heatmaps

# Construct the appropriate matrices/data frames for
# construction of the heat map.
ii.mat <- exprs.eset[ii,]
ii.df <- data.frame(ii.mat)

# Dislplay ii.df
ii.df

# Construct heatmap using reds and blues
library('RColorBrewer')
library('genefilter')
hmcol <- colorRampPalette(brewer.pal(10, 'RdBu'))(256)
tv.Hawaii.Resistant <- dimnames(ii.mat)[[2]][1:3]
spcol <- ifelse(dimnames(ii.mat)[[2]] == tv.Hawaii.Resistant, 'skyblue', 'goldenrod')
heatmap(ii.mat, col = hmcol, ColSideColors = spcol, margins = c(10,15))

# Construct heatmap using a grey (gray) scale
library('RColorBrewer')
library('genefilter')
hmcol <- colorRampPalette(brewer.pal(9, 'Greys'))(256)
tv.Hawaii.Resistant <- dimnames(ii.mat)[[2]][1:3]
spcol <- ifelse(dimnames(ii.mat)[[2]] == tv.Hawaii.Resistant, 'grey10', 'grey80')
heatmap(ii.mat, col = hmcol, ColSideColors = spcol, margins = c(10,15))

(Complete File)(Rout)

Heatmaps are commonly seen in microarray analysis papers. They are generally produced once the data has been analysed and a gene list of interest (eg from limma) has been produced.

At their simplest form they are a 2 dimensional grid of colours, where the colours normally indicate a level of expression of a gene in each condition (chip) in the experiment. Often they are clustered using a hierarchical clustering method to assign similar expression profiles next to each other in the grid.

Often this will have the result (for instance) of showing that post-treatment and pre-treatment cluster together, or that time points in a time course change over time.

In reality they are not that meaningful, unless you are showing a 'per gene' clustering as well, which will cluster both genes and expression profiles which may allow you to identify blocks of genes which are potentially co-regulated.

CSC8309 -- Gene Expression and Proteomics

Heatmaps