Table 1.
Leveraging Both Sequence and Structural Information To Improve Protein–Ligand Affinity Predictionsa
| Vina-Docking | Custom-Docking | Neighbor-Joining-Ranking |
|---|---|---|
| DIG5 | DIG10 | DIG5 |
| DIG6 | DIG1 | DIG18 |
| DIG3 | DIG3 | DIG10 |
| DIG18 | DIG5 | DIG19 |
| DIG8 | DIG8 | DIG4 |
| DIG10 | DIG18 | DIG12 |
| DIG14 | DIG17 | DIG7 |
| DIG12 | DIG19 | DIG17 |
| DIG13 | DIG6 | DIG6 |
| DIG17 | DIG13 | DIG2 |
| DIG19 | DIG4 | DIG14 |
| DIG9 | DIG9 | DIG9 |
| DIG7 | DIG14 | DIG13 |
| DIG2 | DIG12 | DIG8 |
| DIG4 | DIG7 | DIG1 |
| DIG1 | DIG2 | DIG3 |
| AUC = 0.81 | 0.75 | 1.00 |
a
Shown are the predicted rankings of the 14 CSAR 2013 Phase 1 protein sequences using our three methods. The actual binders are shown in bold. The last row indicates the area under the curve (AUC) of our predictions.