Skip to main content

Table 4 Selection performance of different methods in different scenarios (OR = 1.3)

From: Stability SCAD: a powerful approach to detect interactions in large-scale genomic study

 

Method

Scenario

A3

B3

C3

TPR

LASSO

0.681(0.032)

0.664(0.023)

0.777(0.021)

SCAD

0.715(0.028)

0.678(0.023)

0.7853(0.022)

SLASSO

0.651(0.019)

0.692(0.012)

0.874(0.012)

SSCAD

0.835(0.014)

0.808(0.005)

0.878(0.005)

MCC

LASSO

0.173(0.018)

0.298(0.018)

0.303(0.016)

SCAD

0.253(0.016)

0.350(0.017)

0.312(0.014)

SLASSO

0.676(0.012)

0.600(0.015)

0.768(0.013)

SSCAD

0.736(0.012)

0.737(0.014)

0.778(0.019)

AUC

LASSO

0.506(0.018)

0.524(0.003)

0.536(0.006)

SCAD

0.517(0.021)

0.514(0.006)

0.536(0.006)

SLASSO

0.518(0.002)

0.545(0.005)

0.535(0.005)

SSCAD

0.537(0.005)

0.541(0.002)

0.551(0.002)

FDR

LASSO

0.869(0.015)

0.838(0.011)

0.798(0.010)

SCAD

0.832(0.013)

0.819(0.008)

0.726(0.009)

SLASSO

0.188(0.007)

0.164(0.006)

0.182(0.010)

SSCAD

0.169(0.006)

0.163(0.007)

0.174(0.008)

  1. Abbreviations: TPR true positive rate, MCC Matthews correlation coefficient, AUC area under the ROC curve, FDR false discovery rate.
  2. Numbers in each cell represent mean (standard error) by 200 times simulation.