6 <- NNIMaxStep(6)
attr(m6, "subtrees")
attr(m6, "splits")

# Vectorized over leaf counts
NNIMaxStep(4:8)

# Computed for a given tree
library("TreeTools", quietly = TRUE)
NNIMaxStep(BalancedTree(19))

# Normalized
NNIMaxStep(12, normalize = TRUE)