Table 2.
Performances of different feature combination models
| Feature | Recall | Precision | ACC | F | AUC | AUPR |
|---|---|---|---|---|---|---|
| SP | 0.545 | 0.455 | 0.958 | 0.477 | 0.874 | 0.483 |
| SP + Markov | 0.528 | 0.479 | 0.961 | 0.482 | 0.887 | 0.492 |
| SP + Markov + DN | 0.530 | 0.484 | 0.961 | 0.486 | 0.889 | 0.498 |
| SP + Markov + DN + PWM | 0.505 | 0.507 | 0.963 | 0.487 | 0.889 | 0.500 |
| All | 0.532 | 0.478 | 0.961 | 0.484 | 0.884 | 0.494 |
Markov Markov motif profile, PWM position weight matrix profile, DN dinucleotide profile, SP sparse profile, PPT polypyrimidine tract, combination combining all features