Learning to Extend Molecular Scaffolds with Structural Motifs

Recent advancements in deep learning-based modeling of molecules promise to accelerate in silico drug discovery. A plethora of generative models is available, building molecules either atom-by-atom and bond-by-bond or fragment-by-fragment. However, many drug discovery projects require a fixed scaffold to be present in the generated molecule, and incorporating that constraint has only recently been explored. In this work, we propose a new graph-based model that naturally supports scaffolds as initial seed of the generative procedure, which is possible because our model is not conditioned on the generation history. At the same time, our generation procedure can flexibly choose between adding individual atoms and entire fragments. We show that training using a randomized generation order is necessary for good performance when extending scaffolds, and that the results are further improved by increasing the fragment vocabulary size. Our model pushes the state-of-the-art of graph-based molecule generation, while being an order of magnitude faster to train and sample from than existing approaches.