Table 1.

Comparison of predicted and experimental gene essentiality using different networks and different growth conditions.

Network	Experiment Condition	Threshold	Number of Gene Essentiality Predictions				Sensitivity	Specificity	Matthews Correlation Coefficient
			TP	FN	FP	TN
iNJ661	In vitro	<1.0	153	84	71	236	0.65	0.77	0.42
iNJ661m	In vitro	<1.0	153	85	71	237	0.64	0.77	0.42
GSMN-TB	In vitro	<1.0	156	85	58	294	0.65	0.84	0.49
GSMN-TBv	In vitro	<1.0	160	81	75	277	0.66	0.79	0.45
iNJ661v	In vitro	<1.0	140	98	80	228	0.59	0.74	0.33
iNJ661	In vitro	≤0.2	135	102	59	248	0.57	0.81	0.39
iNJ661m	In vitro	≤0.2	135	103	59	249	0.57	0.81	0.39
GSMN-TB	In vitro	≤0.2	152	89	47	305	0.63	0.87	0.52
GSMN-TBv	In vitro	≤0.2	156	85	65	287	0.65	0.82	0.47
iNJ661v	In vitro	≤0.2	123	115	63	245	0.52	0.80	0.33
iNJ661	In vivo	<1.0	16	20	97	242	0.44	0.71	0.10
iNJ661m	In vivo	<1.0	16	20	94	246	0.44	0.72	0.11
GSMN-TB	In vivo	<1.0	10	34	76	317	0.23	0.81	0.03
GSMN-TBv	In vivo	<1.0	16	28	93	300	0.36	0.76	0.09
iNJ661v	In vivo	<1.0	31	5	77	263	0.86	0.77	0.41
iNJ661	In vivo	≤0.2	11	25	76	263	0.31	0.78	0.06
iNJ661m	In vivo	≤0.2	11	25	76	264	0.31	0.78	0.06
GSMN-TB	In vivo	≤0.2	10	34	65	328	0.23	0.83	0.05
GSMN-TBv	In vivo	≤0.2	16	28	83	310	0.36	0.79	0.11
iNJ661v	In vivo	≤0.2	29	7	52	288	0.81	0.85	0.47

A true positive (TP) prediction refers to a gene correctly predicted to be essential, whereas a false negative (FN) prediction refers to a gene incorrectly predicted to be non-essential. A false positive (FP) prediction refers to a gene incorrectly predicted to be essential, whereas a true negative (TN) prediction refers to a gene correctly predicted to be non-essential. Sensitivity = TP/(TP + FN). Specificity = TN/(TN + FP). Matthews correlation coefficient = (TP × TN - FP × FN)/[(TP + FP)(TP + FN)(TN + FP)(TN + FN)]^1/2. GSMN-TBv indicates the GSMN-TB network with its in vivo biomass objective function.