Hierarchical Graph-to-Graph Translation for Molecules
This work addresses the acceleration of drug discovery by providing a more effective tool for molecular optimization, though it appears incremental as it builds on prior graph-to-graph translation methods.
The paper tackled the problem of optimizing precursor molecules for better biochemical properties by extending graph-to-graph translation methods, achieving significant performance improvements over previous state-of-the-art baselines in multiple molecular optimization tasks.
The problem of accelerating drug discovery relies heavily on automatic tools to optimize precursor molecules to afford them with better biochemical properties. Our work in this paper substantially extends prior state-of-the-art on graph-to-graph translation methods for molecular optimization. In particular, we realize coherent multi-resolution representations by interweaving the encoding of substructure components with the atom-level encoding of the original molecular graph. Moreover, our graph decoder is fully autoregressive, and interleaves each step of adding a new substructure with the process of resolving its attachment to the emerging molecule. We evaluate our model on multiple molecular optimization tasks and show that our model significantly outperforms previous state-of-the-art baselines.