Down-weighting overlapping genes improves gene set analysis

BMC Bioinformatics

Table 8 A specificity analysis of the target pathways on 16 cancer data sets

Pathway type	Statistic	GSEA	GSA	PADOG
Target	p med	0.2603	0.087	0.043
Target	rank med	39.56	10.15	6.42
Apoptosis	p med	0.3329	0.203	0.1985
Apoptosis	rank med	37.56	24.24	28.76
Cell cycle	p med	0.3133	0.325	0.227
Cell cycle	rank med	26.29	36.35	28.61
Pathways in cancer	p med	0.351	0.114	0.0465
Pathways in cancer	rank med	47.54	13.21	8.41
RNA polymerase	p med	0.5	0.681	0.6485
RNA polymerase	rank med	57.78	71.51	63.33

The table shows a comparison between the pathways specifically designed by KEGG for each type of cancer (Target pathways) and other pathways that are commonly involved in many cancers. The table shows statistics computed from nominal p-values, and ranks of each type of pathway for the 16 cancer datasets shown in Table 1. PADOG gives the most significant p-values and best ranks to the target pathways. For each analysis method, the values for type of pathway with the smallest median p-values and ranks (strongest association with the phenotype) are shown in bold, while the second smallest values are italicized.

ISSN: 1471-2105