Method |
MCC
| Ac(%) | Sn(%) | Sp(%) |
|---|
DDa,b | 0.541 | 82.4 | 78.8 | 83.3 |
NN_AACa,b | 0.716 | 91.0 | 79.3 | 93.8 |
SVM_AAC_DPCa,b | 0.816 | 93.9 | 90.9 | 94.7 |
SSEA-OMPc | 0.772 | 90.9 | 72.9 | 98.1 |
SSEA-OMPd | 0.906 | 96.2 | 91.5 | 98.1 |
- aDD, NN_AAC and SVM_AAC_DPC were developed in Suwa's group[5, 6, 15].
- bThe corresponding results are directly cited from [5, 6, 15].
- cBased on the stringent sequence-filtering method. Briefly, only the remaining sequences sharing a sequence identity less than 25%, a BLAST e-value greater than 0.01 and a PSI-BLAST e-value greater than 0.01 with the test protein were kept in the sequence library.
- dOnly the first sequence-filtering procedure was employed. Briefly, only the remaining sequences sharing a sequence identity less than 25% and a BLAST e-value greater than 0.01 with the test protein were kept in the sequence library. It should be emphasized that the performance of SSEA-OMP based on the first sequence-filtering procedure could be overestimated. We list the SSEA-OMP performance based on the first sequence-filtering procedure to allow a generally fair comparison between SSEA-OMP and the other three methods, since the performance of the other three methods were characterized by simple sequence identity-based filtering procedure [5, 6, 15].
