Comparative Analysis on Alignment-Based and Pretrained Feature Representations for the Identification of DNA-Binding Proteins

Table 5

Performance comparison of the proposed ensemble FSEiDBP with other predictors validated on the independent dataset PDB186.

Methods	ACC (%)	MCC	SP (%)	SN (%)
DNA-Threader [63]	59.70	0.2790	95.70	23.70
DNAbinder [23]	60.80	0.2160	64.50	57.00
DNA-Prot [64]	61.80	0.2400	53.80	69.90
iDNA-Prot [65]	67.20	0.3440	66.70	67.70
DNABIND [66]	67.70	0.3550	68.80	66.70
Kmer1+ACC [67]	71.00	0.4310	59.10	82.80
iDNAPro-PseAAC [14]	71.50	0.4420	60.2	82.8
Wang’s method [68]	76.30	0.5570	60.20	92.50
DBPPred [24]	76.90	0.5380	74.20	79.60
DPP-PseAAC [20]	77.40	0.5500	70.90	83.00
Local-DPP [69]	79.00	0.6250	65.60	92.50
iDBP-DEP [41]	80.10	0.6250	66.70	93.60
iDNAProt-ES [70]	80.64	0.6130	80.00	81.30
FSEiDBP	83.33	0.6733	76.34	90.32