GAGE: generally applicable gene set enrichment for pathway analysis

BMC Bioinformatics

Table 5 The three comparison schemes of GAGE, 1-on-1, 1-on-grp and grp-on-grp

Gene Sets & Methods		Overlap	Top 10 p-values	Metastasis	Tumor	Sign. Sets
Experimental Sets	1-on-1	4	1.3E-28, 1.2–9	2, 3	5, 5	201 (254), 55 (47)
	1-on-grp	4	4.2E-35, 2.4E-13	3, 5	6, 7	242 (283), 120 (124)
	grp-on-grp	3	6.5E-8, 1.8E-4	3, 4	6, 8	52 (69), 17 (0)
Canonical Pathways	1-on-1	6	7.2E-5, 3.7E-03	9, 9	9, 9	18 (12), 8 (5)
	1-on-grp	5	6.7E-6, 7.5E-4	10, 9	10, 9	20 (16), 10 (8)
	grp-on-grp	0	1.1E-1, 6.1E-2	4, 5	6, 5	0 (0), 0 (0)

Top 10 significantly enriched experimental sets and canonical pathways in poor clinical outcomes vs good outcomes were inferred by GAGE using these three different comparison schemes from two published lung adenocarcinoma data sets [3]. Data columns are overlap between top 10 gene sets for the two studies, top 10 p-values, number of top 10 gene sets related to metastasis (bt) and tumor (t and bt), and numbers of significant gene sets with p-value ≤ 0.001 (or with FDR q-value ≤ 0.01).

ISSN: 1471-2105