Applying machine learning to chemistry problems has a rich history in the context of property prediction (i.e., the development of QSAR/QSPR models), but has only recently been extended to other aspects of organic synthesis. We are developing tools for automated synthesis planning that leverage the collective chemical knowledge contained in large reaction databases like Reaxys and the USPTO. Our goal is to bridge the disconnect between using synthesis planning tools as idea generators and using synthesis planning tools for automated experimentation.
Areas of focus
- Learning retrosynthetic strategies from historical data
- Neural models for the prediction of organic reaction outcomes
- Molecular representation and property prediction
Recent publication: C.W. Coley, W.H. Green, and K.F. Jensen, “Machine learning in computer-aided synthesis planning,” Acc. Chem. Res. 51, 1281-1289 (2018).