Table 3 Performance evaluation of different metrics for the generated drug molecules using the four fragment-based deep generative models
From: DigFrag as a digital fragmentation method used for artificial intelligence-based drug design
Fragment | Valid | Unique | Novelty | IntDiv | Filters | Quantitative Estimation of Drug-likeness | Synthetic Accessibility |
|---|---|---|---|---|---|---|---|
BRICS | 1.000 | 0.977 | 0.998 | 0.816 | 0.710 | 0.661 | 2.703 |
RECAP | 0.999 | 0.932 | 1.000 | 0.821 | 0.751 | 0.644 | 2.709 |
MacFrag | 0.999 | 0.987 | 1.000 | 0.831 | 0.621 | 0.638 | 2.945 |
DigFrag | 1.000 | 0.968 | 1.000 | 0.823 | 0.828 | 0.673 | 2.681 |
