Table 1 Results for two-study simulation data. True integration-driven discovery rate (t IDR) and true integration-driven revision rate (t IRR) for threshold value of posterior probability of differential expression γ = 0.95, and the number of true discovered genes for posterior expected false discovery rate pe FDR = 5%, for the Bayesian standardized expression integration (SEI) and probability integration (PI) models. Results are shown for the three levels of simulated percent differentially expressed genes p _s and three levels of mean inter-study variability ${\tau }_g^2$ for the two-study simulation data.

Two-Study Simulation Data

SEI

PI

SEI

PI

SEI

PI

Low: mean ${\tau }_g^2$ = 0.1 (DE); 0.01 (non DE)

t IDR, γ = 0.95

24.6%

29.9%

12.9%

18.9%

3.8%

9.0%

t IRR, γ = 0.95

3.7%

0%

2.9%

1.4%

6.2%

1.5%

True Genes, pe FDR = 5%

84

92

183

196

479

513

Medium: mean ${\tau }_g^2$ = 0.3 (DE); 0.03 (non DE)

t IDR, γ = 0.95

11.2%

19.5%

10.8%

16.2%

2.6%

6.4%

t IRR, γ = 0.95

16.4%

7.5%

15.9%

4.5%

16.4%

3.1%

True Genes, pe FDR = 5%

80

97

185

211

493

547

High: mean ${\tau }_g^2$ = 0.7(DE); 0.07(non DE)

t IDR, γ = 0.95

11.0%

15.3%

5.4%

7.6%

2.9%

4.9%

t IRR, γ = 0.95

16.7%

7.7%

19.9%

6.6%

16.0%

4.8%

True Genes, pe FDR = 5%

93

105

212

235

572

617