MolUNet++：自适应粒度显式子结构与互作感知分子表示学习

引用本文: 徐凡丁, 杨志伟, 武思睿, 苏武, 王力卓, 孟德宇, 龙建刚. MolUNet++：自适应粒度显式子结构与互作感知分子表示学习[J]. 物理化学学报, 2026, 42(5): 100209. doi: 10.1016/j.actphy.2025.100209 shu

Citation: Fanding Xu, Zhiwei Yang, Sirui Wu, Wu Su, Lizhuo Wang, Deyu Meng, Jiangang Long. MolUNet++: Adaptive-grained explicit substructure and interaction aware molecular representation learning[J]. Acta Physico-Chimica Sinica, 2026, 42(5): 100209. doi: 10.1016/j.actphy.2025.100209 shu

MolUNet++：自适应粒度显式子结构与互作感知分子表示学习

徐凡丁 ^1, ,
杨志伟 ^{2,
,} ,
武思睿 ^3, ,
苏武 ^1, ,
王力卓 ^1, ,
孟德宇 ^{4, 5,
,} ,
龙建刚 ^{1,
,}

1.
西安交通大学生命科学与技术学院, 陕西西安 710049
2.
西安交通大学物理学院, 陕西西安 710049
3.
中国移动通信集团陕西有限公司, 陕西西安 710077
4.
西安交通大学数学与统计学院, 陕西西安 710049
5.
河南大学数学与统计学院, 河南郑州 475004

通讯作者: yzws-123@xjtu.edu.cn (杨志伟); dymeng@mail.xjtu.edu.cn (孟德宇); Email: jglong@xjtu.edu.cn (龙建刚)

摘要: 分子表示学习是人工智能驱动药物研发中的关键任务。尽管图神经网络(GNN)在该领域已表现出优异性能并被广泛应用，但高效提取并显式解析官能团仍是一项挑战。为此，我们提出了MolUNet++模型，该模型通过分子边收缩池化(Molecular Edge Shrinkage Pooling，MESPool)实现分层子结构提取，利用嵌套式UNet框架进行多粒度特征融合，并结合子结构掩蔽解释器实现分子片段的定量分析。我们在分子性质预测、药物-药物相互作用(Drug-Drug Interaction，DDI)预测及药物-靶标相互作用(Drug-Target Interaction，DTI)预测等任务上对MolUNet++进行了评估。实验结果表明，MolUNet++不仅在预测性能上优于传统GNN模型，同时展现出显式、直观且符合化学逻辑的可解释性，为药物设计与优化领域的研究者提供了有价值的启示与工具。

关键词:

English

MolUNet++: Adaptive-grained explicit substructure and interaction aware molecular representation learning

Fanding Xu ^1, ,
Zhiwei Yang ^{2,
,} ,
Sirui Wu ^3, ,
Wu Su ^1, ,
Lizhuo Wang ^1, ,
Deyu Meng ^{4, 5,
,} ,
Jiangang Long ^{1,
,}

1.
School of Life Science and Technology, Xi'an Jiaotong University, Xi'an 710049, Shaanxi Province, China
2.
School of Physics, Xi'an Jiaotong University, Xi'an 710049, Shaanxi Province, China
3.
China Mobile Communications Group Shaanxi Co., Ltd, Xi'an 710077, Shaanxi Province, China
4.
Research Institute for Mathematics and Mathematical Technology, Xi'an Jiaotong University, Xi'an 710049, Shaanxi Province, China
5.
School of Mathematics and Statistics, Henan University, Zhengzhou 475004, Henan Province, China

Corresponding author: yzws-123@xjtu.edu.cn (Zhiwei Yang); dymeng@mail.xjtu.edu.cn (Deyu Meng); Email: jglong@xjtu.edu.cn (Jiangang Long)

Received Date: 22 August 2025
Accepted Date: 22 October 2025
Revised Date: 21 October 2025
Available Online: 15 May 2026

Abstract: Molecular representation learning is a critical task in AI-driven drug development. While graph neural networks (GNNs) have demonstrated strong performance and gained widespread adoption in this field, efficiently extracting and explicitly analyzing functional groups remains a challenge. To address this issue, we propose MolUNet++, a novel model that employs Molecular Edge Shrinkage Pooling (MESPool) for hierarchical substructure extraction, utilizes a Nested UNet framework for multi-granularity feature integration, and incorporates a substructure masking explainer for quantitative fragment analysis. We evaluated MolUNet++ on tasks including molecular property prediction, drug-drug interaction (DDI) prediction, and drug-target interaction (DTI) prediction. Experimental results demonstrate that MolUNet++ not only outperforms traditional GNN models in predictive performance but also exhibits explicit, intuitive, and chemically logical interpretability. This capability provides valuable insights and tools for researchers in drug design and optimization.

