Microarray Normalization Revisited for Reproducible Breast Cancer Biomarkers

Figure 4

RMA single versus RMA global, both in MATLAB implementation (comparison “H”). (a) Distribution profiles of log₂-RMA normalized expression values. (b) Histogram of correlation distances of log₂-normalized expression values between pairs of pipelines. (c) For all 54675 probe sets of each sample, the 80% quantile (q_0.80) is computed and plotted over sample-ID (-axis).

(a)

(b)

(c)