Submission Details
Molecule(s):
O=C(Oc1ccc(Cl)cc1)c1cccs1
Cc1ccc(CN2CCN(C(=O)CCl)CC2)o1
O=C(CCl)N(Cc1ccccc1Cl)c1ccccc1
CC(C)(C)NC(=O)C(N)c1cccc(F)c1
Cc1ccc(C(=O)NCCN2CCOCC2)cc1C
COc1ccccc1CCNC(=O)C(N)c1ccco1
CCC(C)N(C)C(=O)Cn1nnc2ccccc21
O=C(Cn1nnc2ccccc21)NCc1cccs1
Cc1ccc(NC(=S)NN2CCOCC2)cc1
O=C(Oc1cncc(Br)c1)c1ccccn1
Design Rationale:
These molecules were generated by our de novo design neural network. Briefly, we trained a LSTM-based model to generate molecules on the same chemical space as SARS-CoV-1 Mpro inhibitors. We then used the learned features to train a classifier which we used to predict the activity of the generated molecules.
Other Notes:
No fragments were used as inspiration but the submitted molecules do share structural features with known Mpro inhibitors.