Figure 3.

EASE themes are consistent despite the poor overlap of gene lists derived from the same experiment by various analytical methods. (a) Gene lists resulting from the same experiment can differ greatly due to selection criteria. Eight different methods were used to select genes up-regulated in PBMCs of HIV patients upon discontinuation of antiretroviral drug therapy. The various lists resulted from four different array-to-array normalizations, two different methods of intensity calculation and two methods of statistical selection of genes (see text). The grayed boxes on the diagonal show the total number of genes yielded by each method. The non-shaded boxes show the absolute number of genes shared by any two methods, with the percentage of genes in both lists from the combined gene lists is shown in parenthesis. MAS 4 = Affymetrix analysis software version 4.0. dChip = dChip software [6]. Rank remapping=unpublished method of DAH. Non-parametric = method of Sidirov, et al. [7]. t-Test = student T statistic. SAM = Statistical Analysis of Microarrays software [8]. (b) EASE identifies the same themes in spite of the notable variability in gene lists resulting from different methods of data normalization and gene selection criteria (see figure 3a). The six dominant categories over-represented in the genes up-regulated in PBMCs of HIV patients one month after discontinuing therapy all implicate a vigorous immune response. EASE scores (corrected for multiplicity with the bootstrap function using 10,000 random trials) are shown along with the rank (in parentheses) of each category in each of the eight lists. The category outside of the top five for each list is shaded gray. N.S. = p > 0.05

Hosack et al. Genome Biology 2003 4:P4   doi:10.1186/gb-2003-4-6-p4