Skip to main content

Table 4 Results of Simulation 1 (2): The performances of OGS compared with other approaches under gene-gene interactions across two pathways

From: Overlapping group screening for detection of gene-gene interactions: application to gene expression profiles with survival trait

 

Oracle

Uni. Sel.

Ordinary

Lasso

TS-GSIS

Lasso

OGS

Ridge

OGS

Lasso

censoring rate = 50%

 RMSE.M

0.418

0.424

0.382

0.361

0.436

0.349

 T.model

1

0

1

0.905

0.915

0.915

 Tint.model

1

0.035

1

0.905

0.915

0.915

 Sen.

1

0.746

1

0.992

0.994

0.994

 Spe.

1

0.999

0.963

0.965

0.666

0.966

 S.model

45

36.480

165.850

158.610

1139.54

155.235

 Deviance

− 133.967

−102.900

− 219.334

− 241.541

− 42.702

− 252.120

 c-index

0.944

0.899

0.980

0.984

0.842

0.986

censoring rate = 65%

 RMSE.M

0.414

0.422

0.398

0.390

0.436

0.382

 T.model

1

0

0.974

0.909

0.909

0.909

 Tint.model

1

0.035

1

0.909

0.909

0.909

 Sen.

1

0.729

0.999

0.992

0.994

0.992

 Spe.

1

0.999

0.968

0.970

0.558

0.971

 S.model

45

36.970

150.597

141.459

1494.17

140.481

 Deviance

− 127.447

−92.867

− 169.725

− 182.585

−39.018

− 191.849

 c-index

0.949

0.903

0.974

0.979

0.829

0.981