TPpred-LE: therapeutic peptide function prediction based on label embedding

Table 2 The performance of various methods for predicting eight therapeutic peptide functions on the independent dataset

Function	Method	AUC	MCC	F1
AAP	PEPred-Suite^a	0.577	0.02	0.03
	PPTPP^ab	0.604	0.037	0.033
	TPpred-ATMV^a	0.583	0.009	0.027
	TPpred-LE	0.745	0.278	0.285
ABP	PEPred-Suite^a	0.744	0.261	0.367
	PPTPP^ab	0.732	0.261	0.365
	TPpred-ATMV^a	0.731	0.256	0.36
	TPpred-LE	0.834	0.337	0.426
ACP	PEPred-Suite^a	0.56	0.03	0.155
	PPTPP^ab	0.625	0.049	0.162
	TPpred-ATMV^a	0.662	0.096	0.183
	TPpred-LE	0.773	0.328	0.371
AIP	PEPred-Suite^a	0.363	− 0.19	0.18
	PPTPP^ab	0.386	− 0.06	0.168
	TPpred-ATMV^a	0.369	− 0.25	0.196
	TPpred-LE	0.895	0.527	0.594
AVP	PEPred-Suite^a	0.382	− 0.129	0.147
	PPTPP^ab	0.404	− 0.11	0.169
	TPpred-ATMV^a	0.394	− 0.118	0.135
	TPpred-LE	0.835	0.457	0.529
CPP	PEPred-Suite^a	0.813	0.152	0.142
	PPTPP^ab	0.814	0.14	0.139
	TPpred-ATMV^a	0.815	0.152	0.139
	TPpred-LE	0.899	0.477	0.502
PBP	PEPred-Suite^a	0.907	0.153	0.069
	PPTPP^ab	0.829	0.119	0.07
	TPpred-ATMV^a	0.836	0.153	0.086
	TPpred-LE	0.934	0.443	0.430
QSP	PEPred-Suite^a	0.835	0.113	0.043
	PPTPP^ab	0.815	0.08	0.033
	TPpred-ATMV^a	0.772	0.054	0.027
	TPpred-LE	0.879	0.420	0.391

^aThe results are obtained by running their standalone programs
^bPPTPP contains three variant approaches, including PPTPP-cls, PPTPP-prb, and PPTPP-fus, among which only the best results for each metric are reported

ISSN: 1741-7007