Outline

• heatmap.2
• ggplot2
  • qplot()
  • ggplot()

Heatmap.2

• A tutorial: https://sebastianraschka.com/Articles/heatmaps_in_r.html.

library(gplots) ## heatmap.2 is available in gplots

## 
## Attaching package: 'gplots'

## The following object is masked from 'package:stats':
## 
##     lowess

?heatmap.2

Application on Iris data

dataMatrix <- as.matrix(iris[,1:4])
heatmap.2(dataMatrix)

• Remove level trace

dataMatrix <- as.matrix(iris[,1:4])
heatmap.2(dataMatrix, trace = "none")

• Remove level trace

dataMatrix2 <- as.matrix(iris[1:5,1:4])
heatmap.2(dataMatrix2, trace = "none", 
          cellnote = dataMatrix2, notecol = "black")

• Put colSideColorBar
  • RowSideColors
  • legend

dataMatrix <- as.matrix(iris[,1:4])
species <- iris$Species
color0 <- species
levels(color0) <- palette()[1:length(levels(species))]
color <- as.character(color0)

heatmap.2(dataMatrix, trace = "none", RowSideColors = color)
legend("topright", legend=levels(species),fill=levels(color0))

• Remove automatical reordering of rows and columns
  • Colv = NA
  • Rowv = NA

dataMatrix <- as.matrix(iris[,1:4])
species <- iris$Species
color0 <- species
levels(color0) <- palette()[1:length(levels(species))]
color <- as.character(color0)

heatmap.2(dataMatrix, trace = "none", RowSideColors = color,
           Colv=NA, Rowv = NA )

## Warning in heatmap.2(dataMatrix, trace = "none", RowSideColors = color, :
## Discrepancy: Rowv is FALSE, while dendrogram is `both'. Omitting row
## dendogram.

## Warning in heatmap.2(dataMatrix, trace = "none", RowSideColors = color, :
## Discrepancy: Colv is FALSE, while dendrogram is `column'. Omitting column
## dendogram.

legend("topright", legend=levels(species),fill=levels(color0))

• Standardize each row to mean 0 and sd 1
  • scale = “row”

dataMatrix <- as.matrix(iris[,1:4])
species <- iris$Species
color0 <- species
levels(color0) <- palette()[1:length(levels(species))]
color <- as.character(color0)

heatmap.2(dataMatrix, trace = "none", RowSideColors = color,
           Colv=NA, Rowv = NA , scale = "row")

## Warning in heatmap.2(dataMatrix, trace = "none", RowSideColors = color, :
## Discrepancy: Rowv is FALSE, while dendrogram is `both'. Omitting row
## dendogram.

## Warning in heatmap.2(dataMatrix, trace = "none", RowSideColors = color, :
## Discrepancy: Colv is FALSE, while dendrogram is `column'. Omitting column
## dendogram.

legend("topright", legend=levels(species),fill=levels(color0))

• Try other colors
  • heat.colors (default)
  • redgreen
  • greenred
  • grey.colors
  • bluered
  • cm.colors
  • rainbow
  • terrain.colors
  • topo.colors

dataMatrix <- as.matrix(iris[,1:4])
species <- iris$Species
color0 <- species
levels(color0) <- palette()[1:length(levels(species))]
color <- as.character(color0)

heatmap.2(dataMatrix, trace = "none", 
          RowSideColors = color, col = bluered, 
           Colv=NA, Rowv = NA , scale = "row")

## Warning in heatmap.2(dataMatrix, trace = "none", RowSideColors = color, :
## Discrepancy: Rowv is FALSE, while dendrogram is `both'. Omitting row
## dendogram.

## Warning in heatmap.2(dataMatrix, trace = "none", RowSideColors = color, :
## Discrepancy: Colv is FALSE, while dendrogram is `column'. Omitting column
## dendogram.

legend("topright", legend=levels(species),fill=levels(color0))