Key words:

1. [1]
  Z. Guo, K. Guo, B. Nan, Y. Tian, R.G. Iyer, Y. Ma, O. Wiest, X. Zhang, W. Wang, C. Zhang, et al., arXiv: 2207.04869, https://doi.org/10.48550/arXiv.2207.04869.
2. [2]
  T.-H. Nguyen-Vo, P. Teesdale-Spittle, J.E. Harvey, B.P. Nguyen, Memetic Comput. 16 (2024) 519, https://doi.org/10.1007/s12293-024-00414-6. doi: 10.1007/s12293-024-00414-6
3. [3]
  L. David, A. Thakkar, R. Mercado, O. Engkvist, J. Cheminform. 12 (2020) 56, https://doi.org/10.1186/s13321-020-00460-5. doi: 10.1186/s13321-020-00460-5
4. [4]
  Y. Harnik, A. Milo, Chem. Sci. 15 (2024) 5052, https://doi.org/10.1039/d4sc90043j. doi: 10.1039/d4sc90043j
5. [5]
  N. Wen, G. Liu, J. Zhang, R. Zhang, Y. Fu, X. Han, J. Cheminform. 14 (2022) 71, https://doi.org/10.1186/s13321-022-00650-3. doi: 10.1186/s13321-022-00650-3
6. [6]
  D. Baptista, J. Correia, B. Pereira, M. Rocha, J. Integr. Bioinform. 19 (2022) 20220006, https://doi.org/10.1515/jib-2022-0006. doi: 10.1515/jib-2022-0006
7. [7]
  R. Zhang, Y. Lin, Y. Wu, L. Deng, H. Zhang, M. Liao, Y. Peng, Brief. Bioinform. 25 (2024) bbae298, https://doi.org/10.1093/bib/bbae298. doi: 10.1093/bib/bbae298
8. [8]
  K.V. Chuang, L.M. Gunsalus, M.J. Keiser, J. Med. Chem. 63 (2020) 8705, https://doi.org/10.1021/acs.jmedchem.0c00385. doi: 10.1021/acs.jmedchem.0c00385
9. [9]
  D. Jiang, Z. Wu, C.-Y. Hsieh, G. Chen, B. Liao, Z. Wang, C. Shen, D. Cao, J. Wu, T. Hou, J. Cheminform. 13 (2021) 12, https://doi.org/10.1186/s13321-020-00479-8. doi: 10.1186/s13321-020-00479-8
10. [10]
  Y. Wang, Z. Li, A. Barati Farimani, Graph Neural Networks for Molecules, in Machine Learning in Molecular Sciences, C. Qu, H. Liu, Eds., Springer: Cham, Switzerland, 2023, pp. 21–66, https://doi.org/10.1007/978-3-031-37196-7_2.
11. [11]
  O. Wieder, S. Kohlbacher, M. Kuenemann, A. Garon, P. Ducrot, T. Seidel, T. Langer, Drug Discov. Today Technol. 37 (2020) 1, https://doi.org/10.1016/j.ddtec.2020.11.009. doi: 10.1016/j.ddtec.2020.11.009
12. [12]
  Z. Li, K. Meidani, P. Yadav, A. Barati Farimani, J. Chem. Phys. 156 (2022) 144103, https://doi.org/10.1063/5.0083060. doi: 10.1063/5.0083060
13. [13]
  Z. Zhang, L. Chen, F. Zhong, D. Wang, J. Jiang, S. Zhang, H. Jiang, M. Zheng, X. Li, Curr. Opin. Struct. Biol. 73 (2022) 102327, https://doi.org/10.1016/j.sbi.2021.102327. doi: 10.1016/j.sbi.2021.102327
14. [14]
  Z. Wu, J. Wang, H. Du, D. Jiang, Y. Kang, D. Li, P. Pan, Y. Deng, D. Cao, C.-Y. Hsieh, et al., Nat. Commun. 14 (2023) 2585, https://doi.org/10.1038/s41467-023-38192-3. doi: 10.1038/s41467-023-38192-3
15. [15]
  A.Z. Dudek, T. Arodz, J. Gálvez, Comb. Chem. High Throughput Screen. 9 (2006) 213, https://doi.org/10.2174/138620706776055539. doi: 10.2174/138620706776055539
16. [16]
  Y. Liu, B. Guo, X. Zou, Y. Li, S. Shi, Energy Storage Mater. 31 (2020) 434, https://doi.org/10.1016/j.ensm.2020.06.033. doi: 10.1016/j.ensm.2020.06.033
17. [17]
  Y. Liu, X. Zou, Z. Yang, S. Shi, J. Chin. Ceram. Soc. 50 (2022) 863, https://doi.org/10.14062/j.issn.0454-5648.20220093. doi: 10.14062/j.issn.0454-5648.20220093
18. [18]
  Y. Wang, W. Hou, N. Sheng, Z. Zhao, J. Liu, L. Huang, J. Wang, Artif. Intell. Rev. 57 (2024) 294, https://doi.org/10.1007/s10462-024-10918-9. doi: 10.1007/s10462-024-10918-9
19. [19]
