As described in more detail in the response to Reviewer #2, we performed several different analyses to show that the spike-in normalized data are virtually indistinguishable from the RPKM normalized data, although RPKM normalization was more reproducible across biological replicates.
The slope when comparing spike-in normalized data to RPKM normalized data was effectively 1.0, indicating that the values across the transcriptome were nearly identical when normalized with the two approaches.
The same general trends were apparent in the raw CT values for the focal genes as in the normalized data, suggesting Actin5C normalization did not skew the results.
A stringent quality control and inter-array normalization procedure resulted in a normalized data matrix of methylation values (β-values, 0 < β < 1) over 24,039 CpGs and 152 samples (Methods).
As one could anticipate, differences in the normalized data intensities were less obvious, in contrast to the variations in the raw signal intensities normally observed (and we also noted) in different experiments.
All statistically significant correlations in the normalized data were also significant in the unnormalized data, except for one pair: para and IH were correlated in the normalized data (R2 = 0.51), but not in the unnormalized data.
Array probes with less than 100 signal intensity in more than 50% of samples in the normalized data were excluded from the analysis where indicated.
