Center for Quantitative Biology, Academy for Advanced Interdisciplinary Studies, Peking University, Beijing, 100871, PR China.
BNLMS, State Key Laboratory for Structural Chemistry of Unstable & Stable Species, College of Chemistry & Molecular Engineering, Peking University, Beijing, 100871, PR China.
Future Med Chem. 2019 Mar;11(6):567-597. doi: 10.4155/fmc-2018-0358. Epub 2019 Jan 30.
De novo drug design aims to generate novel chemical compounds with desirable chemical and pharmacological properties from scratch using computer-based methods. Recently, deep generative neural networks have become a very active research frontier in de novo drug discovery, both in theoretical and in experimental evidence, shedding light on a promising new direction of automatic molecular generation and optimization. In this review, we discussed recent development of deep learning models for molecular generation and summarized them as four different generative architectures with four different optimization strategies. We also discussed future directions of deep generative models for de novo drug design.
从头药物设计旨在使用基于计算机的方法从零开始生成具有理想化学和药理学性质的新型化合物。最近，深度生成神经网络在从头药物发现的理论和实验证据方面都成为了一个非常活跃的研究前沿，为自动分子生成和优化开辟了一个很有前途的新方向。在这篇综述中，我们讨论了分子生成的深度学习模型的最新进展，并将它们总结为具有四种不同优化策略的四种不同生成架构。我们还讨论了从头药物设计的深度生成模型的未来方向。
