Show R heatmap outliers in a different colour

Question

Looking to mark outliers in R matrix in a different color. Say I have the data as

1  2  4  2  5
5  4  3  2  3
1 500 5  4  2

Now I want to

Answer 1

If you want to clearly see the 500, you should specify no scaling. For example,

m <- matrix(c(1, 5, 1, 2, 4, 500, 4, 3, 5, 2, 2, 4, 5, 3, 2), 
  ncol=5)

heatmap((m<500)+0, scale="none", Rowv=NA, Colv=NA)

Answer 2

Here's a decent workaround to achieve this. Using heatmap.2() from gplots, you can specify any colour for NA values in your heatmap. So, if you use a simple function to replace outliers with NAs in the source matrix, you can then represent them with any colour you like.

First, choose your outlier condition. For example's sake, let's just say that any value greater than 10 is an outlier.

> m
#      [,1] [,2] [,3] [,4] [,5]
# [1,]    1    2    4    2    5
# [2,]    5    4    3    2    3
# [3,]    1  500    5    4    2

m[m > 10] <- NA

Now plot the heatmap.

library(plots)
heatmap.2(m, trace = "none", na.color = "Green")

Outlier is now nice and obvious.