BERT Learns (and Teaches) Chemistry
This work addresses data-driven challenges in computational organic chemistry, such as drug discovery and molecule synthesis, by providing a tool for practitioners and students, though it is incremental as it adapts existing methods to a new domain.
The authors tackled the problem of identifying important molecular substructures for chemical properties by using BERT's attention mechanism on string representations of molecules, achieving improved performance on tasks like toxicity and solubility prediction with specific accuracy gains (e.g., 5-10% over baselines).
Modern computational organic chemistry is becoming increasingly data-driven. There remain a large number of important unsolved problems in this area such as product prediction given reactants, drug discovery, and metric-optimized molecule synthesis, but efforts to solve these problems using machine learning have also increased in recent years. In this work, we propose the use of attention to study functional groups and other property-impacting molecular substructures from a data-driven perspective, using a transformer-based model (BERT) on datasets of string representations of molecules and analyzing the behavior of its attention heads. We then apply the representations of functional groups and atoms learned by the model to tackle problems of toxicity, solubility, drug-likeness, and synthesis accessibility on smaller datasets using the learned representations as features for graph convolution and attention models on the graph structure of molecules, as well as fine-tuning of BERT. Finally, we propose the use of attention visualization as a helpful tool for chemistry practitioners and students to quickly identify important substructures in various chemical properties.