Starr (version 1.28.0)

profileplot: Vizualize clusters

Description

Visualization of a set of ``profiles'' (i.e. a consecutive series of measurements like a time series, or the DNA binding levels along different positions on a gene). The profiles are given as the rows of a (samples x positions) matrix that contains the measurements. Instead of plotting a line for each profile (row of the matrix), the q-quantiles for each position (column of the matrix) are calculated, where q runs through a set of representative quantiles. Then for each q, a line of q-quantiles is plotted along the positions. Color coding of the quantile profiles aids the interpretation of the plot: There is a color gradient from the median profile to the 0 (=min) resp. 1(=max) quantile.

Usage

profileplot(cluster, label=NULL,	at=NULL, main = "", xlim=NULL, xlab = "", xaxt = "s",xlabels = NULL, las = 1, ylim=NULL, ylab = "", fromto = c(0.05,0.95),colpal = "heat", 	nrcolors = 25, outer.col="light grey", add.quartiles = T, add = F, separate = T)

Arguments

cluster

a (samples x columns) matrix with numerical entries. Each sample row is understood as a consecutive series of measurements. Missing values are not allowed so far

label

if multiple clusters should be plotted in one diagram, the cluster labels for each item are given in this vector

at

optional vector of length ncol(cluster), default = 1:ncol(cluster). Specifies the x-values at which the positions will be plotted.

main

the title of the plot, standard graphics parameter

xlim

xlimits, standard graphics parameter

xlab

x-axis legend, standard graphics parameter

xaxt

should an x axis be plotted at all? (="n" if not), standard graphics parameter

xlabels

character vector. If specified, this text will be added at the ``at``-positions as x-axis labels.

las

direction of the xlabels text. las=1: horizontal text, las=2: vertical text

ylim

ylimits, standard graphics parameter

ylab

y-axis legend, standard graphics parameter

fromto

determines the smallest and the largest quantile that are plotted in colors, more distant values are plotted as outliers

colpal

either "red","green","blue" (predefined standard color palettes in profileplot), or a vector of colors to be used instead.

nrcolors

not very important. How many colors will the color palette contain? Usually, the default = 25 is sufficient

outer.col

color of the outlier lines, default = "light grey". For no outliers, choose outer.col="none"

add.quartiles

should the quartile lines be plotted (grey/black)? default=TRUE

add

should the profile plot be added to the current plot? Defaults to FALSE

separate

should each cluster, be plotted in a separate window? Defaults to TRUE

Examples

sampls = 100
probes = 63
at = (-31:31)*14
clus = matrix(rnorm(probes*sampls,sd=1),ncol=probes)
clus= rbind( t(t(clus)+sin(1:probes/10))+1:nrow(clus)/sampls , t(t(clus)+sin(pi/2+1:probes/10))+1:nrow(clus)/sampls )
labs = paste("cluster",kmeans(clus,4)$cluster)

profileplot(clus,main="All data",fromto=c(0,1))
profileplot(clus,label=labs,main="Clustered data",colpal=c("heat","blue","red","topo"),add.quartiles=FALSE)
profileplot(clus,main="Same data, 4 clusters in one plot\n color gradient fromto = c(0.4,0.6), no outliers plotted",label=labs,separate=FALSE,xaxt="n",fromto=c(0.4,0.6),
	colpal=c("heat","blue","red","green"),outer.col="none")