StackDPP: a stacking ensemble based DNA-binding protein prediction model

BMC Bioinformatics

Table 3 Cross-validation (CV) performance of our reproduction of state-of-the-art models using the PDB1075 dataset

Models	ACC(%)		SN(%)		SP(%)		MCC
Models	P	R	P	R	P	R	P	R
DPP-PseAAC [17]	95.91	96.59	94.10	94.67	97.64	98.36	0.98	0.93
LocalDPP [11]	79.10	77.85	84.80	81.34	73.60	74.54	0.59	0.56
Adilina et al. [20](Group)	70.21	72.43	61	68.32	79.7	76.84	0.41	0.45
Adilina et al. [20](RFE)	71.04	71.12	62	69.44	79.9	77.81	0.43	0.44
iDNAProt-ES [14]	90.18	88.21	90.38	90.39	90	86.05	0.94	0.76

For the predictors of [20], 10-fold CV results are cited; for others we have shown the jackknife CV results. The column P represents results collected from the corresponding publications. The column R represents our reproduced results of the corresponding models

ISSN: 1471-2105