MSTSegments: Add minimum spanning tree to plot, colouring by stress

In TreeDist: Calculate and Map Distances Between Phylogenetic Trees

Add minimum spanning tree to plot, colouring by stress

Description

To identify strain in a multidimensional scaling of distances, it can be useful to plot a minimum spanning tree \insertCiteGower1966,SmithSpaceTreeDist. Colouring each edge of the tree according to its strain can identify areas where the mapping is stretched or compressed.

Usage

MSTSegments(mapping, mstEnds, ...)

StrainCol(
  distances,
  mapping,
  mstEnds = MSTEdges(distances),
  palette = rev(hcl.colors(256L, "RdYlBu"))
)

Arguments

mapping	Two-column matrix giving x and y coordinates of plotted points.
mstEnds	Two-column matrix identifying rows of mapping at end of each edge of the MST, as output by TreeTools::MSTEdges().
...	Additional arguments to segments().
distances	Matrix or dist object giving original distances between each pair of points.
palette	Vector of colours with which to colour edges.

Value

StrainCol() returns a vector in which each entry is selected from palette, with an attribute logStrain denoting the logarithm of the mapped over original distance, shifted such that the median value is zero. Palette colours are assigned centred on the median value, with entries early in palette assigned to edges in which the ratio of mapped distance to original distance is small.

Author(s)

Martin R. Smith ([e-mail])

References

\insertAllCited

See Also

Other tree space functions: MapTrees(), MappingQuality(), SpectralEigens(), cluster-statistics, median.multiPhylo()

Examples

set.seed(0)
library("TreeTools", quietly = TRUE)
distances <- ClusteringInfoDist(as.phylo(5:16, 8))
mapping <- cmdscale(distances, k = 2)
mstEnds <- MSTEdges(distances)

# Set up blank plot
plot(mapping, asp = 1, frame.plot = FALSE, ann = FALSE, axes = FALSE,
     type = "n")
# Add MST
MSTSegments(mapping, mstEnds, 
            col = StrainCol(distances, mapping, mstEnds))
# Add points at end so they overprint the MST
points(mapping)
PlotTools::SpectrumLegend(
 "bottomleft",
 legend = c("Extended", "Median", "Contracted"),
 bty = "n",     # No box
 y.intersp = 2, # Expand in Y direction
 palette = hcl.colors(256L, "RdYlBu", rev = TRUE)
)

TreeDist documentation built on Oct. 26, 2023, 1:07 a.m.