Junction Tree Variational Autoencoder for Molecular Graph Generation
Wengong Jin Regina Barzilay Tommi Jaakkola
Abstract
We seek to automate the design of molecules based on specific chemical properties. In computational terms, this task involves continuous embedding and generation of molecular graphs. Our primary contribution is the direct realization of molecular graphs, a task previously approached by generating linear SMILES strings instead of graphs. Our junction tree variational autoencoder generates molecular graphs in two phases, by first generating a tree-structured scaffold over chemical substructures, and then combining them into a molecule with a graph message passing network. This approach allows us to incrementally expand molecules while maintaining chemical validity at every step. We evaluate our model on multiple tasks ranging from molecular generation to optimization. Across these tasks, our model outperforms previous state-of-the-art baselines by a significant margin.
中文速览
药物发现的核心难题是怎样自动设计出具备目标化学性质的分子，过去的做法是先把分子转成SMILES线性字符串再生成，但这种方式无法很好地捕捉分子间的结构相似性，生成的分子也难以保证化学合法性。本文提出了一种叫做"结合树变分自编码器"（Junction Tree VAE, JT-VAE）的方法，把分子图的生成分为两个阶段：先生成一棵以化学子结构为节点的"结合树"作为骨架，再用图消息传递网络将这些子结构拼装成完整分子，从而在每一步都保证中间产物的化学合法性。在多项评测任务上，该模型从先验分布采样时能生成100%合法的分子，在分子性质优化任务上也比此前最优基线提升了约30%。这项工作第一次实现了直接对分子图进行端到端的生成建模，为计算机辅助药物设计提供了更可靠、更高效的分子生成框架。
原文 arXiv:1802.04364;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1802.04364v4