Skip to main content

Table 2 Molecules having desired properties among valid and unique SMILES not in training data

From: Predicting chemical structure using reinforcement learning with a stack-augmented conditional variational autoencoder

 

Stack-CVAE

SSVAE

cRNN

CVAE

Stack-RNN

Binding Affinity on on-target proteins

652 (80.39%)

133 (80.12%)

145(57.54%)

368 (92.23%)

11 (91.67%)

Binding Affinity on off-target proteins

810 (99.88%)

139 (83.73%)

236(93.65%)

392 (98.25%)

12 (100%)

RAscore

566 (69.79%)

158 (95.18%)

248(98.41%)

198 (49.62%)

1 (8.33%)

Molecular Weight

532 (65.60%)

147 (88.55%)

198(78.57%)

134 (33.58%)

0 (0%)

LogP

623 (76.82%)

129 (77.71%)

180(71.43%)

238 (59.65%)

0 (0%)

TPSA

594 (73.24%)

138 (83.13%)

194(76.98%)

278 (69.67%)

0 (0%)