Node-Structured Integrative Gaussian Graphical Model Guided by Pathway Information

Table 1

Performance comparisons of the nsiGGM with the JGGM and GGM using data simulated along with predefined module genes.

Methods	# of noise genes	Sensitivity (s.e.)	Specificity (s.e.)	Youden (s.e.)
nsiGGM	30	0.2217 (0.0253)	0.9433 (0.0036)	0.1650 (0.0257)
	40	0.2125 (0.0133)	0.9472 (0.0053)	0.1598 (0.0117)
	50	0.2034 (0.019)	0.9481 (0.0035)	0.1515 (0.0175)
JGGM	30	0.2433 (0.04)	0.8685 (0.0273)	0.1118 (0.0161)
	40	0.2815 (0.0418)	0.8321 (0.0309)	0.1136 (0.0146)
	50	0.1920 (0.0425)	0.8733 (0.0318)	0.0653 (0.0124)
GGM	30	0.2593 (0.0264)	0.8325 (0.0214)	0.0918 (0.0094)
	40	0.2752 (0.029)	0.8050 (0.0257)	0.0802 (0.0074)
	50	0.2177 (0.0303)	0.8431 (0.0268)	0.0608 (0.0085)