Table 5.

A comparison of our AMP prediction method with state-of-the-art methods on AUC-ROC, AUC-PR, MCC, and κ by means of datasets C^train and C^test.

Method	ML algorithm	Number of features	AUC-ROC	AUC-PR	MCC	κ
iAMPpred#	SVM	66	0.98	0.99	0.91	—
iAMP-2L#	FKNN	46	0.95	—	0.9	—
AmPEP (DF)	RF	105	0.995	0.957	0.920	0.962
AmPEP (DF_PCC < 0.7)	RF	80	0.994	0.950	0.914	0.913
AmPEP (DF_PCC < 0.6)	RF	43	0.994	0.934	0.919	0.918
AmPEP (DF_PCC < 0.5)	RF	23	0.995	0.905	0.924	0.923

^#Results were taken from refs^5,6.