Citation: Xi Xue, Kai Chen, Hanyu Sun, Xiangying Liu, Xue Liu, Shize Li, Jingjie Yan, Yu Peng, Mohammad S. Mubarak, Ahmed Al-Harrasi, Hai-Yu Hu, Yafeng Deng, Xiandao Pan, Xiaojian Wang. Bidirectional Chemical Intelligent Net: A unified deep learning–based framework for predicting chemical reaction[J]. Chinese Chemical Letters, ;2025, 36(11): 110968. doi: 10.1016/j.cclet.2025.110968 shu

Bidirectional Chemical Intelligent Net: A unified deep learning–based framework for predicting chemical reaction

Xi Xue ^{a, b, 1} ,
Kai Chen ^{a, 1} ,
Hanyu Sun ^a ,
Xiangying Liu ^a ,
Xue Liu ^a ,
Shize Li ^a ,
Jingjie Yan ^a ,
Yu Peng ^a ,
Mohammad S. Mubarak ^c ,
Ahmed Al-Harrasi ^d ,
Hai-Yu Hu ^a ,
Yafeng Deng ^e ,
Xiandao Pan ^{a, b, *,} ,
Xiaojian Wang ^{a, b, *,}

a.
State Key Laboratory of Bioactive Substances and Functions of Natural Medicines, Institute of Materia Medica, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing 100050, China
b.
Beijing Key Laboratory of Active Substances Discovery and Druggability Evaluation, Department of Medicinal Chemistry, Institute of Materia Medica, Peking Union Medical College and Chinese Academy of Medical Sciences, Beijing 100050, China
c.
Department of Chemistry, The University of Jordan, Amman 11942, Jordan
d.
Natural and Medical Sciences Research Center, University of Nizwa, Nizwa 616, Oman
e.
CarbonSilicon AI Technology Co. Ltd., Beijing 100080, China

xdp@imm.ac.cn

wangxiaojian@imm.ac.cn

Received Date: 30 September 2024
Revised Date: 4 February 2025
Accepted Date: 17 February 2025
Available Online: 19 February 2025

Figures(6)

