PCR Array Heatmap - Scaling and Clustering

Question

[Biogear230]

Hello everyone,

I have lately been trying to perform heatmap analysis of PCR array data, and have been wondering what the standard approach would be. I first centered the data using the mean, scaled the array and set the heatmap to be scaled by row. I was wondering what would be the best method for scaling and clustering my data.

1. Is it preferred to center the PCR data using the median or the mean?

2. When performing the heatmap.2 function, should scale be set to 'none' as the data has been scaled once already?

3. Why do you obtain different results when using web based PCR array analysis?

4. Does the default clustering(euclidean) method for heatmap2 generate the best results or should I perform a different method of clustering?

Thank you very much!

Below is the code I used to generate a heatmap:

library(gplots)

datatable <- read.table("PCR_Array_Mean.txt")
data <- scale(as.matrix(datatable))


heatmap.2(data,scale='none',
                  trace="none", 
                  key = TRUE, 
                  keysize = 1,
                  key.ylab = NA,
                  key.xlab = NA, 
                  density.info = 'none',
                  margins = c(7,10),
                  lhei=c(1,10),
                  col=greenred(1000), 
                  symbreaks=TRUE)