TY - JOUR
T1 - Masked graph modeling for molecule generation
AU - Mahmood, Omar
AU - Mansimov, Elman
AU - Bonneau, Richard
AU - Cho, Kyunghyun
N1 - Publisher Copyright:
© 2021, The Author(s).
PY - 2021/12/1
Y1 - 2021/12/1
N2 - De novo, in-silico design of molecules is a challenging problem with applications in drug discovery and material design. We introduce a masked graph model, which learns a distribution over graphs by capturing conditional distributions over unobserved nodes (atoms) and edges (bonds) given observed ones. We train and then sample from our model by iteratively masking and replacing different parts of initialized graphs. We evaluate our approach on the QM9 and ChEMBL datasets using the GuacaMol distribution-learning benchmark. We find that validity, KL-divergence and Fréchet ChemNet Distance scores are anti-correlated with novelty, and that we can trade off between these metrics more effectively than existing models. On distributional metrics, our model outperforms previously proposed graph-based approaches and is competitive with SMILES-based approaches. Finally, we show our model generates molecules with desired values of specified properties while maintaining physiochemical similarity to the training distribution.
AB - De novo, in-silico design of molecules is a challenging problem with applications in drug discovery and material design. We introduce a masked graph model, which learns a distribution over graphs by capturing conditional distributions over unobserved nodes (atoms) and edges (bonds) given observed ones. We train and then sample from our model by iteratively masking and replacing different parts of initialized graphs. We evaluate our approach on the QM9 and ChEMBL datasets using the GuacaMol distribution-learning benchmark. We find that validity, KL-divergence and Fréchet ChemNet Distance scores are anti-correlated with novelty, and that we can trade off between these metrics more effectively than existing models. On distributional metrics, our model outperforms previously proposed graph-based approaches and is competitive with SMILES-based approaches. Finally, we show our model generates molecules with desired values of specified properties while maintaining physiochemical similarity to the training distribution.
UR - http://www.scopus.com/inward/record.url?scp=85106916973&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85106916973&partnerID=8YFLogxK
U2 - 10.1038/s41467-021-23415-2
DO - 10.1038/s41467-021-23415-2
M3 - Article
C2 - 34039973
AN - SCOPUS:85106916973
SN - 2041-1723
VL - 12
JO - Nature communications
JF - Nature communications
IS - 1
M1 - 3156
ER -