Table 6.

Comparison of the differential expression analysis on Hugene 1.0 ST arrays preprocessed with several popular algorithms

Method	Data	Diff. expr.	m2 m8	m3 m7	m4 m6	m1 m6	m4 m9	m1 m4	m6 m9	m1 m5a	m5a m9	m1 m5b	m5b m9	m1 m5c	m5c m9
MAS 5 1-Step Tukey-Biweight	PM	Student	0.72	0.74	0.61	0.83	0.83	0.67	0.54	0.61	0.65	0.73	0.72	0.75	0.74
		Win. t	0.82	0.82	0.69	0.89	0.86	0.70	0.57	0.71	0.69	0.80	0.77	0.80	0.77
		Reg. t	0.79	0.81	0.64	0.87	0.83	0.65	0.56	0.67	0.69	0.78	0.77	0.76	0.72
	PM	Student	0.80	0.80	0.65	0.88	0.83	0.71	0.60	0.83	0.79	0.84	0.80	0.81	0.78
	Norm	Win. t	0.85	0.85	0.69	0.90	0.86	0.71	0.59	0.85	0.81	0.86	0.83	0.84	0.79
		Reg. t	0.86	0.85	0.65	0.90	0.84	0.67	0.57	0.83	0.80	0.85	0.82	0.82	0.75
Plier	Raw	Student	0.86	0.74	0.75	0.90	0.85	0.78	0.52	0.74	0.71	0.83	0.77	0.85	0.80
		Win. t	0.90	0.83	0.79	0.92	0.89	0.80	0.59	0.77	0.74	0.85	0.81	0.87	0.82
		Reg. t	0.91	0.84	0.81	0.94	0.89	0.83	0.60	0.79	0.74	0.87	0.82	0.89	0.82
dChip Li-Wong MBEI	PM	Student	0.80	0.74	0.65	0.86	0.84	0.71	0.49	0.63	0.71	0.77	0.77	0.79	0.77
	Norm	Win. t	0.85	0.80	0.71	0.88	0.86	0.72	0.54	0.69	0.73	0.81	0.81	0.81	0.78
		Reg. t	0.82	0.79	0.65	0.88	0.84	0.70	0.53	0.64	0.75	0.80	0.82	0.80	0.73
RMA Median-polish	Norm	Student	0.83	0.83	0.70	0.86	0.83	0.72	0.60	0.80	0.78	0.81	0.78	0.81	0.77
		Win. t	0.89	0.88	0.75	0.88	0.86	0.73	0.60	0.83	0.80	0.84	0.81	0.83	0.79
		Reg. t	0.93	0.92	0.80	0.93	0.90	0.77	0.63	0.87	0.84	0.89	0.85	0.88	0.84
GCRMA Median-polish	Raw	Student	0.84	0.86	0.76	0.82	0.88	0.53	0.64	0.41	0.69	0.75	0.76	0.69	0.81
		Win. t	0.87	0.89	0.79	0.86	0.87	0.56	0.60	0.49	0.68	0.80	0.75	0.73	0.79
		Reg. t	0.90	0.91	0.84	0.88	0.90	0.65	0.62	0.44	0.70	0.84	0.77	0.74	0.82
	Norm	Student	0.85	0.85	0.78	0.89	0.86	0.71	0.68	0.83	0.82	0.84	0.82	0.80	0.80
		Win. t	0.89	0.88	0.80	0.89	0.86	0.70	0.62	0.84	0.80	0.84	0.81	0.82	0.79
		Reg. t	0.94	0.94	0.86	0.95	0.92	0.75	0.66	0.89	0.86	0.90	0.86	0.88	0.84

We analyzed the Hugene 1.0 ST Tissue Mixture Study (Affymetrix) for differential expression, using Student t-test, Regularized t-test, and Window t-test. Several previously published preprocessing methods were tested. Comparison between heart (mix 1) and brain (mix 9) is assumed to provide the reference list of p-values. For each combination of preprocessing/analysis steps (rows), Pearson's correlation coefficient has been computed on log10(p-values) between the reference analysis and several comparisons of mixtures (columns). Underlined characters highlights the top 5 correlation coefficients for each column. Raw/Norm labels refer to raw and normalized data, PM stands for PM-only methods.