Proportion statistics to detect differentially expressed genes: a comparison with log-ratio statistics

BMC Bioinformatics

Table 5 A summary of the missing values for each of the tests and the number of significant genes detected by other methods within those missing values.

Significant Genes from Sets of Missing Genes
Method	EBA (Affymetrix)	(LRT)	EBA (RNA-Seq)	edgeR	DESeq
EBA (Affymetrix)	14292	561	17	589	450	16	2550	1316
(LRT)	75	3979	0	235	197	0	3064	2208
EBA (RNA-Seq)	81	0	4726	235	197	0	3064	2208
edgeR	0	0	0	0	0	0	0	0
DESeq	0	0	0	0	0	0	0	0
	422	973	5	1166	997	8947	4171	2406
	2	0	0	0	0	0	915	0
	3	0	0	0	0	0	856	1832

The diagonal gives the number of genes with missing tests. The off-diagonals indicate those genes that are significant for one method amongst the missing calls for another method. The acronym LRT denotes the likelihood ratio test based on the Poisson distribution, as described in the Methods. The acronym EBA denotes the empirical Bayes analysis performed on both the Affymetrix and RNA-Seq data.

ISSN: 1471-2105