Conditionally coloring the fold change values to plot a heatmap using pheatmap

Question

[chetana]

Hi all,

I'm trying to plot a heat map with fold change (log2 values) values ranging from almost -100 to 100 across 6 time points. There are actually very few points in the extreme ranges of fold change. I was planning on plotting just the -10 to 10 range in the scale with differing color intensity where as the points above or below that will be color coded in a single color. I tried using breaks with (-100,-10,-5,0,5,10,100) but in the legend the color code for -10 to 10 range is not visible. Can you please suggest any other solution. I've been looking for it everywhere but haven't found a solution. I would really appreciate your help. Thanks in advance!

[Kevin Blighe]

Can you paste the exact code that you have been trying via the specification of breaks? - thanks!

[chetana]

Please find the code below. Patterns are the way set of few genes fall into a category based on their expression across the timepoints and I just used it for the row annotation.

library(pheatmap)
library(RColorBrewer)
geneExp <- read.table("ICL004-R_Pattern1-10.txt",
                  header=T, sep="\t")
head(geneExp)
names(geneExp)
geneExp_matrix <- as.matrix(geneExp[3:8])
rownames(geneExp_matrix)<-geneExp$Host.Factor.ID
Patterns<- c(1,2,3,4,5,6,7,8,9,10)
Pattern_color_options<- c(brewer.pal(10, "Set3"))
Pattern_colors <- unlist(lapply(geneExp$Pattern_number,function(x){
 for(P_num in 1:length(Patterns)){
   if(grepl(paste("Pattern_",Patterns[P_num],sep=""),x)) return(Pattern_color_options[P_num])
 }
}))
breaks_heatmap<- c(-100,-10,-5,0,5,10,100)
row_annotation<-data.frame(geneExp$Pattern_number)
rownames(row_annotation)<-geneExp$Host.Factor.ID
col.pal <- RColorBrewer::brewer.pal(6, "RdBu")
pheatmap::pheatmap(geneExp_matrix,
               legend=TRUE,
               cluster_row = FALSE,
               cluster_cols = FALSE, scale = "none",
               breaks = breaks_heatmap,
               annotation_col = NA,
               annotation_row = row_annotation, gaps_row = c(2921, 5685, 7790, 9764, 11407, 13028, 14122, 15053, 15972),
               annotation_legend = TRUE, 
               annotation_names_row = TRUE, annotation_names_col = FALSE,
               show_rownames = FALSE, show_colnames = TRUE,
               legend_breaks = c(-100,-10,-5,0,5,10,100),
               color = col.pal, gaps_col = c(1,2,3,4,5),
               fontsize = 6.5)

[Kevin Blighe]

Hello chetana my friend, you are specifying the breaks incorrectly.

To specify breaks for a heatmap, you have to do something like this:

breaks <- c(seq(-100, -10, by=5), seq(-9, -5, by=1), seq(-4, 4, by=0.5), seq(5, 9, by=1), seq(10, 100, by=5))

[Kevin Blighe]

Please take a look here: Keeping pheatmap colour range the same for different data sets