Review Article
A Review of Soft Computing Techniques for Gene Prediction
Table 2
Performance of soft computing techniques for protein-coding gene prediction.
| Program (soft computing technique used) | Exon level | Nucleotide level | Sensitivity (ESn) | Specificity (ESp) | Sensitivity (Sn) | Specificity (Sp) |
| GRAIL-I (NN1) | 53% | 90% | — | — | GeneParser (NN) | — | — | 83% | 83% | GRAIL-II (NN) | 89% | 91% | 91% | 90% | CODEX (NN) | 72% | 89% | — | — | GIN (NN) | 78% | 80% | 92% | 99% | MLFANN (NN) | — | — | 96.65% | 96.18% | SpyMGASLacGenePred (NN) | — | — | 100% | 76.90% | RescueNet (NN) | — | — | 89.39% | 89.04% | EG-MLP (NN) | — | — | 79% | 78% | Evolutionary algorithm (GA2) | — | — | 43% | 66% | MultiNNProm (NN + GA) | — | — | 98% | 97% | RBFN-combining (NN + GA) | 77% | 79% | 89% | 90% |
|
|
NN (Neural network), 2GA (Genetic algorithms).
|