clusterplot {LSD}

Visualize two-dimensional data clusters

Description

Depict a numeric matrix or list utilizing the underlying distribution quantiles of one dimension in a color encoded fashion.

Usage

clusterplot(
  input,
  label = NULL,
  at = NULL,
  main = NULL,
  xlim = NULL,
  ylim = NULL,
  xlabels = NULL,
  fromto = c(0.05, 0.95),
  colpal = "standardheat",
  simulate = FALSE,
  daltonize = FALSE,
  cvd = "p",
  nrcol = 25,
  outer.col = "lightgrey",
  quartiles.col = c("grey", "black", "grey"),
  add.quartiles = TRUE,
  separate = TRUE,
  rev = FALSE,
  size = TRUE,
  alpha = NULL,
  axes = TRUE,
  ...
)

Arguments

input

matrix or list with numerical entries.

label

a character vector assigning rows/elements of 'input' to clusters (if specified, multiple clusters can be depicted in different colors and/or subsequent plots).

at

a integer vector containing the x-positions corresponding to the columns of 'input'.

main

title(s) of the plot, standard graphics parameter.

xlim

x limits, standard graphics parameter.

ylim

y limits, standard graphics parameter.

xlabels

a character vector containing labels for the x-axis.

fromto

a numeric vector containing the range of quantiles (between 0 and 1) to be plotted.

colpal

a character vector containing R built-in color names or a name of a LSD colorpalette as a character string (see disco() or disco).

simulate

logical: if TRUE (FALSE by default), a converted colorpalette is used to simulate dichromat vision according to http://www.daltonize.org (see daltonize).

daltonize

logical: if TRUE (FALSE by default), a converted colorpalette is used to enhance dichromat vision according to http://www.daltonize.org (see daltonize).

cvd

character string implying the type of color vision deficiency ("p" for protanope, "d" for deuteranope or "t" for tritanope).

nrcol

a non-negative integer specifying the number of colors to be used (defaults to 25, if not specified).

outer.col

R built-in color to be used for outlier lines (lines outside of 'fromto').

quartiles.col

a character vector containing three R built-in colors for quartile lines (c('0.25','0.5','0.75')).

add.quartiles

logical: if TRUE (by default), lines are plotted corresponding to the quartiles.

separate

if TRUE (by default), different clusters are depicted in subsequent plots.

rev

logical: if TRUE (FALSE by default), a reversed colorpalette is used.

size

logical: if TRUE (by default), the size of each cluster is added to the title of the respective plot.

alpha

alpha value: a two-digit integer between 01 and 99 for color opacity, i.e. appearance of partial or full transparency (usage omitted by default).

axes

logical: if TRUE (by default), a box and axes are added to the plot (if FALSE, custom specification of axes can be achieved via basic R graphics functions).

...

additional parameters to be passed to points and plot.

Author(s)

Achim Tresch, Bjoern Schwalb

See Also

singleclusterplot, align, demotour, disco, colorpalette

Examples

samples = 100
probes = 75
at = 1:probes
clus = matrix(rnorm(probes*samples,sd=1),ncol=probes)

clus = rbind(
	t(t(clus)+sin(1:probes/10))+1:nrow(clus)/samples,
	t(t(clus)+sin(pi/2+1:probes/10))+1:nrow(clus)/samples)

quartiles.col = c("transparent","black","transparent")
colpal = c("standardheat","crazyblue","crazyred","crazygreen")

labs = paste("cluster",kmeans(clus,4)$cluster)
clusterplot(clus,fromto=c(0,1))

clusterplot(clus,labs,separate=FALSE,xaxt="n",fromto=c(0.4,0.6),colpal=colpal,
	outer.col="none",ylim=c(-2,3),quartiles.col = quartiles.col)
 
clusterplot(clus,labs,colpal=colpal)

labs = paste("cluster",kmeans(clus,2)$cluster)
colpal = c("greens","purples")
clusterplot(clus,labs,separate=FALSE,xaxt="n",fromto=c(0.3,0.7),colpal=colpal,
	outer.col="none",ylim=c(-1,2),alpha=50,quartiles.col = quartiles.col)