Table 4.

Predictive performance of different prediction methods for the evaluation data sets.

	EV153_low
Localization	HYBRID		CELLO II		PSORTb v.2.0		PSLpred		P-CLASSIFIER
	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC
CP	91.7	0.67	91.7	0.70	63.5	-0.61	89.6	0.59	91.7	0.66
IM	65.4	0.73	46.2	0.64	46.2	-0.58	38.5	0.41	30.8	0.48
PP	45.5	0.25	81.8	0.49	00.0	-0.03	54.5	0.34	81.8	0.49
OM	44.4	0.58	33.3	0.34	22.2	-0.46	44.4	0.58	22.2	0.17
EC	27.3	0.43	45.5	0.50	09.1	-0.29	45.5	0.54	27.3	0.33
Overall	76.5	-	76.5	-	49.7	-	72.5	-	71.9	-
	EV90_high
Localization	HYBRID		CELLO II		PSORTb v.2.0		PSLpred		P-CLASSIFIER
	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC
CP	100.0	0.95	92.9	0.83	100.0	1.00	096.4	0.88	92.9	0.78
IM	96.2	0.97	73.1	0.75	100.0	1.00	92.3	0.92	80.8	0.84
PP	100.0	0.96	61.5	0.58	100.0	1.00	92.3	0.83	46.2	0.46
OM	94.7	0.97	73.7	0.67	94.7	0.97	68.4	0.79	73.7	0.69
EC	75.0	0.86	75.0	0.54	100.0	1.00	100.0	0.81	75.0	0.54
Overall	96.7	-	77.8	-	98.9	-	88.9	-	77.8	-
	EV243_all
Localization	HYBRID		CELLO II		PSORTb v.2.0		PSLpred		P-CLASSIFIER
	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC	Acc (%)	MCC
CP	93.5	0.80	91.9	0.77	71.8	0.73	91.1	0.72	91.9	0.73
IM	80.8	0.85	59.6	0.70	73.1	0.80	65.4	0.68	55.8	0.67
PP	75.0	0.56	70.8	0.51	54.2	0.66	75.0	0.57	62.5	0.45
OM	78.6	0.85	60.7	0.58	71.4	0.83	60.7	0.73	57.1	0.53
EC	40.0	0.57	53.3	0.50	33.3	0.57	60.0	0.62	40.0	0.39
Overall	84.0	-	77.0	-	67.9	-	78.6	-	74.1	-

§ The best performance of overall and individual localization sites is underlined. HYBRID is trained on the PS1444 data set.