Chemical reactions, which transform one set of substances to another, drive research in chemistry and biology. Recently, computer-aided chemical reaction prediction has spurred rapidly growing interest, and various deep learning–based algorithms have been proposed. However, current efforts primarily focus on developing models that support specific applications, with less emphasis on building unified frameworks that predict chemical reactions. Here, we developed Bidirectional Chemical Intelligent Net (BiCINet), a prediction framework based on Bidirectional and Auto-Regressive Transformers (BARTs), for predicting chemical reactions in various tasks, including the bidirectional prediction of organic synthesis and enzyme-mediated chemical reactions. This versatile framework was trained using general chemical reactions and achieved top-1 forward and backward accuracies of 80.7% and 48.6%, respectively, for the public benchmark dataset USPTO_50K. By multitask transfer learning and integrating various task prompts into the model, BiCINet enables retrosynthetic planning and metabolic prediction for small molecules, as well as retrosynthetic analysis and enzyme-catalyzed product prediction for natural products. These results demonstrate the superiority of our multifunctional framework for comprehensively understanding chemical reactions.
- Machine learning,
- Deep learning,
- Chemical reaction prediction,
- Retrosynthesis,
- Metabolism,
- Biocatalytic reactions
1. [1]
  Z. Tu, T. Stuyver, C.W. Coley, Chem. Sci. 14 (2023) 226–244. doi: 10.1039/d2sc05089g
2. [2]
  M. Meuwly, Chem. Rev. 121 (2021) 10218–10239. doi: 10.1021/acs.chemrev.1c00033
3. [3]
  Testa, B., Pedretti, A., Vistoli, G. Drug Discov. Today 17 (2012) 549–560. doi: 10.1016/j.drudis.2012.01.017
4. [4]
  W. Zhong, Z. Yang, C.Y.C. Chen, Nat. Commun. 14 (2023) 3009. doi: 10.1038/s41467-023-38851-5
5. [5]
  S. Chen, Y. Jung, Nat. Mach. Intell. 4 (2022) 772–780. doi: 10.1038/s42256-022-00526-z
6. [6]
  D. Fooshee, A. Mood, E. Gutman, et al., Mol. Syst. Des. Eng. 3 (2018) 442–452. doi: 10.1039/c7me00107j
7. [7]
  B. Li, S. Su, C. Zhu, et al., J. Cheminform. 15 (2023) 72. doi: 10.54097/ajmss.v2i3.8719
8. [8]
  M.H.S. Segler, M.P. Waller, Chem. Eur. J. 23 (2017) 5966–5971. doi: 10.1002/chem.201605499
9. [9]
  Y. Wang, C. Pang, Y. Wang, et al., Nat. Commun. 14 (2023) 6155. doi: 10.1038/s41467-023-41698-5
10. [10]
  Z. Zhong, et al., WIREs Comput. Mol. Sci. 14 (2023) e1694.
11. [11]
  J. Dong, M. Zhao, Y. Liu, Y. Su, X. Zeng, Brief Bioinform. 23 (2022) bbab391.
12. [12]
  S. Zheng, J. Rao, Z. Zhang, J. Xu, Y. Yang, J. Chem. Inf. Model. 60 (2019) 47–55. doi: 10.1504/ijiids.2019.10023833
13. [13]
  C.W. Coley, R. Barzilay, T.S. Jaakkola, et al., ACS Cent. Sci. 3 (2017) 434–443. doi: 10.1021/acscentsci.7b00064
14. [14]
  C.W. Coley, L. Rogers, W.H. Green, K.F. Jensen, ACS Cent. Sci. 3 (2017) 1237–1245. doi: 10.1021/acscentsci.7b00355
15. [15]
  P. Karpov, G. Godin, I.V. Tetko, Chapter 78, artificial neural networks and machine learning – ICANN 2019: workshop and special sessions, Lecture Notes in Computer Science, Springer, 2019, pp. 817–830.
16. [16]
  K. Lin, Y. Xu, J. Pei, L. Lai, Chem. Sci. 11 (2020) 3355–3364. doi: 10.1039/c9sc03666k
17. [17]
  P. Schwaller, T. Laino, T. Gaudin, et al., ACS Cent. Sci. 5 (2019) 1572–1583. doi: 10.1021/acscentsci.9b00576
18. [18]
  M. Sacha, M. Błaz, P. Byrski, et al., J. Chem. Inf. Model. 61 (2021) 3273–3284. doi: 10.1021/acs.jcim.1c00537
19. [19]
  J. Lu, Y. Zhang, J. Chem. Inf. Model. 62 (2022) 1376–1387. doi: 10.1021/acs.jcim.1c01467
20. [20]
  B. Liu, B. Ramsundar, P. Kawthekar, et al., ACS Cent. Sci. 3 (2017) 1103–1113. doi: 10.1021/acscentsci.7b00303
21. [21]
  C. Shi, M. Xu, H. Guo, M. Zhang, J. Tang, arXiv: (2021), https://doi.org/10.48550/arXiv.2003.12725.
22. [22]
  Z. Tu, C.W. Coley, J. Chem. Inf. Model. 62 (2022) 3503–3513. doi: 10.1021/acs.jcim.2c00321
23. [23]
  I.V. Tetko, P. Karpov, R. Van Deursen, et al., Nat. Commun. 11 (2020) 5575. doi: 10.1038/s41467-020-19266-y
24. [24]