  D. Mesquita, A. Souza, S. Kaski, Adv. Neural Inf. Process. Syst. 33 (2020) 2220, https://doi.org/10.48550/arXiv.2010.11418. doi: 10.48550/arXiv.2010.11418
20. [20]
  D. Grattarola, D. Zambon, F.M. Bianchi, C. Alippi, IEEE Trans. Neural Netw. Learn. Syst. 35 (2022) 2708, https://doi.org/10.1109/TNNLS.2022.3190922. doi: 10.1109/TNNLS.2022.3190922
21. [21]
  C. Liu, Y. Zhan, J. Wu, C. Li, B. Du, W. Hu, T. Liu, D. Tao, Graph Pooling for Graph Neural Networks: Progress, Challenges, and Opportunities, in Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence (IJCAI 2023), 2023, pp. 6712–6722, https://doi.org/10.24963/ijcai.2023/752.
22. [22]
  F. Xu, Z. Yang, L. Wang, D. Meng, J. Long, Brief. Bioinform. 25 (2024) bbad423, https://doi.org/10.1093/bib/bbad423. doi: 10.1093/bib/bbad423
23. [23]
  Z. Zhou, M.M. Rahman Siddiquee, N. Tajbakhsh, J. Liang, UNet++: A Nested U-Net Architecture for Medical Image Segmentation, in Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, Springer: Cham, Switzerland, 2018, pp. 3–11, https://link.springer.com/chapter/10.1007/978-3-030-00889-5_1.
24. [24]
  D.J. Pearce, An Improved Algorithm for Finding the Strongly Connected Components of a Directed Graph, (2005), https://api.semanticscholar.org/CorpusID:7781255.
25. [25]
  O. Ronneberger, P. Fischer, T. Brox, U-Net: Convolutional Networks for Biomedical Image Segmentation, in Medical Image Computing and Computer-Assisted Intervention - MICCAI 2015, Springer: Cham, Switzerland, 2015, pp. 234–241, https://doi.org/10.1007/978-3-319-24574-4_28.
26. [26]
  H. Gao, S. Ji, IEEE Trans. Pattern Anal. Mach. Intell. 44 (2022) 4948, https://doi.org/10.1109/TPAMI.2021.3081010. doi: 10.1109/TPAMI.2021.3081010
27. [27]
  C. Cangea, P. Veličković, N. Jovanović, T. Kipf, P. Liò, arXiv: 1811.01287, https://doi.org/10.48550/arXiv.1811.01287.
28. [28]
  G. Corso, L. Cavalleri, D. Beaini, P. Liò, P. Veličković, Adv. Neural Inf. Process. Syst. 33 (2020) 13260, https://doi.org/10.48550/arXiv.2004.05718. doi: 10.48550/arXiv.2004.05718
29. [29]
  W. Hu, B. Liu, J. Gomes, M. Zitnik, P. Liang, V. Pande, J. Leskovec, arXiv: 1905.12265, https://doi.org/10.48550/arXiv.1905.12265.
30. [30]
  Y. You, T. Chen, Y. Sui, T. Chen, Z. Wang, Y. Shen, Adv. Neural Inf. Process. Syst. 33 (2020) 5812, https://proceedings.neurips.cc/paper/2020/file/3fe230348e9a12c13120749e3f9fa4cd-Paper.pdf. https://proceedings.neurips.cc/paper/2020/file/3fe230348e9a12c13120749e3f9fa4cd-Paper.pdf
31. [31]
  R. Sun, H. Dai, A.W. Yu, Adv. Neural Inf. Process. Syst. 35 (2022) 12096, https://dl.acm.org/doi/10.5555/3600270.3601149. doi: 10.5555/3600270.3601149
32. [32]
  S. Zhang, Z. Yan, Y. Huang, L. Liu, D. He, W. Wang, X. Fang, X. Zhang, F. Wang, H. Wu, et al., Bioinformatics 38 (2022) 3444, https://doi.org/10.1093/bioinformatics/btac342. doi: 10.1093/bioinformatics/btac342
33. [33]
  W. Lin, C.C.A. Fung, ACS Omega 9 (2024) 20832, https://doi.org/10.1021/acsomega.3c09512. doi: 10.1021/acsomega.3c09512
34. [34]
  E. Inae, G. Liu, M. Jiang, arXiv: 2309.04589, https://doi.org/10.48550/arXiv.2309.04589.
35. [35]
  B. Fabian, T. Edlich, H. Gaspar, M. Segler, J. Meyers, M. Fiscato, M. Ahmed, arXiv: 2011.13230, https://doi.org/10.48550/arXiv.2011.13230.
36. [36]
  Y. Liu, Z. Yang, X. Zou, S. Ma, D. Liu, M. Avdeev, S. Shi, Natl. Sci. Rev. 10 (2023) nwad125, https://doi.org/10.1093/nsr/nwad125. doi: 10.1093/nsr/nwad125
