Search papers, labs, and topics across Lattice.
ChemFlow, a novel hierarchical graph neural network, is introduced to predict physicochemical properties of molecular mixtures by simultaneously embedding intramolecular interactions and mixture composition. It integrates atomic, functional group, and molecular-level features, facilitating information flow across these levels. The model uses an atomic-level feature fusion module and bidirectional group-to-molecule attention mechanisms, outperforming state-of-the-art models in predicting concentration-dependent properties.
By dynamically adjusting representations based on concentration and composition, ChemFlow accurately models complex chemical mixtures, significantly outperforming existing methods.
Accurate prediction of the physicochemical properties of molecular mixtures using graph neural networks remains a significant challenge, as it requires simultaneous embedding of intramolecular interactions while accounting for mixture composition (i.e., concentrations and ratios). Existing approaches are ill-equipped to emulate realistic mixture environments, where densely coupled interactions propagate across hierarchical levels - from atoms and functional groups to entire molecules - and where cross-level information exchange is continuously modulated by composition. To bridge the gap between isolated molecules and realistic chemical environments, we present ChemFlow, a novel hierarchical framework that integrates atomic, functional group, and molecular-level features, facilitating information flow across these levels to predict the behavior of complex chemical mixtures. ChemFlow employs an atomic-level feature fusion module, Chem-embed, to generate context-aware atomic representations influenced by the mixture state and atomic characteristics. Next, bidirectional group-to-molecule and molecule-to-group attention mechanisms enable ChemFlow to capture functional group interactions both within and across molecules in the mixture. By dynamically adjusting representations based on concentration and composition, ChemFlow excels at predicting concentration-dependent properties and significantly outperforms state-of-the-art models in both concentration-sensitive and concentration-independent systems. Extensive experiments demonstrate ChemFlow's superior accuracy and efficiency in modeling complex chemical mixtures.