  Z. Zhong, J. Song, Z. Feng, et al., Chem. Sci. 13 (2022) 9023–9034. doi: 10.1039/d2sc02763a
25. [25]
  A.A. Lee, Q. Yang, V. Sresht, et al., Chem. Commun. 55 (2019) 12152–12155. doi: 10.1039/c9cc05122h
26. [26]
  W. Qian, X. Wang, Y. Kang, et al., J. Cheminform. 16 (2024) 38. doi: 10.1186/s13321-024-00827-y
27. [27]
  E. Heid, D. Probst, W.H. Green, et al., Chem. Sci. 14 (2023) 14229–14242. doi: 10.1039/d3sc02048g
28. [28]
  T.W. Thorpe, J.R. Marshall, N.J. Turner, J. Am. Chem. Soc. 146 (2024) 7876–7884. doi: 10.1021/jacs.3c09542
29. [29]
  D. Kreutter, P. Schwaller, J.L. Reymond, Chem. Sci. 12 (2021) 8648–8659. doi: 10.1039/d1sc02362d
30. [30]
  D. Probst, M. Manica, Y.G.N. Teukam, et al., Nat. Commun. 13 (2022) 964. doi: 10.1038/s41467-022-28536-w
31. [31]
  T.T.V. Tran, H. Tayara, K.T. Chong, Pharmaceutics 15 (2023) 1260. doi: 10.3390/pharmaceutics15041260
32. [32]
  Z. Chen, L. Zhang, P. Zhang, et al., J. Chem. Inf. Model. 64 (2024) 2528–2538. doi: 10.1021/acs.jcim.3c01396
33. [33]
  J.D. Tyzack, J. Kirchmair, Chem. Biol. Drug Des. 93 (2019) 377–386. doi: 10.1111/cbdd.13445
34. [34]
  Y. Djoumbou-Feunang, J. Fiamoncini, A. Gil-de-la-Fuente, et al., J. Cheminform. 11 (2019) 2. doi: 10.1186/s13321-018-0324-5
35. [35]
  J. Hafner, V. Hatzimanikatis, Bioinformatics 37 (2021) 3560–3568. doi: 10.1093/bioinformatics/btab368
36. [36]
  D. Wang, W. Liu, Z. Shen, et al., Front. Pharmacol. 10 (2019) 1586. doi: 10.1007/s10114-019-8312-x
37. [37]
  E.E. Litsa, P. Das, L.E. Kavraki, Chem. Sci. 11 (2020) 12777–12788. doi: 10.1039/d0sc02639e
38. [38]
  M. Lewis, Y. Liu, N. Goyal, et al., arXiv (2019), https://doi.org/10.48550/arXiv.1910.13461.
39. [39]
  C. Choudhury, V. Kumar, R. Kumar, Eur. J. Med. Chem. 249 (2023) 115153. doi: 10.1016/j.ejmech.2023.115153
40. [40]
  S.W. Seo, Y.Y. Song, J.Y. Yang, et al., GTA: Graph truncated attention for retrosynthesis, in: Proceedings of the AAAI Conference on Artificial Intelligence, 35, 2021, pp. 531–539.
1. [1]
  Yunzhe Zheng , Si Sun , Jiali Liu , Qingyu Zhao , Heng Zhang , Jing Zhang , Peng Zhou , Zhaokun Xiong , Chuan-Shu He , Bo Lai . Application of machine learning for material prediction and design in the environmental remediation. Chinese Chemical Letters, 2025, 36(9): 110722-. doi: 10.1016/j.cclet.2024.110722
2. [2]
  Lu Li , Jianing Shen , Qinkun Xiao , Chaozheng He , Jinzhou Zheng , Chaoqin Chu , Chen Chen . Stable crystal structure prediction using machine learning-based formation energy and empirical potential function. Chinese Chemical Letters, 2025, 36(11): 110421-. doi: 10.1016/j.cclet.2024.110421
3. [3]
  Zonglin Li , Shihua Zou , Zining Wang , Georgeta Postole , Liang Hu , Hongying Zhao . Machine learning in electrochemical oxidation process: A mini-review. Chinese Chemical Letters, 2025, 36(8): 110526-. doi: 10.1016/j.cclet.2024.110526
4. [4]
  Zixing Xu , Ruiying Chen , Chuanming Hao , Qionghong Xie , Chunhui Deng , Nianrong Sun . Peptidome data-driven comprehensive individualized monitoring of membranous nephropathy with machine learning. Chinese Chemical Letters, 2024, 35(5): 108975-. doi: 10.1016/j.cclet.2023.108975
5. [5]
  Qingyun Hu , Wei Wang , Junyuan Lu , He Zhu , Qi Liu , Yang Ren , Hong Wang , Jian Hui . High-throughput screening of high energy density LiMn_1-xFe_xPO₄ via active learning. Chinese Chemical Letters, 2025, 36(2): 110344-. doi: 10.1016/j.cclet.2024.110344
6. [6]
  Yuting Wu , Haifeng Lv , Xiaojun Wu . Design of two-dimensional porous covalent organic framework semiconductors for visible-light-driven overall water splitting: A theoretical perspective. Chinese Journal of Structural Chemistry, 2024, 43(11): 100375-100375. doi: 10.1016/j.cjsc.2024.100375
7. [7]
  Feihu Wu , Gengwen Chen , Kaitao Lai , Shiqing Zhang , Yingchao Liu , Ruijian Luo , Xiaocong Wang , Pinzhi Cao , Yi Ye , Jiarong Lian , Junle Qu , Zhigang Yang , Xiaojun Peng . Non-specific/specific SERS spectra concatenation for precise bacteria classifications with few samples using a residual neural network. Chinese Chemical Letters, 2025, 36(1): 109884-. doi: 10.1016/j.cclet.2024.109884
