Analysis of intra- and inter-laboratory reproducibility. (A) Box-and-whisker plots display, for each laboratory, the intra-laboratory squared correlation coefficients (r²) of all probe sets represented on the HG-U133 Plus 2.0 microarray for the HepG2 cell line sample. The signal used is DS. Each laboratory analysed six HepG2 samples using various amounts of starting total RNA: 1·0 μg, 1·5 μg, 3·0 μg, 5·0 μg (duplicate), or 8·0 μg, respectively. Thus, all possible different pairwise comparisons were performed (Count). Mean r² values (black arrow) and standard deviation (SD) values are given for each of the series of comparisons for each laboratory. Outliers are represented as red boxes. Note: more comparisons were performed in Centres 9 and 11 because multiple operators contributed microarray data (Appendix SII). (B) Repeatability of expression signal within laboratories. The CV of the expression signal values between centre replicates of the same sample type was calculated for all generally detected genes (left y-axis). The distributions of replicate CVs are presented in a series of eleven box-and-whisker plots: one for each of the two sample types HepG2 (left) or MCF-7 (right) at the eleven distinct laboratories. The median (line), interquartile range as well as the 10th and 90th percentile values are indicated in each plot. Only genes that were generally detected were included in the box plots and CV calculations. The number of generally detected genes was defined as being called present in at least one third of the samples, e.g., at least two out of the six replicates per centre. This number varied by sample and laboratory and is noted as the line plot with the y-axis on the right. (C) Box-and-whisker plots display the inter-laboratory squared correlation coefficients (r²) of all probe sets represented on the HG-U133 Plus 2.0 microarray for the HepG2 cell line sample. The signal used is DS. Each centre analysed six HepG2 samples using various amounts of starting total RNA: 1·0 μg, 1·5 μg, 3·0 μg, 5·0 μg (duplicate), or 8·0 μg, respectively. Here, microarray data from Centre 3 is compared with all other laboratories. Each inter-laboratory analysis with different pairwise comparisons is represented by a single box plot (Count). Mean r² values (black arrow) and standard deviation (SD) values are given for each series of comparisons. Outliers are represented as red boxes. Note: more comparisons were performed in Centres 9 and 11 because multiple operators contributed microarray data (Appendix SII). (D) Scatter plot analysis of inter-laboratory reproducibility. The graph shows 10 distinct scatter plot analyses, each displaying a comparison between Centre 3 and the other laboratories for the 5·0 μg HepG2 sample run at the stage of proficiency testing. The r² value calculation is based on DS intensity signals from all probe sets on the HG-U133 Plus 2.0 microarray.

Unsupervised principal component analysis (PCA). A total of 204 experiments are included in the three-dimensional PCA and each sphere represents the gene expression profile for a cell line or leukaemia sample. The signal used is DQN1. The first three principal components (PC) account for 41·0% of variation of the data (PC1 = 18·1%, PC2 = 14·9%, PC3 = 8·0%). The analysis is based on all probe sets represented on the HG-U133 Plus 2.0 microarray without any filtering process (n = 54 613). Outliers are marked with arrows. (A) The same sample types are represented by the same colour spheres. Distinct manufacturing batch numbers of the cell lines are given in Appendix SI. (B) Samples processed within the same centre are represented by the same colour spheres.