aCGH (version 1.50.0)

clusterGenome: clustering and heatmap

Description

This function clusters samples and shows their heatmap

Usage

clusterGenome(aCGH.obj, response = as.factor(rep("All", ncol(aCGH.obj))), chrominfo = human.chrom.info.Jul03, cutoff=1, lowCol = "red", highCol = "green", midCol = "black", ncolors = 50, byclass = FALSE, showaber = FALSE, amplif = 1, homdel = -0.75, samplenames = sample.names(aCGH.obj), vecchrom = 1:23, titles = "Image Plot", methodS = "ward", dendPlot = TRUE, imp = TRUE, categoricalPheno = TRUE)

Arguments

aCGH.obj

object of class aCGH here

response

phenotype of interest. defaults to the same phenotype assigned to all samples

chrominfo

a chromosomal information associated with the mapping of the data

cutoff

maximum absolute value. all the values are floored to +/-cutoff depending on whether they are positive of negative. defaults to 1

ncolors

number of colors in the grid. input to maPalette. defaults to 50

lowCol

color for the low (negative) values. input to maPalette. defaults to "red"

highCol

color for the high (positive) values. input to maPalette. defaults to "green"

midCol

color for the values close to 0. input to maPalette. defaults to "black"

byclass

logical indicating whether samples should be clustered within each level of the phenotype or overall. defaults to F

showaber

logical indicating whether high level amplifications and homozygous deletions should be indicated on the plot. defaults to F

amplif

positive value that all observations equal or exceeding it are marked by yellow dots indicating high-level changes. defaults to 1

homdel

negative value that all observations equal or below it are marked by light blue dots indicating homozygous deletions. defaults to -0.75

samplenames

sample names

vecchrom

vector of chromosomal indeces to use for clustering and to display. defaults to 1:23

titles

plot title. defaults to "Image Plots"

methodS

clustering method to cluster samples. defaults to "ward"

dendPlot

logical indicating whether dendogram needs to be drawn. defaults to T.

imp

logical indicating whether imputed or original values should be used. defaults to T, i.e. imputed.

categoricalPheno

logical indicating whether phenotype is categorical. Continious phenotypes are treated as "no groups" except that their values are dispalyed.defaults to TRUE.

Details

This functions is a more flexible version of the heatmap. It can cluster within levels of categorical phenotype as well as all of the samples while displaying phenotype levels in different colors. It also uses any combination of chromosomes that is requested and clusters samples based on these chromosomes only. It draws the chromosomal boundaries and displays high level changes and homozygous deletions. If phenotype if not categical, its values may still be displayed but groups are not formed and byclass = F. Image plot has the samples reordered according to clustering order.

See Also

aCGH heatmap

Examples

data(colorectal)

#cluster all samples using imputed data on all chromosomes (autosomes and X):

clusterGenome(colorectal)

#cluster samples within sex groups based on 3 chromosomes individually. 
#use non-imputed data and  do not show dendogram. Indicate amplifications and 
#homozygous deletions.

clusterGenome(colorectal, response = phenotype(colorectal)$sex,
                   byclass = TRUE, showaber = TRUE, vecchrom = c(4,8,9),
                   dendPlot = FALSE, imp = FALSE)

#cluster samples based on each chromosome individualy and display age. Show
#gains in red and losses in green. Show aberrations and use values < -1
#to identify homozgous deletions. Do not show dendogram.

pdf("plotimages.pdf", width = 11, height = 8.5)
for (i in 1:23)
    clusterGenome(colorectal,
                       response = phenotype(colorectal)$age,
                       chrominfo = human.chrom.info.Jul03,
                       cutoff = 1, ncolors = 50, lowCol="green",
                       highCol="red", midCol="black", byclass = FALSE,
                       showaber = TRUE, homdel = -1, vecchrom = i,
                       titles = "Image Plot", methodS = "ward",
                       dendPlot = FALSE, categoricalPheno = FALSE)
dev.off()