Table 2.

Benchmark results of the thirteen prediction tools. Equations 1-4 are used to measure the predictive performance of these tools. (Abbreviations used: Sn = Sensitivity; Spc = Specificity; Acc = Accuracy; MCC = Matthews' Correlation Coefficient). ¹ Used with HMMER 2.3.2 with cut-off score set at -5 [26] and the updated model [20]; ² Version 3.0; ³ Authors updated system with UniProt 14.6 (Swiss-Prot Release 57.0) [3,62]; ⁴ Version 1.0.1.

	Experiment 1				Experiment 2				Experiment 3
Methods	Sn	Spc	Acc	MCC	Sn	Spc	Acc	MCC	Sn	Spc	Acc	MCC
Philius	0.704	0.952	0.828	0.677	0.742	0.968	0.855	0.729	0.728	0.961	0.844	0.708
Phobius	0.637	0.978	0.807	0.654	0.749	0.982	0.865	0.752	0.711	0.987	0.849	0.726
PrediSi	0.726	0.974	0.850	0.723	0.768	0.986	0.877	0.773	0.750	0.974	0.862	0.742
RPSP	0.730	0.989	0.859	0.744	0.805	0.996	0.901	0.816	0.794	1.000	0.897	0.811
SigCleave	0.541	0.878	0.709	0.445	0.613	0.823	0.718	0.446	0.618	0.860	0.739	0.493
SigHMM1	0.707	0.937	0.822	0.662	0.561	0.963	0.762	0.572	0.596	0.952	0.774	0.587
SignalP2 ANN	0.785	0.959	0.872	0.756	0.856	0.965	0.910	0.826	0.842	0.987	0.914	0.838
SignalP2 HMM	0.759	0.952	0.856	0.725	0.832	0.974	0.903	0.814	0.833	0.969	0.901	0.810
Signal-BLAST3	0.978	0.815	0.896	0.803	0.881	0.809	0.845	0.692	0.825	0.794	0.809	0.619
Signal-CF	0.648	0.900	0.774	0.566	0.768	0.905	0.836	0.679	0.750	0.890	0.820	0.647
Signal-3L	0.737	0.889	0.813	0.633	0.786	0.920	0.853	0.712	0.715	0.934	0.825	0.665
SOSUIsignal	0.189	0.926	0.557	0.170	0.232	0.925	0.578	0.217	0.232	0.921	0.577	0.212
SPOCTOCUS4	0.393	0.907	0.650	0.350	0.502	0.902	0.702	0.441	0.408	0.899	0.654	0.352