Table 1 Results by directly reconstructing molecules using random reconstruction algorithm
From: t-SMILES: a fragment-based molecular representation framework for de novo ligand design
Code Algorithm | Dataset | Model | Valid | Unique | Novelty | KLD | FCD | FBTs |
|---|---|---|---|---|---|---|---|---|
TSSA | ChEMBL | TSSA_J | 1.000 | 0.982 | 0.833 | 0.974 | 0.704 | 5029 |
TSSA_B | 1.000 | 0.992 | 0.682 | 0.981 | 0.720 | 610633 | ||
TSSA_M | 1.000 | 0.993 | 0.856 | 0.986 | 0.823 | 88938 | ||
TSSA_S | 1.000 | 1.000 | 0.882 | 0.969 | 0.816 | 515329 | ||
Zinc | TSSA_J | 1.000 | 0.971 | 0.835 | 0.985 | 0.827 | 61786 | |
TSSA_B | 1.000 | 0.971 | 0.755 | 0.975 | 0.740 | 1197 | ||
TSSA_M | 1.000 | 0.970 | 0.842 | 0.988 | 0.858 | 18989 | ||
TSSA_S | 1.000 | 0.976 | 0.876 | 0.972 | 0.840 | 485 | ||
QM9 | TSSA_J | 1.000 | 0.929 | 0.304 | 0.977 | 0.971 | 279 | |
TSSA_B | 1.000 | 0.945 | 0.056 | 0.998 | 0.983 | 32 | ||
TSSA_M | 1.000 | 0.911 | 0.141 | 0.997 | 0.979 | 238 | ||
TSSA_S | 1.000 | 0.898 | 0.156 | 0.996 | 0.975 | 69 | ||
TSDY | ChEMBL | TSDY_B | 1.000 | 0.996 | 0.210 | 0.997 | 0.915 | 4267 |
TSDY_M | 1.000 | 0.996 | 0.711 | 0.987 | 0.897 | 85409 | ||
TSDY_S | 1.000 | 0.996 | 0.459 | 0.995 | 0.913 | 39 | ||
Zinc | TSDY_B | 1.000 | 0.978 | 0.365 | 0.995 | 0.909 | 190 | |
TSDY_M | 1.000 | 0.978 | 0.688 | 0.996 | 0.916 | 5681 | ||
TSDY_S | 1.000 | 0.978 | 0.393 | 0.997 | 0.939 | 18 | ||
TSID | ChEMBL | TSID_B | 1.000 | 0.996 | 0.004 | 0.998 | 0.925 | 4267 |
Zinc | TSID_B | 1.000 | 0.978 | 0.010 | 0.999 | 0.945 | 190 |
