| SENS | SPEC | PPV | MCC |  |
---|
PDBinder (1)
| 0.295 | 0.983 | 0.413 | 0.313 |
HOLO
|
Q-siteFinder (2)
| 0.466 | 0.934 | 0.279 | 0.306 | Â |
(1) AND (2)
| 0.154 | 0.997 | 0.483 | 0.245 | Â |
(1) OR (2)
| 0.622 | 0.913 | 0.286 | 0.365 | Â |
PDBinder
| 0.251 | 0.984 | 0.372 | 0.271 |
APO
|
Q-siteFinder (2)
| 0.324 | 0.931 | 0.200 | 0.199 | Â |
(1) AND (2)
| 0.098 | 0.997 | 0.375 | 0.168 | Â |
(1) OR (2)
| 0.497 | 0.909 | 0.231 | 0.280 | Â |
- Integration of PDBinder with Q-SiteFinder. These results refer to the Apo/Holo Test Set. The first two rows detail the performance values achieved by PDBinder (1) and Q-SiteFinder (2) alone. The third row reports the results obtained if a residue is considered positive when it is predicted by both methods. The fourth row details the opposite situation in which a residue is considered positive when it is predicted by either method. The last row reports the results obtained with the two threshold systems. We performed a ten-fold cross validation to derive two different propensity thresholds, one for the residues predicted as positive by Q-SiteFinder and the other one for the negatives. The results are the average of the ten cross validation runs ± the standard error.