Table 4. Performance of various scoring functions in predicting the quality of the model on CASP10 targets for a blind test.
RFMQA | dDFIRE | GOAP | DFIRE | RWplus | OPUS | |
CCTM | 0.984 | 0.967 | 0.971 | 0.956 | 0.782 | 0.954 |
0.985 | 0.964 | 0.966 | 0.964 | 0.960 | 0.966 | |
Average TMloss | 0.038 | 0.052 | 0.048 | 0.057 | 0.052 | 0.061 |
CCRank | 0.395 | 0.403 | 0.365 | 0.378 | 0.357 | 0.323 |
Note: The first, the second and the third rows respectively represents the correlation coefficient (CCTM), Spearman's correlation coefficient () and the average TM-score loss (TMloss) between TMmethod (TM-score of the model selected by a QA method) and TMbest (TM-score of the most native-like structure among decoys). The final row represents the average correlation coefficient between the predicted ranking and the actual ranking (CCRank) of 95 CASP10 targets. Bold fonts denote the best result.