Preferred analysis methods for Affymetrix GeneChips. II. An expanded, balanced, wholly-defined spike-in dataset

BMC Bioinformatics

Table 3 One-sided p-values of the Wilcoxon singed rank test for sample size study^a.

Routes^b	The number of arrays
	2	3	4	5	6	7
gcrma-reb.scaling.all.pmonly.medianpolish.vsn.all.CyberT	3.44e-47^c	5.08e-12	ns	ns	ns	ns
none.vsn.all.pmonly.medianpolish.constant.all.SAMR	8.53e-59	1.30e-42	1.43e-06	ns	ns	ns
gcrma-reb.scaling.all.pmonly.medianpolish.vsn.all.SAMR	6.22e-61	4.04e-49	1.59e-12	ns	ns	ns
gcrma-reb.constant.all.pmonly.medianpolish.vsn.all.CyberT	2.15e-52	9.56e-20	ns	ns	ns	ns
gcrma-reb.scaling.all.pmonly.medianpolish.vsn.all.LIMMA	6.22e-61	3.58e-56	2.04e-26	ns	ns	ns
gcrma-reb.constant.all.pmonly.medianpolish.vsn.all.LIMMA	6.22e-61	4.36e-59	7.32e-36	5.56e-13	ns	ns
none.vsn.all.pmonly.medianpolish.constant.all.LIMMA	4.36e-59	5.60e-39	1.03e-03	ns	ns	ns
none.vsn.all.pmonly.medianpolish.quantiles.all.FoldChange	2.77e-32	ns	ns	ns	ns	ns
rma.vsn.all.pmonly.medianpolish.scaling.all.SAMR	6.22e-61	3.64e-44	4.05e-05	ns	ns	ns

^ap-values for testing whether the relative AUC values corresponding to a specific number of arrays is less than 95% of the relative AUC value achieved by the same route in the Platinum Spike dataset.
^bThe routes used for sample size study. They are modified from nine of the top 10 routes (excluding Alchemy) assessed in the Platinum Spike dataset by using all arrays for normalization.
^cns: not significant, based on the multiple hypothesis correction using Holm's method.

ISSN: 1471-2105