Table 2.
Comparison of predictive quality measured on the MxD and CASP8 datasets when considering all disordered regions
| Dataset | Predictor | MCC |
Sw |
ACC |
SENS |
SPEC |
AUC |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Value | SD | Signif- | Value | SD | Signif- | Value | SD | Value | SD | Value | SD | Value | SD | Signif- | ||
| icance | icance | icance | ||||||||||||||
| MxD | MFDp | 0.44 | ±0.02 | 0.51 | ±0.01 | 0.75 | ±0.01 | 0.76 | ±0.01 | 0.75 | ±0.01 | 0.81 | ±0.01 | |||
| MD | 0.43 | ±0.02 | + | 0.48 | ±0.01 | + | 0.74 | ±0.01 | 0.68 | ±0.01 | 0.80 | ±0.01 | 0.81 | ±0.01 | = | |
| DISOPRED2 | 0.40 | ±0.02 | + | 0.44 | ±0.02 | + | 0.72 | ±0.01 | 0.66 | ±0.01 | 0.78 | ±0.01 | 0.78 | ±0.01 | + | |
| IUPredL | 0.39 | ±0.02 | + | 0.42 | ±0.01 | + | 0.71 | ±0.01 | 0.60 | ±0.01 | 0.82 | ±0.01 | 0.78 | ±0.01 | + | |
| IUPredS | 0.37 | ±0.01 | + | 0.38 | ±0.01 | + | 0.69 | ±0.01 | 0.53 | ±0.01 | 0.85 | ±0.01 | 0.78 | ±0.01 | + | |
| NORSnet | 0.34 | ±0.02 | + | 0.37 | ±0.02 | + | 0.68 | ±0.01 | 0.55 | ±0.02 | 0.81 | ±0.01 | 0.74 | ±0.01 | + | |
| DISOclust | 0.33 | ±0.01 | + | 0.40 | ±0.01 | + | 0.70 | ±0.01 | 0.78 | ±0.01 | 0.62 | ±0.01 | 0.77 | ±0.01 | + | |
| Ucon | 0.31 | ±0.01 | + | 0.34 | ±0.01 | + | 0.67 | ±0.01 | 0.57 | ±0.01 | 0.77 | ±0.01 | 0.74 | ±0.01 | + | |
| PROFbval | 0.19 | ±0.01 | + | 0.22 | ±0.01 | + | 0.61 | ±0.00 | 0.84 | ±0.01 | 0.38 | ±0.01 | 0.69 | ±0.01 | + | |
| CASP8 | MFDp | 0.61 | ±0.06 | 0.63 | ±0.06 | 0.82 | ±0.03 | 0.68 | ±0.06 | 0.95 | ±0.00 | 0.89 | ±0.02 | |||
| 379 | 0.59 | ±0.06 | + | 0.65 | ±0.05 | – | 0.82 | ±0.03 | 0.71 | ±0.05 | 0.94 | ±0.00 | 0.91 | ±0.02 | – | |
| DISOPRED2 | 0.59 | ±0.06 | + | 0.61 | ±0.06 | + | 0.80 | ±0.03 | 0.65 | ±0.06 | 0.95 | ±0.00 | 0.88 | ±0.02 | + | |
| 297 | 0.57 | ±0.05 | + | 0.66 | ±0.05 | – | 0.83 | ±0.02 | 0.74 | ±0.05 | 0.92 | ±0.00 | 0.90 | ±0.02 | – | |
| 97 | 0.56 | ±0.05 | + | 0.65 | ±0.05 | – | 0.82 | ±0.02 | 0.73 | ±0.05 | 0.92 | ±0.00 | 0.91 | ±0.02 | – | |
| 153 | 0.55 | ±0.05 | + | 0.67 | ±0.05 | – | 0.83 | ±0.02 | 0.76 | ±0.05 | 0.90 | ±0.00 | 0.91 | ±0.02 | – | |
| IUPredS | 0.54 | ±0.06 | + | 0.52 | ±0.06 | + | 0.76 | ±0.03 | 0.56 | ±0.06 | 0.96 | ±0.00 | 0.85 | ±0.03 | + | |
| IUPredL | 0.53 | ±0.09 | + | 0.45 | ±0.09 | + | 0.73 | ±0.05 | 0.48 | ±0.09 | 0.98 | ±0.00 | 0.82 | ±0.03 | + | |
| 69 | 0.51 | ±0.05 | + | 0.66 | ±0.04 | – | 0.83 | ±0.02 | 0.80 | ±0.04 | 0.86 | ±0.00 | 0.90 | ±0.02 | – | |
| NORSnet | 0.48 | ±0.12 | + | 0.37 | ±0.11 | + | 0.69 | ±0.06 | 0.39 | ±0.11 | 0.98 | ±0.00 | 0.79 | ±0.04 | + | |
| DISOclust | 0.42 | ±0.05 | + | 0.59 | ±0.04 | + | 0.80 | ±0.02 | 0.78 | ±0.04 | 0.82 | ±0.01 | 0.86 | ±0.02 | + | |
| MD | 0.42 | ±0.06 | + | 0.56 | ±0.06 | + | 0.78 | ±0.03 | 0.71 | ±0.06 | 0.85 | ±0.01 | 0.85 | ±0.03 | + | |
| Ucon | 0.29 | ±0.06 | + | 0.34 | ±0.07 | + | 0.67 | ±0.03 | 0.47 | ±0.07 | 0.87 | ±0.00 | 0.74 | ±0.04 | + | |
| PROFbval | 0.19 | ±0.03 | + | 0.31 | ±0.03 | + | 0.65 | ±0.01 | 0.86 | ±0.03 | 0.45 | ±0.01 | 0.78 | ±0.03 | + | |
We report the averages and corresponding SDs for bootstrapping with 1000 repetitions of 80% of chains. Underlined methods are used as inputs for MFDp. The methods are sorted by the MCC values and the highest values are shown in bold. Results of tests of significance of the differences between MFDp and the other methods are given in the ‘significance’ columns. The tests compare average values from 1000 bootstrapping repetitions. The + and – mean that MFDp is statistically significantly better/worse with P < 0.01, and = means that results are not significantly different.