Table 1.

System performance measures

Program	Positives		Negatives		Program performance evaluation
	TP	FP	TN	FN	Rc	Pr	Acc	MCC
SignaLPv4-Da	74	2	398	4	0.949	0.995	0.987	0.954
SignalPv3-HMM-S/TMHMMv2	76	9	373	20	0.792	0.976	0.939	0.805
Phobius	74	8	392	4	0.949	0.980	0.975	0.910
ProtComp-v9	59	6	391	22	0.728	0.985	0.941	0.781
WolfPsort	56	7	392	23	0.709	0.982	0.937	0.759
SignalPv3-NN-D	76	34	366	2	0.974	0.915	0.925	0.781
SignalPv3-HMM-S	76	32	368	2	0.974	0.920	0.929	0.790
SignalPv3-NN-D/SignalPv3-HMM-S	75	15	370	18	0.806	0.961	0.931	0.777
WolfPsort/SignalPv3-NN-D	56	32	367	23	0.709	0.920	0.885	0.602
WolfPsort/SignalPv3-HMM-S	56	23	377	22	0.718	0.943	0.906	0.657
WolfPsort/Phobius	55	12	387	24	0.696	0.970	0.925	0.713

Performance was measured based on the program’s ability to correctly discriminate extracellular proteins from non-extracellular proteins

Abbreviations: TP true positive prediction, FP false positive prediction, TN true negative prediction, FN false negative prediction, Rc recall, Pr precision, Acc accuracy, MCC Mathew’s Correlation Coefficient

^aMost accurate method to identify extracellular proteins