{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,27]],"date-time":"2026-08-27T17:24:35Z","timestamp":1787851475184,"version":"build-2784847793"},"reference-count":41,"publisher":"MDPI AG","issue":"17","license":[{"start":{"date-parts":[[2026,8,27]],"date-time":"2026-08-27T00:00:00Z","timestamp":1787788800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"Hubei Provincial Key Research and Development Program","award":["2025BAB061"],"award-info":[{"award-number":["2025BAB061"]}]},{"name":"Hubei Provincial Major Science and Technology Program","award":["2025BEA001"],"award-info":[{"award-number":["2025BEA001"]}]},{"name":"Wuhan Natural Science Foundation Focused Program","award":["2026040101030022"],"award-info":[{"award-number":["2026040101030022"]}]},{"name":"Foundation of Hubei Provincial Department of Education","award":["Q20241006"],"award-info":[{"award-number":["Q20241006"]}]},{"award":["Q20241006"],"award-info":[{"award-number":["Q20241006"]}],"id":[{"id":"https:\/\/ror.org\/05yaa9j15","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Molecules"],"abstract":"<jats:p>Molecular property prediction is a fundamental task in drug discovery and chemical biology, where effective molecular representations are essential for accurate prediction. Learning transferable motif-level representations remains challenging because explicit motif annotations are scarce and existing representations may be altered during downstream supervised optimization. In this study, we propose MSGRL, a motif-driven self-supervised graph representation learning framework for interpretable molecular property prediction. MSGRL represents each molecule through a hierarchical graph structure, consisting of a motif-based graph for inter-motif organization and motif-specific atom-based graphs for intra-motif atomic structure. Its central design is to decouple label-agnostic intra-motif structural learning from label-dependent inter-motif property learning. An MPNN-GRU encoder is pretrained on motif-specific atom-based graphs using a variational motif graph autoencoder (VMGAE), which reconstructs the internal bond topology of motifs in a self-supervised manner. After pretraining, the intra-motif encoder is kept frozen, while the downstream module pools atom-level feature matrices into motif vectors, propagates information over the motif-based graph, and applies attention-based pooling for molecular property prediction. This design keeps the pretrained intra-motif representations fixed while allowing the inter-motif network and prediction head to adapt to individual downstream tasks. Experiments on eight MoleculeNet benchmark datasets show that MSGRL achieves the highest ROC-AUC scores on all five classification datasets and the lowest RMSE on Lipophilicity, while its performance on ESOL and FreeSolv is more mixed. Ablation studies further support the contributions of encoder freezing, motif-based graph construction, and attention-based pooling. Motif-level attribution analyses provide qualitative and dataset-level evidence regarding the substructures emphasized by the model. These results demonstrate the effectiveness of the proposed hierarchical representation strategy, particularly for the evaluated molecular classification tasks.<\/jats:p>","DOI":"10.3390\/molecules31173008","type":"journal-article","created":{"date-parts":[[2026,8,27]],"date-time":"2026-08-27T16:47:13Z","timestamp":1787849233000},"page":"3008","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["MSGRL: A Motif-Driven Self-Supervised Graph Representation Learning Framework for Interpretable Molecular Property Prediction"],"prefix":"10.3390","volume":"31","author":[{"given":"You","family":"Wu","sequence":"first","affiliation":[{"name":"School of Artificial Intelligence, Hubei University, Wuhan 430062, China"},{"name":"School of Computer Science, Hubei University, Wuhan 430062, China"},{"name":"College of Informatics, Huazhong Agricultural University, Wuhan 430070, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0007-5357-8550","authenticated-orcid":false,"given":"Yuxin","family":"Jiang","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence, Hubei University, Wuhan 430062, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaoyun","family":"Qi","sequence":"additional","affiliation":[{"name":"School of Chemistry and