Table 4 Performances of the nine models or methods for different applications.

Application	Evaluation	MFR	Logit	LDA	PCC	SRC	MI	PPC	CMI	CXP
10-fold cross-validation	AUC	0.819	0.818	0.818	0.699	0.692	0.664	0.695	0.484	0.686
10-fold cross-validation	B0 + B1	0.927	0.916	0.916	0.495	0.469	0.456	0.366	0.152	0.455
Test verification	AUC	0.823	0.822	0.822	0.696	0.690	0.658	0.691	0.484	0.682
Test verification	B0 + B1	0.873	0.856	0.856	0.440	0.518	0.428	0.270	0.172	0.477
GeneFriends verification	AUC	0.816	0.821	0.821	0.815	0.764	0.733	0.823	0.484	0.782
GeneFriends verification	B0 + B1	0.962	0.957	0.957	0.571	0.471	0.483	0.613	0.091	0.485
DIP verification	AUC	0.727	0.724	0.724	0.604	0.617	0.586	0.602	0.487	0.600
DIP verification	B0 + B1	0.727	0.713	0.713	0.544	0.507	0.519	0.438	0.142	0.463
Constructing a cancer gene network	NPP	15	12	14	10	10	11	12	8	10
Predicting gene function	L0 + L1	33.07	32.45	32.45	4.83	8.89	6.42	7.18	0.16	1.89

B0 + B1 indicates the average value of PPVs of B0- and B1-matched genes; NPP indicates the number of predicted metabolic pathways; L0 + L1 indicates the average number of L0- and L1-matched genes

Quick links

Search