iPromoter-Seqvec: identifying promoters using bidirectional long short-term memory and sequence-embedded features

BMC Genomics

Table 1 Model performance on the independent test sets of iPromoter-Seqvec and other state-of-the-art methods

Dataset	Method	AUCROC	AUCPR	BA	SN	SP	PR	MCC	F1
HS-TApro	iPro-EP	0.89	0.87	0.81	0.84	0.78	0.79	0.62	0.81
	DeePromoter	-	-	0.67	0.94	0.39	0.61	0.40	0.74
	iPromoter-Seqvec (Ours)	0.99	0.99	0.94	0.90	0.99	0.99	0.89	0.94
HS-nonTApro	iPro-EP	0.73	0.74	0.65	0.73	0.56	0.63	0.30	0.67
	DeePromoter	-	-	0.51	0.90	0.12	0.51	0.04	0.65
	iPromoter-Seqvec (Ours)	0.86	0.86	0.75	0.62	0.89	0.85	0.53	0.72
MM-TApro	DeePromoter	-	-	0.59	0.84	0.34	0.56	0.21	0.67
MM-TApro	iPromoter-Seqvec (Ours)	0.99	0.99	0.93	0.88	0.98	0.97	0.86	0.92
MM-nonTApro	DeePromoter	-	-	0.64	0.87	0.40	0.59	0.31	0.71
MM-nonTApro	iPromoter-Seqvec (Ours)	0.91	0.91	0.83	0.74	0.91	0.90	0.67	0.81

ISSN: 1471-2164