Chemical Engineering, Hubei University, Wuhan 430062, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9673-6845","authenticated-orcid":false,"given":"Haitao","family":"Fu","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence, Hubei University, Wuhan 430062, China"},{"name":"Key Laboratory of Intelligent Sensing System and Security (Hubei University), Ministry of Education, Wuhan 430062, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Qiyu","family":"Tang","sequence":"additional","affiliation":[{"name":"The Roslin Institute, University of Edinburgh, Edinburgh EH25 9RG, UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0446-4040","authenticated-orcid":false,"given":"Wen","family":"Wang","sequence":"additional","affiliation":[{"name":"Institute for Neuroscience and Cardiovascular Research, Centre for Cardiovascular Science, University of Edinburgh, Edinburgh EH16 4TJ, UK"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Cheng","family":"Zeng","sequence":"additional","affiliation":[{"name":"School of Artificial Intelligence, Hubei University, Wuhan 430062, China"},{"name":"Key Laboratory of Intelligent Sensing System and Security (Hubei University), Ministry of Education, Wuhan 430062, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Guosheng","family":"Zhu","sequence":"additional","affiliation":[{"name":"School of Computer Science, Hubei University, Wuhan 430062, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,8,27]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","unstructured":"Dana, D., Gadhiya, S.V., St. Surin, L.G., Li, D., Naaz, F., Ali, Q., Paka, L., Yamin, M.A., Narayan, M., and Goldberg, I.D. (2018). Deep Learning in Drug Discovery and Medicine; Scratching the Surface. Molecules, 23.","DOI":"10.3390\/molecules23092384"},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"463","DOI":"10.1038\/s41573-019-0024-5","article-title":"Applications of machine learning in drug discovery and development","volume":"18","author":"Vamathevan","year":"2019","journal-title":"Nat. Rev. Drug Discov."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"2017","DOI":"10.1016\/j.drudis.2019.07.006","article-title":"Deep learning in drug discovery: Opportunities, challenges and future prospects","volume":"24","author":"Lavecchia","year":"2019","journal-title":"Drug Discov. Today"},{"key":"ref_4","unstructured":"Kipf, T.N., and Welling, M. (2017, January 24\u201326). Semi-Supervised Classification with Graph Convolutional Networks. Proceedings of the International Conference on Learning Representations, Toulon, France."},{"key":"ref_5","unstructured":"Xu, K., Hu, W., Leskovec, J., and Jegelka, S. (2019, January 6\u20139). How Powerful are Graph Neural Networks?. Proceedings of the International Conference on Learning Representations, New Orleans, LA, USA."},{"key":"ref_6","first-page":"1263","article-title":"Neural message passing for Quantum chemistry","volume":"Volume 70","author":"Gilmer","year":"2017","journal-title":"ICML\u201917: Proceedings of the 34th International Conference on Machine Learning"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"8749","DOI":"10.1021\/acs.jmedchem.9b00959","article-title":"Pushing the Boundaries of Molecular Representation for Drug Discovery with the Graph Attention Mechanism","volume":"63","author":"Xiong","year":"2020","journal-title":"J. Med. Chem."},{"key":"ref_8","unstructured":"Dwivedi, V.P., Luu, A.T., Laurent, T., Bengio, Y., and Bresson, X. (2022, January 25\u201329). Graph Neural Networks with Learnable Structural and Positional Representations. Proceedings of the International Conference on Learning Representations, Online."},{"key":"ref_9","unstructured":"Li, G., Xiong, C., Qian, G., Thabet, A., and Ghanem, B. (2022). DeeperGCN: All You Need to Train Deeper GCNs. arXiv."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"11614","DOI":"10.1109\/TPAMI.2025.3598461","article-title":"Identifying Semantic Component for Robust Molecular Property Prediction","volume":"47","author":"Li","year":"2025","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"119471","DOI":"10.1016\/j.ces.2023.119471","article-title":"Insights into deep learning framework for molecular property prediction based on different tokenization algorithms","volume":"285","author":"Yan","year":"2024","journal-title":"Chem. Eng. Sci."