37. [37]
  Y. Liu, S. Ma, Z. Yang, X. Zou, S. Shi, J. Chin. Ceram. Soc. 51 (2023) 427, https://doi.org/10.14062/j.issn.0454-5648.20220991. doi: 10.14062/j.issn.0454-5648.20220991
38. [38]
  Y. Liu, Z. Yang, X. Zou, Y. Lin, S. Ma, W. Zuo, Z. Zou, H. Wang, M. Avdeev, S. Shi, Mater. Sci. Eng.: R: Rep. 166 (2025) 101050, https://doi.org/10.1016/j.mser.2025.101050. doi: 10.1016/j.mser.2025.101050
39. [39]
  Z. Hou, X. Liu, Y. Cen, Y. Dong, H. Yang, C. Wang, J. Tang, GraphMAE: Self-Supervised Masked Graph Autoencoders, in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, Association for Computing Machinery: New York, NY, USA, 2022, pp. 594–604, https://doi.org/10.1145/3534678.3539321.
40. [40]
  J. Xia, C. Zhao, B. Hu, Z. Gao, C. Tan, Y. Liu, S. Li, S. Z. Li, Mole-BERT: Rethinking Pre-training Graph Neural Networks for Molecules, in the Eleventh International Conference on Learning Representations (ICLR2023), 2023 poster, https://openreview.net/forum?id=jevY-DtiZTR.
41. [41]
  Z. Zhang, Q. Liu, H. Wang, C. Lu, C.-K. Lee, Motif-based Graph Self-Supervised Learning for Molecular Property Prediction, in Proceedings of the 35th International Conference on Neural Information Processing Systems (NIPS2021), Curran Associates Inc. : Red Hook, NY, USA, 2021 pp. 15870–15882, https://dl.acm.org/doi/10.5555/3540261.3541475.
42. [42]
  G.W. Bemis, M.A. Murcko, J. Med. Chem. 39 (1996) 2887, https://doi.org/10.1021/jm9602928. doi: 10.1021/jm9602928
43. [43]
  Y. Liu, S. Ma, Z. Yang, D. Wu, Y. Zhao, M. Avdeev, S. Shi, J. Materiomics 11 (2025) 101066, https://doi.org/10.1016/j.jmat.2025.101066. doi: 10.1016/j.jmat.2025.101066
44. [44]
  H. Zhu, T.M. Martin, L. Ye, A. Sedykh, D.M. Young, A. Tropsha, Chem. Res. Toxicol. 22 (2009) 1913, https://doi.org/10.1021/tx900189p. doi: 10.1021/tx900189p
45. [45]
  M.L. Landry, J.J. Crawford, ACS Med. Chem. Lett. 11 (2019) 72, https://doi.org/10.1021/acsmedchemlett.9b00489. doi: 10.1021/acsmedchemlett.9b00489
46. [46]
  A.K. Nyamabo, H. Yu, J.-Y. Shi, Brief. Bioinform. 22 (2021) bbab133, https://doi.org/10.1093/bib/bbab133. doi: 10.1093/bib/bbab133
47. [47]
  P. Bai, F. Miljković, B. John, H. Lu, Nat. Mach. Intell 5 (2023) 126, https://doi.org/10.1038/s42256-022-00605-1. doi: 10.1038/s42256-022-00605-1
48. [48]
  M. Nickel, V. Tresp, H.-P. Kriegel, A Three-Way Model for Collective Learning on Multi-Relational Data, in Proceedings of the 28th International Conference on Machine Learning PMLR97 (ICML2021), Omnipress: Madison, WI, USA, 2011 pp. 809–816, https://dl.acm.org/doi/10.5555/3104482.3104584.
49. [49]
  J. Lee, I. Lee, J. Kang, Self-Attention Graph Pooling, in Proceedings of the 36th International Conference on Machine Learning (ICML2019), PMLR, 2019, pp. 3734-3743, https://proceedings.mlr.press/v97/lee19c/lee19c.pdf.
50. [50]
  T. Liu, Y. Lin, X. Wen, R.N. Jorissen, M.K. Gilson, Nucleic Acids Res. 35 (2007) D198, https://doi.org/10.1093/nar/gkl999. doi: 10.1093/nar/gkl999
51. [51]
  M. Zitnik, R. Sosic, S. Maheshwari, J. Leskovec, BioSNAP Datasets: Stanford Biomedical Network Dataset Collection, (2018), https://snap.stanford.edu/biodata/.
52. [52]
  H. Liu, J. Sun, J. Guan, J. Zheng, S. Zhou, Bioinformatics 31 (2015) i221, https://doi.org/10.1093/bioinformatics/btv256. doi: 10.1093/bioinformatics/btv256

