De novo molecule design towards biased properties via a deep generative framework and iterative transfer learning

被引:1
|
作者
Sattari, Kianoosh [1 ]
Li, Dawei [1 ]
Kalita, Bhupalee [4 ]
Xie, Yunchao [1 ]
Lighvan, Fatemeh Barmaleki [5 ]
Isayev, Olexandr [4 ]
Lin, Jian [1 ,2 ,3 ]
机构
[1] Dept Mech & Aerosp Engn, Buffalo, NY 65409 USA
[2] Dept Elect Engn & Comp Sci, Cambridge, MA 02139 USA
[3] Univ Missouri, Dept Phys & Astron, Columbia, MO 65211 USA
[4] Carnegie Mellon Univ, Dept Chem, Pittsburgh, PA 15213 USA
[5] Southern Illinois Univ Edwardsville, Dept Biol Sci, Edwardsville, IL 62026 USA
来源
DIGITAL DISCOVERY | 2024年 / 3卷 / 02期
基金
美国国家科学基金会;
关键词
INVERSE DESIGN; PROJECT;
D O I
10.1039/d3dd00210a
中图分类号
O6 [化学];
学科分类号
0703 ;
摘要
De novo design of molecules with targeted properties represents a new frontier in molecule development. Despite enormous progress, two main challenges remain: (i) generating novel molecules conditioned on targeted, continuous property values; (ii) obtaining molecules with property values beyond the range in the training data. To tackle these challenges, we propose a reinforced regressional and conditional generative adversarial network (RRCGAN) to generate chemically valid molecules with targeted HOMO-LUMO energy gap (Delta EH-L) as a proof-of-concept study. As validated by density functional theory (DFT) calculation, 75% of the generated molecules have a relative error (RE) of <20% of the targeted Delta EH-L values. To bias the generation toward the Delta EH-L values beyond the range of the original training molecules, transfer learning was applied to iteratively retrain the RRCGAN model. After just two iterations, the mean Delta EH-L of the generated molecules increases to 8.7 eV from the mean value of 5.9 eV shown in the initial training dataset. Qualitative and quantitative analyses reveal that the model has successfully captured the underlying structure-property relationship, which agrees well with the established physical and chemical rules. These results present a trustworthy, purely data-driven methodology for the highly efficient generation of novel molecules with different targeted properties.
引用
收藏
页码:410 / 421
页数:12
相关论文
共 50 条
  • [21] Prospective de novo drug design with deep interactome learning
    Atz, Kenneth
    Cotos, Leandro
    Isert, Clemens
    Hakansson, Maria
    Focht, Dorota
    Hilleke, Mattis
    Nippa, David F.
    Iff, Michael
    Ledergerber, Jann
    Schiebroek, Carl C. G.
    Romeo, Valentina
    Hiss, Jan A.
    Merk, Daniel
    Schneider, Petra
    Kuhn, Bernd
    Grether, Uwe
    Schneider, Gisbert
    NATURE COMMUNICATIONS, 2024, 15 (01)
  • [22] Improving de novo protein binder design with deep learning
    Nathaniel R. Bennett
    Brian Coventry
    Inna Goreshnik
    Buwei Huang
    Aza Allen
    Dionne Vafeados
    Ying Po Peng
    Justas Dauparas
    Minkyung Baek
    Lance Stewart
    Frank DiMaio
    Steven De Munck
    Savvas N. Savvides
    David Baker
    Nature Communications, 14
  • [23] MolScore: a scoring, evaluation and benchmarking framework for generative models in de novo drug design
    Thomas, Morgan
    O'Boyle, Noel M.
    Bender, Andreas
    De Graaf, Chris
    JOURNAL OF CHEMINFORMATICS, 2024, 16 (01):
  • [24] RELATION: A Deep Generative Model for Structure-Based De Novo Drug Design
    Wang, Mingyang
    Hsieh, Chang-Yu
    Wang, Jike
    Wang, Dong
    Weng, Gaoqi
    Shen, Chao
    Yao, Xiaojun
    Bing, Zhitong
    Li, Honglin
    Cao, Dongsheng
    Hou, Tingjun
    JOURNAL OF MEDICINAL CHEMISTRY, 2022, : 9478 - 9492
  • [25] A survey of generative AI for de novo drug design: new frontiers in molecule and protein generation
    Tang, Xiangru
    Dai, Howard
    Knight, Elizabeth
    Wu, Fang
    Li, Yunyang
    Li, Tianxiao
    Gerstein, Mark
    BRIEFINGS IN BIOINFORMATICS, 2024, 25 (04)
  • [26] De novo drug design by iterative multiobjective deep reinforcement learning with graph-based molecular quality assessment
    Fang, Yi
    Pan, Xiaoyong
    Shen, Hong-Bin
    BIOINFORMATICS, 2023, 39 (04)
  • [27] De novo molecular design using deep reinforcement learning methods
    Chen, Hongming
    Olivercrona, Markus
    Blaschke, Thomas
    Engkvist, Ola
    Kogej, Thierry
    Tyrchan, Christian
    ABSTRACTS OF PAPERS OF THE AMERICAN CHEMICAL SOCIETY, 2018, 256
  • [28] Molecular de-novo design through deep reinforcement learning
    Olivecrona, Marcus
    Blaschke, Thomas
    Engkvist, Ola
    Chen, Hongming
    JOURNAL OF CHEMINFORMATICS, 2017, 9
  • [29] Molecular de-novo design through deep reinforcement learning
    Marcus Olivecrona
    Thomas Blaschke
    Ola Engkvist
    Hongming Chen
    Journal of Cheminformatics, 9
  • [30] Deep Reinforcement Learning for Multiparameter Optimization in de novo Drug Design
    Stahl, Niclas
    Falkman, Goran
    Karlsson, Alexander
    Mathiason, Gunnar
    Bostrom, Jonas
    JOURNAL OF CHEMICAL INFORMATION AND MODELING, 2019, 59 (07) : 3166 - 3176