Skip to main content

Table 1 Sensitivity and Specificity of the " large n small p" simulation

From: Network-based group variable selection for detecting expression quantitative trait loci (eQTL)

   Sensitivity Specificity
Model Gene NGVS GLasso Lasso NGVS GLasso Lasso
1 1 0.81
(0.17)
0.76
(0.21)
0.71
(0.24)
0.86
(0.13)
0.82
(0.16)
0.79
(0.18)
  2 0.78
(0.16)
0.77
(0.21)
0.69
(0.23)
0.84
(0.12)
0.83
(0.16)
0.77
(0.17)
2 1 0.97
(0.10)
0.97
(0.09)
0.85
(0.18)
0.98
(0.08)
0.98
(0.07)
0.89
(0.14)
  2 0.96
(0.11)
0.97
(0.10)
0.83
(0.18)
0.97
(0.08)
0.98
(0.08)
0.88
(0.14)
3 1 0.75
(0.21)
0.71
(0.20)
0.59
(0.22)
0.81
(0.16)
0.78
(0.15)
0.70
(0.16)
  2 0.76
(0.18)
0.74
(0.22)
0.58
(0.19)
0.82
(0.13)
0.81
(0.16)
0.69
(0.14)
4 1 0.82
(0.20)
0.94
(0.13)
0.67
(0.17)
0.87
(0.15)
0.96
(0.10)
0.76
(0.13)
  2 0.84
(0.18)
0.91
(0.16)
0.66
(0.16)
0.88
(0.14)
0.94
(0.12)
0.75
(0.12)
  1. NGVS: the network-based group variable selection method; GLasso: the group lasso without gene network and loci structure; Lasso: only considering additive effect of each locus. Sensitivity and specificity are calculated based on 50 simulations, with standard errors reported in parentheses.