},{"key":"ref_12","doi-asserted-by":"crossref","first-page":"938","DOI":"10.1016\/j.neunet.2023.06.030","article-title":"Domain-informed graph neural networks: A quantum chemistry case study","volume":"165","author":"Morgan","year":"2023","journal-title":"Neural Netw."},{"key":"ref_13","doi-asserted-by":"crossref","first-page":"94","DOI":"10.1016\/j.neunet.2023.05.039","article-title":"3D graph neural network with few-shot learning for predicting drug-drug interactions in scaffold-based cold start scenario","volume":"165","author":"Lv","year":"2023","journal-title":"Neural Netw."},{"key":"ref_14","unstructured":"Xu, G., Jiang, Y., Lei, P., Yang, Y., and Chen, J. (2024, January 7\u201311). GTMGC: Using Graph Transformer to Predict Molecule\u2019s Ground-State Conformation. Proceedings of the Twelfth International Conference on Learning Representations, Vienna, Austria."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"7104","DOI":"10.1038\/s41467-024-51321-w","article-title":"Data-driven quantum chemical property prediction leveraging 3D conformations with Uni-Mol+","volume":"15","author":"Lu","year":"2024","journal-title":"Nat. Commun."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"1527","DOI":"10.1016\/j.chempr.2020.05.014","article-title":"Learning to Make Chemical Predictions: The Interplay of Feature Representation, Data, and Machine Learning Methods","volume":"6","author":"Haghighatlari","year":"2020","journal-title":"Chem"},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"161","DOI":"10.1080\/17460441.2026.2623154","article-title":"The expectations of in silico fragment-based drug design and future challenges","volume":"21","author":"Chikhale","year":"2026","journal-title":"Expert Opin. Drug Discov."},{"key":"ref_18","doi-asserted-by":"crossref","first-page":"bbad305","DOI":"10.1093\/bib\/bbad305","article-title":"HimGNN: A novel hierarchical molecular graph representation learning framework for property prediction","volume":"24","author":"Han","year":"2023","journal-title":"Brief. Bioinform."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"2981","DOI":"10.1093\/bioinformatics\/btab195","article-title":"FraGAT: A fragment-oriented multi-scale graph attention model for molecular property prediction","volume":"37","author":"Zhang","year":"2021","journal-title":"Bioinformatics"},{"key":"ref_20","first-page":"15870","article-title":"Motif-based Graph Self-Supervised Learning for Molecular Property Prediction","volume":"Volume 34","author":"Ranzato","year":"2021","journal-title":"Proceedings of the Advances in Neural Information Processing Systems"},{"key":"ref_21","doi-asserted-by":"crossref","first-page":"34","DOI":"10.1038\/s42004-023-00825-5","article-title":"Hierarchical Molecular Graph Self-Supervised Learning for Property Prediction","volume":"6","author":"Zang","year":"2023","journal-title":"Commun. Chem."},{"key":"ref_22","first-page":"15980","article-title":"MOTIF-Driven Contrastive Learning of Graph Representations","volume":"35","author":"Zhang","year":"2021","journal-title":"Proc. AAAI Conf. Artif. Intell."},{"key":"ref_23","doi-asserted-by":"crossref","unstructured":"Long, W., Zhai, S., Liu, Y., Li, M., and Wen, Z. (2026). Motif-Level Graph Learning Enables Interpretable Prediction of Drug-Induced QT Prolongation via Cooperative Substructural Determinants. Int. J. Mol. Sci., 27.","DOI":"10.3390\/ijms27114706"},{"key":"ref_24","doi-asserted-by":"crossref","first-page":"513","DOI":"10.1039\/C7SC02664A","article-title":"MoleculeNet: A benchmark for molecular machine learning","volume":"9","author":"Wu","year":"2018","journal-title":"Chem. Sci."},{"key":"ref_25","unstructured":"Kipf, T.N., and Welling, M. (2016). Variational Graph Auto-Encoders. arXiv."},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Zhang, X., Chen, C., Wang, X., Jiang, H., Zhao, W., and Cui, X. (2025). MetaGIN: A lightweight framework for molecular property prediction. Front. Comput. Sci., 19.","DOI":"10.1007\/s11704-024-3784-y"},{"key":"ref_27","unstructured":"Gasteiger, J., Becker, F., and G\u00fcnnemann, S. (2021, January 6\u201314). GemNet: Universal Directional Graph Neural Networks for Molecules. Proceedings of the Conference on Neural Information Processing Systems (NeurIPS), Online."