Search papers, labs, and topics across Lattice.
RetroGEF is a novel flow-based generative model designed for single-step retrosynthesis, addressing the challenge of molecular graph transformations that alter connectivity and size. By learning to add atoms and modify bonds directly from product-reactant pairs without a predefined edit order, it effectively models dynamic graph evolution. Experiments on benchmark datasets confirm RetroGEF's state-of-the-art performance in predicting viable synthetic routes.
A single generative flow model can dynamically transform molecular graphs, handling variable size and connectivity changes for retrosynthesis without explicit edit ordering.
Retrosynthesis enables the discovery of viable synthetic routes to target molecules. It plays a central role in modern drug discovery and materials design. Retrosynthesis involves molecular graph transformations that can change both connectivity and graph size. These transformations may introduce reactant components absent from the target while revising the product-derived structure. To model these transformations, we propose RetroGEF, a flow-based generative model for single-step retrosynthesis. Starting from the target molecule, it constructs possible reactants by adding atoms and changing bonds in the molecular graph. RetroGEF models molecular transformations and changes in graph size within the same generative process, rather than relying on a fixed-size graph canvas. It learns this process directly from product--reactant pairs without requiring a prescribed edit order. Experiments on representative retrosynthesis benchmarks demonstrate that RetroGEF achieves state-of-the-art performance.
