Table 2.
Conditional molecular graph generation with proposed model
| Dataset | Target condition | G-mean | Unique count | MolWt | LogP |
|---|---|---|---|---|---|
| QM9 | Training set | – | 100,000 | 122.97 ± 7.61 | 0.14 ± 1.16 |
| Unconditional generation | 0.639 | 3243 | 123.01 ± 8.04 | − 0.06 ± 1.36 | |
| MolWt = 120 | 0.583 | 2316 | 121.85 ± 5.11 | 0.02 ± 1.36 | |
| MolWt = 125 | 0.543 | 1947 | 125.11 ± 4.56 | − 0.27 ± 1.22 | |
| MolWt = 130 | 0.482 | 1475 | 128.98 ± 4.27 | − 0.41 ± 1.33 | |
| LogP = − 0.4 | 0.576 | 2399 | 122.97 ± 8.26 | − 0.40 ± 0.73 | |
| LogP = 0.2 | 0.543 | 2099 | 122.53 ± 8.17 | 0.19 ± 0.75 | |
| LogP = 0.8 | 0.537 | 1989 | 122.17 ± 8.09 | 0.83 ± 0.72 | |
| ZINC | Training set | – | 100,000 | 357.94 ± 65.48 | 2.62 ± 1.36 |
| Unconditional generation | 0.888 | 7000 | 366.44 ± 51.63 | 2.49 ± 1.43 | |
| MolWt = 300 | 0.742 | 4090 | 313.12 ± 13.72 | 1.91 ± 1.50 | |
| MolWt = 350 | 0.796 | 5045 | 356.22 ± 12.66 | 2.24 ± 1.36 | |
| MolWt = 400 | 0.805 | 5212 | 400.95 ± 13.66 | 2.78 ± 1.30 | |
| LogP = 1.5 | 0.865 | 6470 | 352.33 ± 46.78 | 1.66 ± 0.94 | |
| LogP = 2.5 | 0.860 | 6356 | 366.64 ± 48.30 | 2.46 ± 0.92 | |
| LogP = 3.5 | 0.827 | 5658 | 381.96 ± 48.46 | 3.22 ± 0.85 |