Figure 2.

Affymetrix and Illumina data from the Microarray Quality Control project can be directly integrated.A) Pairwise Pearson correlation heatmaps (left) demonstrate cross platform bias and the effects of three correction methods, mean-centering, distance-weighted discrimination (DWD) and an Empirical Bayes method (ComBat). R values range from low correlation (red) to high correlation (white) through shades of orange and yellow reflecting the overall similarity of expression profiles based upon biological and platform-specific variation. The shades of purple to pink indicate the samples (A = 100% UHRR, B = 100% HBRR, C = 75% UHRR + 25% HBRR, D = 25% UHRR + 75% HBRR). Samples are ordered by replicate and lab name rather than by platform. Green bars for Affymetrix samples and blue for Illumina samples. Boxplots of correlation coefficients within and between labs are shown in (Additional file 1. B) Cross-platform correction minimises technical variation whilst maintaining biological variation and differential expression. C) Venn diagrams demonstrate the overlap between the 1000 most differentially expressed genes between the MAQC UHRR and HBRR (A and B samples) using significance analysis of Microarrays (SAM) method with either Affymetrix (Green) or Illumina (Blue) alone, or Affymetrix and Illumina together (Purple).

Turnbull et al. BMC Medical Genomics 2012 5:35   doi:10.1186/1755-8794-5-35
Download authors' original image