Reproducibility and reliability assays of the gene expression-measurements

Table 1 Reproducibility of the microarray data

	Uncorrected data				After quantile correction
Experiment ^a	Average number of genes with >50% inter-replicate variation	r _(P) ^b	Slope ^c	A.D ^d %	Average number of genes with >50% inter-replicate variation	r _(P) ^b	Slope ^c	A.D ^d %
E-GEOD-53990: *Arabidopsis* data with 6 treatments and 3 replicates	8488/22810 (37%)	0.975	0.804	91%	2000/22810 (8.8%)	0.980	0.982	26%
E-GEOD-27822: Barley data with 2 treatments and 3 replicates	6433/22840 (28%)	0.992	1.22	34%	519/22840 (2.3%)	0.994	0.992	8.2%
E-GEOD-39950: Yeast data with 2 treatments and 3 replicates	1645/10928 (15%)	0.997	1.257	26%	396/10928 (3.6%)	0.997	0.980	12%
E-GEOD-49918: *E. coli* data with 4 treatments and 3 replicates	1059/10208 (10%)	0.996	1.032	28%	299/10208 (2.9%)	0.997	0.981	10%
E-GEOD-24524: *E. coli* data with 4 treatments and 2 replicates	837/10208 (8.2%)	0.995	0.999	22%	300/1028 (2.9%)	0.996	0.990	15%

Different publicly available microarray data were analyzed using the Affymetrix Expression Console build 1.3.1.187 following the 3 Expression Arrays-RMA protocol. Both the corrected and uncorrected data were extracted. The Pearson correlations between the replicates were calculated on the log₂ transformed data. Following our method, the average of the ratios’ deviations was also calculated in order to evaluate its usefulness compared to the correlation coefficient.
^aThe experiments are available at http://www.ebi.ac.uk/arrayexpress/experiments/browse.html, ^bThe average of Pearson correlations between the replicates in each experiment, ^cThe average of slopes of inter-replicate regression lines in each experiment, ^dThe average of the ratios’ deviations.

ISSN: 2241-5793