Modelling p-value distributions to improve theme-driven survival analysis of cancer transcriptome datasets

BMC Bioinformatics

Table 2 Significant genesets and CSR comparison.

Geneset Name	Survival Estimate	Chang *et al*.(p-value)	p ₁	*FDR* ₁	p ₂	*FDR* ₂
GO "fatty acid metabolism"	Breast OS	-	0.0000005	0.006	0.00014	0.30
CSR	Breast OS	0.0410	0.0321000	0.280	0.14929	0.99
CSR	Breast RFS	0.0130	0.0144000	0.210	0.09614	0.99
GO "receptor mediated endocytosis"	Lung OS	-	0.0000020	0.003	0.00003	0.02
GO "brain development"	Lung OS	-	0.0000020	0.003	0.00003	0.02
GO "apical plasma membrane"	Lung OS	-	0.0001020	0.040	0.00078	0.29
KEGG "MAPK signaling pathway"	Lung OS	-	0.0001020	0.040	0.00080	0.29
CSR	Lung OS	0.0014	0.0026000	0.230	0.00325	0.53

For each survival estimate, a description of the significant biologically-related genesets (p₁ < 0.01; FDR₁ < 0.30; p₂ < 0.01; FDR₂ < 0.30) as well as a comparison of the p-values reported by Chang et al. and our log-rank (p₁) and empirical (p₂) p-values for the CSR gene expression signature are shown.

ISSN: 1471-2105