Table 5.

Comparison of SCRIP with other state-of-the-art predictors on benchmark testing datasets.

Predictor		Sensitivity	Specificity	Precision	Accuracy	MCC	F1	AUC
SecretomeP	Average stdev p-value	0.700 ±0.025 l.1e-08	0.726 ± 0.033 1.0e-06	0.719 ± 0.025 4.5e-08	0.713 ± 0.021 4.4e-09	0.426 ± 0.042 4.4e-09	0.709 ± 0.020 2.5e-09	0.709 ± 0.011 9.6e-20
SRTpred	Average stdev p-value	0.710 ± 0.021 1.0e-07	0.721 ± 0.026 2.8e-05	0.718 ± 0.017 2.0e-06	0.715 ± 0.012 2.1e-07	0.431 ± 0.024 2.2e-07	0.714 ± 0.012 9.1e-08	0.714 ± 0.018 1.0e-14
iMSP-U	Average stdev p-value	0.718 ± 0.031 1.2e-07	0.730 ± 0.026 6.6e-05	0.727 ± 0.026 8.3e-06	0.724 ± 0.026 1.2e-06	0.449 ± 0.052 1.2e-06	0.723 ± 0.027 4.3e-07	0.795 ± 0.009 4.4e-13
iMSP-H	Average stdev p-value	0.733 ± 0.027 3.9e-06	0.735 ± 0.025 1.4e-04	0.735 ± 0.019 3.5e-05	0.734 ± 0.018 8.9e-06	0.469 ± 0.036 9.2e-06	0.734 ± 0.019 5.1e-06	0.817 ± 0.012 1.4e-10
SCRIP	Average stdev p-value	0.754 ± 0.027 N/A	0.765 ± 0.036 N/A	0.763 ± 0.029 N/A	0.759 ± 0.024 N/A	0.519 ± 0.047 N/A	0.758 ± 0.023 N/A	0.844 ± 0.010 N/A

The threshold is set where the MCC achieve the maximum value.