扫一扫看文章

计量

PDF下载量: 0
文章访问数: 144
HTML全文浏览量: 13

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

MolUNet++：自适应粒度显式子结构与互作感知分子表示学习

通讯作者: yzws-123@xjtu.edu.cn (杨志伟); dymeng@mail.xjtu.edu.cn (孟德宇); Email: jglong@xjtu.edu.cn (龙建刚)

English

MolUNet++: Adaptive-grained explicit substructure and interaction aware molecular representation learning

Corresponding author: yzws-123@xjtu.edu.cn (Zhiwei Yang); dymeng@mail.xjtu.edu.cn (Deyu Meng); Email: jglong@xjtu.edu.cn (Jiangang Long)

CCS Chemistry：嵌段共聚物自组装成 “三明治”二维有机材料，实现纳米级压力传感功能

CCS Chemistry：颜徐州、鲍哲南：你的皮肤可能是一片SPMs

CCS Chemistry：工作高效不疲劳，只须让催化剂动起来

CCS Chemistry：静电组装导向聚合- 聚电解质纳米凝胶量化制备新策略

CCS Chemistry：金黄色葡萄球菌醛基脱氢酶是如何识别特异性底物的？

计量

通讯作者: 陈斌, bchen63@163.com

中国化学会秘书处

MolUNet++：自适应粒度显式子结构与互作感知分子表示学习

通讯作者: yzws-123@xjtu.edu.cn (杨志伟); dymeng@mail.xjtu.edu.cn (孟德宇); Email: jglong@xjtu.edu.cn (龙建刚)

English

MolUNet++: Adaptive-grained explicit substructure and interaction aware molecular representation learning

Corresponding author: yzws-123@xjtu.edu.cn (Zhiwei Yang); dymeng@mail.xjtu.edu.cn (Deyu Meng); Email: jglong@xjtu.edu.cn (Jiangang Long)

CCS Chemistry：嵌段共聚物自组装成 “三明治”二维有机材料，实现纳米级压力传感功能

CCS Chemistry：颜徐州、鲍哲南： 你的皮肤可能是一片SPMs

CCS Chemistry：工作高效不疲劳，只须让催化剂动起来

CCS Chemistry：静电组装导向聚合- 聚电解质纳米凝胶量化制备新策略

CCS Chemistry：金黄色葡萄球菌醛基脱氢酶是如何识别特异性底物的？

计量

文章相关

通讯作者: 陈斌, bchen63@163.com

中国化学会秘书处

CCS Chemistry：颜徐州、鲍哲南：你的皮肤可能是一片SPMs