},{"key":"ref_28","doi-asserted-by":"crossref","first-page":"5021","DOI":"10.1038\/s41467-025-59917-6","article-title":"DeepDTAGen: A multitask deep learning framework for drug-target affinity prediction and target-aware drugs generation","volume":"16","author":"Shah","year":"2025","journal-title":"Nat. Commun."},{"key":"ref_29","unstructured":"Schr\u00f6dinger, LLC (2015). The PyMOL Molecular Graphics System, Version 1.8, Schr\u00f6dinger, LLC."},{"key":"ref_30","doi-asserted-by":"crossref","first-page":"3891","DOI":"10.1021\/acs.jcim.1c00203","article-title":"AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings","volume":"61","author":"Eberhardt","year":"2021","journal-title":"J. Chem. Inf. Model."},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"1936","DOI":"10.1021\/acs.jcim.6b00290","article-title":"Computational Modeling of \u03b2-Secretase 1 (BACE-1) Inhibitors Using Ligand Based Approaches","volume":"56","author":"Subramanian","year":"2016","journal-title":"J. Chem. Inf. Model."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"1686","DOI":"10.1021\/ci300124c","article-title":"A Bayesian approach to in silico blood-brain barrier penetration modeling","volume":"52","author":"Martins","year":"2012","journal-title":"J. Chem. Inf. Model."},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"1294","DOI":"10.1016\/j.chembiol.2016.07.023","article-title":"A Data-Driven Approach to Predicting Successes and Failures of Clinical Trials","volume":"23","author":"Gayvert","year":"2016","journal-title":"Cell Chem. Biol."},{"key":"ref_34","doi-asserted-by":"crossref","unstructured":"Artemov, A.V., Putin, E., Vanhaelen, Q., Aliper, A., Ozerov, I.V., and Zhavoronkov, A. (2016). Integrated deep learned transcriptomic and structure-based predictor of clinical trials outcomes. bioRxiv.","DOI":"10.1101\/095653"},{"key":"ref_35","doi-asserted-by":"crossref","first-page":"D1075","DOI":"10.1093\/nar\/gkv1075","article-title":"The SIDER database of drugs and side effects","volume":"44","author":"Kuhn","year":"2016","journal-title":"Nucleic Acids Res."},{"key":"ref_36","unstructured":"National Institutes of Health (NIH) (2025, July 26). Tox21 Challenge, Available online: https:\/\/tripod.nih.gov\/tox21\/challenge\/."},{"key":"ref_37","doi-asserted-by":"crossref","first-page":"1000","DOI":"10.1021\/ci034243x","article-title":"ESOL: Estimating Aqueous Solubility Directly from Molecular Structure","volume":"44","author":"Delaney","year":"2004","journal-title":"J. Chem. Inf. Comput. Sci."},{"key":"ref_38","doi-asserted-by":"crossref","first-page":"711","DOI":"10.1007\/s10822-014-9747-x","article-title":"FreeSolv: A database of experimental and calculated hydration free energies, with input files","volume":"28","author":"Mobley","year":"2014","journal-title":"J. Comput.-Aided Mol. Des."},{"key":"ref_39","doi-asserted-by":"crossref","first-page":"D930","DOI":"10.1093\/nar\/gky1075","article-title":"ChEMBL: Towards direct deposition of bioassay data","volume":"47","author":"Mendez","year":"2019","journal-title":"Nucleic Acids Res."},{"key":"ref_40","doi-asserted-by":"crossref","first-page":"1503","DOI":"10.1002\/cmdc.200800178","article-title":"On the Art of Compiling and Using \u2018Drug-Like\u2019 Chemical Fragment Spaces","volume":"3","author":"Degen","year":"2008","journal-title":"ChemMedChem"},{"key":"ref_41","doi-asserted-by":"crossref","unstructured":"Ansel, J., Yang, E., He, H., Gimelshein, N., Jain, A., Voznesensky, M., Bao, B., Bell, P., Berard, D., and Burovski, E. (2024). PyTorch 2: Faster Machine Learning Through Dynamic Python Bytecode Transformation and Graph Compilation. Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, (ASPLOS \u201924), ACM.","DOI":"10.1145\/3620665.3640366"}],"container-title":["Molecules"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1420-3049\/31\/17\/3008\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,27]],"date-time":"2026-08-27T16:47:22Z","timestamp":1787849242000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1420-3049\/31\/17\/3008"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,27]]},"references-count":41,"journal-issue":{"issue":"17","published-online":{"date-parts":[[2026,9]]}},"alternative-id":["molecules31173008"],"URL":"https:\/\/doi.org\/10.3390\/molecules31173008","relation":{},"ISSN":["1420-3049"],"issn-type":[{"value":"1420-3049","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,27]]}}}