Table 8 Performance of StackDPP and state-of-the-art predictors on independent test sets subsampled from UNIPROT356 to mimic real-world proportion of DNA-BPs
From: StackDPP: a stacking ensemble based DNA-binding protein prediction model
Model | Accuracy | Sensitivity | Specificity | MCC | Precision | F1-score | AUROC | AUPR |
|---|---|---|---|---|---|---|---|---|
DPP-PseAAC | 0.85±0.009 | 0.82±0.137 | 0.85±0.000 | 0.42±0.079 | 0.27±0.035 | 0.41±0.057 | 0.84±0.068 | 0.38±0.093 |
LocalDPP | 0.94±0.007 | 0.31±0.117 | 0.98±0.000 | 0.35±0.120 | 0.46±0.113 | 0.37±0.121 | 0.65±0.059 | 0.38±0.100 |
Adelina (Group) | 0.88±0.007 | 0.80±0.112 | 0.88±0.000 | 0.45±0.067 | 0.31±0.031 | 0.45±0.049 | 0.84±0.056 | 0.66±0.145 |
Adelina (RFE) | 0.86±0.006 | 0.85±0.091 | 0.86±0.000 | 0.44±0.052 | 0.29±0.022 | 0.43±0.036 | 0.85±0.045 | 0.59±0.147 |
iDNAProt-ES | 0.08±0.000 | 1.00±0.000 | 0.02±0.000 | 0.03±0.000 | 0.06±0.000 | 0.12±0.000 | 0.51±0.000 | 0.15±0.048 |
StackDPP | 0.95±0.006 | 0.90±0.089 | 0.96±0.000 | 0.69±0.056 | 0.57±0.026 | 0.70±0.047 | 0.93±0.045 | 0.81±0.081 |