data is always log2cpm transformed in correlation.R

[nrachman]

from looking at the source code, I am under the impression that data is always transformed to log2cpm when performing the correlation analyses in correlation.R even when log2cpm_transform = FALSE

I was using batchQC on microarray data and my data are already log transformed. Though subtle, the additional log transformation does affect the median pairwise correlations

using batchqc_corscatter - matches output from from batchQC(..., log2transfrom = FALSE)

modified batchqc_corscatter to not log2 CPM transform