The chemical space of aromatic molecules is vast and diverse, which presents an opportunity for data-driven investigation. To facilitate this, our group has established The COMPAS Project: the first COMputational database of Polycyclic Aromatic Systems. We are actively expanding this chemical database in a methodical manner using high-throughput computational chemistry methods.
Polycyclic aromatic systems are well-known for being the workhorses of organic electronics. Our group is interested in using Machine Learning and Deep Learning models to enable the design and/or identification of new polycyclic aromatic molecules that can provide improved performance and stability.
The COMPAS Project is one component of this avenue of investigation; the second is the development of new chemical representations that are amenable to interpretation, and which allow exploration of the data for structure-property relationships.
Using these new representations, we train forward-models for property prediction, and inverse generative models for designing new molecules with targeted properties.
