You likely need to scale your data, or set the breaks to match the data distribution that you currently have.

If you simply set scale = 'row', for all intents and purposes, that may improve the heatmap from an aesthetics standpoint. This will result in heamap.2() Z-scaling your data per row. However, your current chosen values for breaks seems strange to me.

Another way to do this is to scale your input data via:

heatmap.2(
  t(scale(t(dataSub))),
  ...)

This has the exact same effect as setting scale = 'row'.

Take a quick look at my example here, ignoring the zFPKM part, though: Heatmap based with FPKM values

Kevin

I think you can try to remove col=bluered(100), because in default, they use yellow-red schema. P.S. You can create a vector of n contiguous colors using the functions rainbow(n), heat.colors(n), terrain.colors(n), topo.colors(n), and cm.colors(n).