Improving the statistical detection of regulated genes from microarray data using intensity-based variance estimation

Table 2 Accuracy of each test statistic when compared to a "gold standard" t-statistic.

	M_avg	t	d_{90th percentile}	d_SAM	Z(I_avg)	Z(I_min)
Dataset 4 #1	0.69	0.09	0.80	0.80	0.79	0.83
Dataset 4 #2	0.66	0.08	0.78	0.78	0.79	0.82
Dataset 4 #3	0.54	0.04	0.79	0.70	0.77	0.84
Dataset 4 #4	0.68	0.02	0.80	0.79	0.80	0.83
Dataset 4 #5	0.64	0.06	0.80	0.74	0.78	0.84
Dataset 4 #6	0.67	0.01	0.80	0.80	0.79	0.83

Each column contains the R² value calculated between each experimental test statistic and the "gold standard" t-statistic for (left to right): the average logged ratio M_avg, the standard t-statistic, the 90^th percentile penalized t-statistic, the SAM penalized t-statistic, the Z-statistic using the I_avg pooling metric and the Z-statistic using the I_min pooling metric. Data are from six independent three-array subsets of Dataset 4. Although M_avg is not a statistical test, it is included in this table for comparison.

ISSN: 1471-2164