Dynamic probabilistic threshold networks to infer signaling pathways from time-course perturbation data

BMC Bioinformatics

Table 2 Performance comparison ERBB signaling network

	Sensitivity	Specificity	Accuracy	Precision	AUC _ROC	AUC _PR
D-PBTN	0.44	0.83	0.75	0.41	0.72	0.64
PBTN	0.20	0.77	0.66	0.20	0.52	0.31
DEPN	0.26	0.86	0.73	0.33	not avl.	not avl.
D-PBTN (+Prior)	0.59	0.90	0.84	0.62	0.78	0.70
DEPN (+Prior)	0.59	0.87	0.81	0.55	not avl.	not avl.
Prior network	0.48	0.87	0.79	0.5	-	-

Achieved sensitivity, specificity, accuracy and precision for the inference of ERBB signaling [34]. Shown are results of D-PBTN, PBTN and DEPN, with and without prior information (+Prior). The last row shows results of the prior network alone. Network inference was assessed using STRING as gold standard. Results for DEPN were taken from Froehlich et al. [34], authors do not report AUC_ROC or AUC_PR values.

ISSN: 1471-2105