Table 2 Randomly selected SMILES generated by the different models
From: Molecular de-novo design through deep reinforcement learning
Model | Sampled SMILES |
|---|---|
Prior | CCOC(=O)C1=C(C)OC(N)=C(C#N)C1c1ccccc1C(F)(F)F |
COC(=O)CC(C)=NNc1ccc(N(C)C)cc1[N+](=O)[O-] | |
Cc1ccccc1CNS(=O)(=O)c1ccc2c(c1)C(=O)C(=O)N2 | |
Agent | CC(C)(C)NC(=O)c1ccc(OCc2ccccc2C(F)(F)F)nc1-c1ccccc1 |
CC(=O)NCC1OC(=O)N2c3ccc(-c4cccnc4)cc3OCC12 | |
OCCCNCc1cccc(-c2cccc(-c3nc4ccccc4[nH]3)c2OCCOc2ncc(Cl)cc2Br)c1 | |
Action level | CCN1CC(C)(C)OC(=O)c2cc(-c3ccc(Cl)cc3)ccc21 |
CCC(CC)C(=O)Nc1ccc2cnn(-c3ccc(C(C)=O)cc3)c2c1 | |
CCCCN1C(=O)c2ccccc2NC1c1ccc(OC)cc1 | |
REINFORCE | CC1CCCCC12NC(=O)N(CC(=O)Nc1ccccc1C(=O)O)C2=O |
CCCCCCCCCCCCCCCCCCCCCCCCCCCCNC(=O)OCCCCCC | |
CCCCCCCCCCCCCCCCCCCCCC1CCC(O)C1(CCC)CCCCCCCCCCCCCCC | |
REINFORCE + Prior | Nc1ccccc1C(=O)Oc1ccccc1 |
O=c1cccccc1Oc1ccccc1 | |
Nc1ccc(-c2ccccc2O)cc1 |