Conditional Coloring Of Heatmaps

Question

[Assa Yeroslaviz]

Hi all,

I am trying to create a heatmap for my data. Here is a short part of the matrix

                1       2       3          A          B          C         D
  1        8.1733  6.4126  6.2126 16.59719550 16.285273 16.1536943 16.226814
  2            NA      NA      NA 11.05089616 11.702181 10.6578633 10.607349
  3            NA      NA      NA  6.60984359  4.234195  3.6587827  6.665620
  4       13.2216 13.2145 13.2326 14.08834851 14.214714 14.6321250 12.590646
  5       14.3365 14.2590 14.3201 15.69146410 14.873010 13.6182272 15.692651
  6            NA      NA      NA  0.00000000  0.000000  0.0000000  1.310340
  7            NA      NA      NA  0.07038933  0.000000  0.2016339  0.000000

I am using the heatmap.2 command from the gplots package in R. What I would like to have is the color grey for the NA, the color white for 0 and the rst should follow a certain color palette I am creating.

This is what I am doing at the moment

pairs.breaks <- c(0, seq(1, 18.23145, length.out=30))
color.palette= rev(colorRampPalette(c("yellow","orange","red","black"))(length(pairs.breaks) - 2))

heatmap.2( matrix,
Rowv=FALSE,na.rm=F, na.color="grey",  Colv=FALSE, 
trace="none",# tracecol="gray50", margin=c(4,5),
dendrogram = "none", 
col=c("white",color.palette),
colsep=c(3), sepcolor="white",
lhei=c(1,12), lwid=c(2, 8),
breaks=pairs.breaks)

The problem is, the in the last line for eaxmple the values 0.2016339 is also white. I would like white to be explicitly for 0 values. Anything else must have a color in my color-range.

Is it possible?

I would appreciate any help.

Assa

[Irsan]

use ggplot2 for nice R-graphics (reference guide here). You can make heatmaps and map the values of a column in your dataframe to a continuous colour gradient, you can give NA-values a specific colour and in this post you can get advice on how to deal with outliers in your data range that mess up the scale of the heatmap coloring

Answer 1

[Biesterfeld]

Your breaks look like

0.000000  1.000000  1.594188  2.188376

which means, that every value in [0,1) will be assigned by color white. Hence, you have to adjust your breaks:

matrix.min  <- min(matrix[ matrix!=0 ], na.rm=TRUE)
matrix.max <- max( matrix, na.rm = TRUE )
pairs.breaks <- c(0, seq( matrix.min, matrix.max, length.out=30) )