# Controlling heatmap colors with ggplot2

November 22, 2012
By

(This article was first published on mintgene » R, and kindly contributed to R-bloggers)

One of the most popular posts on this blog is the very first one, solving the issue of mapping certain ranges of values to particular colors in heatmaps. Given the abundance of ggplot2 usage in R plotting, I thought I’d give it a try and do similar job within the context of graphics grammar.

## required packages (plot, melt data frame, and rolling function)
library(ggplot2)
library(reshape)
library(zoo)

## repeat random selection
set.seed(1)

## create 50x10 matrix of random values from [-1, +1]
random_matrix <- matrix(runif(500, min = -1, max = 1), nrow = 50)

## set color representation for specific values of the data distribution
quantile_range <- quantile(random_matrix, probs = seq(0, 1, 0.2))

## use http://colorbrewer2.org/ to find optimal divergent color palette (or set own)
color_palette <- colorRampPalette(c("#3794bf", "#FFFFFF", "#df8640"))(length(quantile_range) - 1)

## prepare label text (use two adjacent values for range text)
label_text <- rollapply(round(quantile_range, 2), width = 2, by = 1, FUN = function(i) paste(i, collapse = " : "))

## discretize matrix; this is the most important step, where for each value we find category of predefined ranges (modify probs argument of quantile to detail the colors)
mod_mat <- matrix(findInterval(random_matrix, quantile_range, all.inside = TRUE), nrow = nrow(random_matrix))

## remove background and axis from plot
theme_change <- theme(
plot.background = element_blank(),
panel.grid.minor = element_blank(),
panel.grid.major = element_blank(),
panel.background = element_blank(),
panel.border = element_blank(),
axis.line = element_blank(),
axis.ticks = element_blank(),
axis.text.x = element_blank(),
axis.text.y = element_blank(),
axis.title.x = element_blank(),
axis.title.y = element_blank()
)

## output the graphics
ggplot(melt(mod_mat), aes(x = X1, y = X2, fill = factor(value))) +
geom_tile(color = "black") +
scale_fill_manual(values = color_palette, name = "", labels = label_text) +
theme_change


Result of the quantile color representation:

We can predefine ranges, and create skewed colorsets:
Trick was to discretize the matrix of continuous values. Alternatively, you can use “breaks” argument in functions such as scale_fill_gradientn, but such method will assign continuous list of colors within set range.

Cheers.