Package 'fastnntr'

Compute a NeighbourNet split network

Description

Runs the NeighbourNet algorithm (Bryant & Moulton, 2004) on a distance matrix and returns the resulting split network in a networx-compatible list, ready to plot with tanggle/ggplot2 or base phangorn.

Usage

run_neighbornet_networkx(
  x,
  flip_y = TRUE,
  labels = NULL,
  max_iterations = 5000,
  ordering_method = NULL,
  inference_method = NULL
)

run_neighbournet_networkx(
  x,
  flip_y = TRUE,
  labels = NULL,
  max_iterations = 5000,
  ordering_method = NULL,
  inference_method = NULL
)

Arguments

x	A distance matrix. May be a dist object, a numeric matrix, or a data.frame/data.table of distances; non-matrix inputs are coerced with as.matrix(). The matrix must be square and symmetric.
flip_y	Logical; if TRUE (default) the vertical axis of the computed layout is flipped, matching the orientation used by SplitsTree.
labels	Optional character vector of taxon labels. If NULL (default) the column names of x are used (and, for a dist object, its labels).
max_iterations	Integer; maximum NNLS iterations for split-weight estimation (default 5000).
ordering_method	Split ordering algorithm: "multiway" (default) or "closest-pair". NULL uses the default.
inference_method	Split-weight solver: "active-set" (default) or "cg". NULL uses the default.

Details

NeighbourNet produces an implicit (split) network: a planar diagram that summarises conflicting signal in the data for exploratory analysis. Unlike explicit networks, the boxes do not represent specific reticulation events such as hybridisation or introgression.

run_neighbournet_networkx() is an identical spelling alias of run_neighbornet_networkx().

Value

A list of class c("networx", "phylo") containing edge, tip.label, edge.length, Nnode, splitIndex, splits, translate, and a .plot sublist whose vertices matrix holds 2-D node coordinates from the equal-angle layout. The structure is compatible with phangorn's networx objects and tanggle's ggplot2 layers.

See Also

tanggle and phangorn for plotting networx objects; stats::dist() and phangorn::dist.ml() for computing distances.

Examples

# A minimal workflow from raw distances to a plotted network.
d <- dist(matrix(rnorm(50), nrow = 5, dimnames = list(LETTERS[1:5], NULL)))
net <- run_neighbornet_networkx(d)
str(net$tip.label)

# Plot with tanggle/ggplot2 (if installed):
if (requireNamespace("tanggle", quietly = TRUE)) {
  library(tanggle)
  ggplot2::ggplot(net) + geom_splitnet() + geom_tiplab2()
}

---

Set the global thread count for fastnnt

Description

Configures the size of the internal Rayon thread pool used for split-weight estimation. The pool can only be initialised once per R session; subsequent calls emit a warning and leave the existing pool unchanged.

Usage

set_fastnnt_threads(threads)

Arguments

threads

Integer (>= 1); number of worker threads to use.

Value

Invisibly NULL; called for its side effect.

Examples

set_fastnnt_threads(4)

---