Table 4.
Comparison among different folding rate prediction methods based on “IVANKOVDATA”
Methods | Method Type | Fold kinetic Classification Accuracy | Correlation | MAD |
---|---|---|---|---|
Effective length method | sequence | NA | 0.70 | 0.96 |
LRCO method | sequence | NA | 0.61 | 0.81 |
FOLD-RATE | sequence | NA | 0.91 | 1.1 |
K-Fold | structure | 81% | 0.74 | 0.75 |
Multi-predictor SVM (two-state) | sequence | 80% | 0.81 | 0.79 |
Multi-predictor SVM (multi-state) | sequence | 80% | 0.80 | 0.68 |
Method 1: Effective length method [14]
Method 2: LRCO method [19]
Method 3: FOLD-RATE [16]
Method 4: K-Fold [7]
Method 5: Our multi-predictor SVM (two-state)
Method 6: Our multi-predictor SVM (multi-state)
Method-Type means if the method is using experimental structural data (structure) or using only sequence data (sequence). Correlation here means the correlation value between predicted rates and experimental rates. MAD is mean absolute difference between predicted rates and experimental rates.