8. [8]
  Chengde Wang , Liping Huang , Shanshan Wang , Lihao Wu , Yi Wang , Jun Dong . A distinction of gliomas at cellular and tissue level by surface-enhanced Raman scattering spectroscopy. Chinese Chemical Letters, 2024, 35(5): 109383-. doi: 10.1016/j.cclet.2023.109383
9. [9]
  Xiaoxiao Wang , Bolun Wang , Fenfen Ji , Jie Yan , Jiacheng Fang , Doudou Zhang , Ji Xu , Jing Ji , Xinran Hao , Hemi Luan , Yanjun Hong , Shulan Qiu , Min Li , Zhu Yang , Wenlan Liu , Xiaodong Cai , Zongwei Cai . Discovery of plasma biomarkers for Parkinson’s disease diagnoses based on metabolomics and lipidomics. Chinese Chemical Letters, 2024, 35(11): 109653-. doi: 10.1016/j.cclet.2024.109653
10. [10]
  Ze Zhang , Lei Yang , Jin-Ru Liu , Hao Hu , Jian-Li Mi , Chao Su , Bei-Bei Xiao , Zhi-Min Ao . Improved oxygen electrocatalysis at FeN₄ and CoN₄ sites via construction of axial coordination. Chinese Chemical Letters, 2025, 36(2): 110013-. doi: 10.1016/j.cclet.2024.110013
11. [11]
  Xinyue Han , Yunhan Yang , Jiayin Lu , Yuxiang Lin , Dongxue Zhang , Ling Lin , Liang Qiao . Efficient serum lipids profiling by TiO₂-dopamin-assisted MALDI-TOF MS for breast cancer detection. Chinese Chemical Letters, 2025, 36(5): 110183-. doi: 10.1016/j.cclet.2024.110183
12. [12]
  Xiangyue Li , Dexin Zhu , Kunmin Pan , Xiaoye Zhou , Jiaming Zhu , Yingxue Wang , Yongpeng Ren , Hong-Hui Wu . Identifying key determinants of discharge capacity in ternary cathode materials of lithium-ion batteries. Chinese Chemical Letters, 2025, 36(5): 109870-. doi: 10.1016/j.cclet.2024.109870
13. [13]
  Hao Li , Hanzhi Lu , Linlin Hu , Xueli Zhang , Hua Shao , Fulun Li , Yanfei Shen . Dynamic surface-enhanced Raman spectroscopy-based metabolic profiling: A novel pathway to overcoming antifungal resistance. Chinese Chemical Letters, 2025, 36(7): 110342-. doi: 10.1016/j.cclet.2024.110342
14. [14]
  Yicheng Li , Qian Liu , Tianhao Li , Hao Bi , Zhurui Shen . Recent achievements in rare earth modified metal oxides for environmental and energy applications: A review. Chinese Chemical Letters, 2025, 36(9): 110698-. doi: 10.1016/j.cclet.2024.110698
15. [15]
  Xibao Li , Yiyang Wan , Fang Deng , Yingtang Zhou , Pinghua Chen , Fan Dong , Jizhou Jiang . Advances in Z-scheme and S-scheme heterojunctions for photocatalytic and photoelectrocatalytic H₂O₂ production. Chinese Chemical Letters, 2025, 36(10): 111418-. doi: 10.1016/j.cclet.2025.111418
16. [16]
  Yiwen Lin , Yijie Chen , Chunhui Deng , Nianrong Sun . Integration of resol/block-copolymer carbonization and machine learning: A convenient approach for precise monitoring of glycan-associated disorders. Chinese Chemical Letters, 2024, 35(12): 109813-. doi: 10.1016/j.cclet.2024.109813
17. [17]
  Zhaojing Huang , Hao Li , Jiayi Luo , Shunxing Li , Ming Zhao , Fengjiao Liu , Haijiao Xie . Deep learning-based simultaneous bioavailability assessment and speciation analysis of dissolved organic copper. Chinese Chemical Letters, 2025, 36(5): 110209-. doi: 10.1016/j.cclet.2024.110209
18. [18]
  Yao-Hua Gu , Yu Chen , Qing Li , Neng-Bin Xie , Xue Xing , Jun Xiong , Min Hu , Tian-Zhou Li , Ke-Yu Yuan , Yu Liu , Tang Tang , Fan He , Bi-Feng Yuan . Metabolome profiling by widely-targeted metabolomics and biomarker panel selection using machine-learning for patients in different stages of chronic kidney disease. Chinese Chemical Letters, 2024, 35(11): 109627-. doi: 10.1016/j.cclet.2024.109627
19. [19]
  Tong Zhou , Liyi Xie , Chuyu Liu , Xiyan Zheng , Bao Li . Between Sobriety and Intoxication: The Fascinating Journey of Sauce-Flavored Latte. University Chemistry, 2024, 39(9): 55-58. doi: 10.12461/PKU.DXHX202312048
20. [20]
  Chang Guo , Haipeng Yang , Hui Fang , Yingguo Zhao , Yating Li . 基于深度学习的物理化学课程DOK教学实践初探——以弯曲液面附加压力和蒸气压教学为例. University Chemistry, 2025, 40(6): 28-36. doi: 10.12461/PKU.DXHX202408049

Get Citation

PDF

XML

Metrics

PDF Downloads(0)
Abstract views(22)
HTML views(3)

通讯作者: 陈斌, bchen63@163.com

1.
沈阳化工大学材料科学与工程学院沈阳 110142