Table 1.
The comparison of the properties of the molecules in the transfer learning dataset and the molecules generated by different methods.
| Docking Score | QED | SA Score | LogP | |
|---|---|---|---|---|
| SBMolGen | −7.416 ± 0.36 | 0.509 ± 0.01 | 3.233 ± 0.11 | 4.582 ± 0.88 |
| Transfer Learning Dataset | −7.493 ± 0.46 | 0.415 ± 0.03 | 2.769 ± 0.22 | 4.749 ± 1.90 |
| DUD-E | −7.366 ± 0.42 | 0.443 ± 0.03 | 2.747 ± 0.21 | 4.608 ± 1.35 |
| PETrans (without transfer learning) | −7.689 ± 0.60 | 0.421 ± 0.03 | 2.920 ± 0.48 | 5.185 ± 2.82 |
| PETrans (Ours) | −7.969 ± 0.58 | 0.452 ± 0.05 | 2.736 ± 0.57 | 4.567 ± 4.17 |