{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,28]],"date-time":"2026-08-28T07:57:46Z","timestamp":1787903866196,"version":"build-2784847793"},"reference-count":46,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T00:00:00Z","timestamp":1753833600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Mol. Biosci."],"abstract":"<jats:p>Accurate molecular property prediction is fundamental to modern drug discovery and materials design. However, prevailing computational methods are often insufficient, as they rely on single-granularity structural representations that fail to capture the hierarchical complexity of molecular systems. To address this challenge, we propose a new approach to molecular representation learning that incorporates structural information across multiple scales. We design DFusMol (Dual Fusion with Global and Local Attention), a novel framework inspired by multi-modal learning. DFusMol employs graph encoders to capture features from both atomic-level molecular graphs and motif-level graphs derived from chemical rules. A customized global-local attention mechanism then blends these diverse features to build comprehensive molecular representations. Experiments on nine public benchmark datasets reveal that DFusMol delivers top-tier predictive performance across all tasks, outperforming state-of-the-art self-supervised learning models on six of them. By effectively integrating atomic- and motif-level information, DFusMol provides an innovative and efficient solution for molecular property prediction, enhancing representation learning methodologies and demonstrating strong potential for applications in drug design and lead compound screening.<\/jats:p>","DOI":"10.3389\/fmolb.2025.1623620","type":"journal-article","created":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T04:13:36Z","timestamp":1753848816000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["DFusMol: predicting molecular properties based on dual-channel attention"],"prefix":"10.3389","volume":"12","author":[{"given":"Xuan","family":"Liu","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wei","family":"Du","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Haibao","family":"Tang","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yingjian","family":"Gu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Zhibang","family":"Li","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiaoyang","family":"Fu","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,7,30]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"186","DOI":"10.1126\/science.aar5169","article-title":"Predicting reaction performance in c\u2013n cross-coupling using machine learning","volume":"360","author":"Ahneman","year":"2018","journal-title":"Science"},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1706.03762","article-title":"Attention is all you need","volume":"30","author":"Ashish","year":"2017","journal-title":"Adv. neural Inf. Process. Syst."},{"key":"B3","doi-asserted-by":"crossref","first-page":"144","DOI":"10.1145\/130385.130401","article-title":"A training algorithm for optimal margin classifiers","volume-title":"Proceedings of the fifth annual workshop on Computational learning theory","author":"Boser","year":"1992"},{"key":"B4","doi-asserted-by":"publisher","first-page":"5","DOI":"10.1023\/a:1010933404324","article-title":"Random forests","volume":"45","author":"Breiman","year":"2001","journal-title":"Mach. Learn."},{"key":"B5","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1409.1259","article-title":"On the properties of neural machine translation: encoder-decoder approaches","author":"Cho","year":"2014","journal-title":"arXiv Prepr. arXiv:1409"},{"key":"B6","doi-asserted-by":"publisher","first-page":"6395","DOI":"10.1038\/s41467-023-41948-6","article-title":"A systematic study of key elements underlying molecular property prediction","volume":"14","author":"Deng","year":"2023","journal-title":"Nat. Commun."},{"key":"B7","doi-asserted-by":"publisher","first-page":"495","DOI":"10.1038\/d41573-019-00074-z","article-title":"Trends in clinical success rates and therapeutic focus","volume":"18","author":"Dowden","year":"2019","journal-title":"Nat. Rev. Drug Discov."},{"key":"B8","doi-asserted-by":"publisher","first-page":"127","DOI":"10.1038\/s42256-021-00438-4","article-title":"Geometry-enhanced molecular representation learning for property prediction","volume":"4","author":"Fang","year":"2022","journal-title":"Nat. Mach. Intell."},{"key":"B9","doi-asserted-by":"publisher","first-page":"542","DOI":"10.1038\/s42256-023-00654-0","article-title":"Knowledge graph-enhanced molecular contrastive learning with functional prompt","volume":"5","author":"Fang","year":"2023","journal-title":"Nat. Mach. Intell."},{"key":"B10","doi-asserted-by":"publisher","first-page":"653","DOI":"10.1038\/s41570-022-00416-3","article-title":"Extending machine learning beyond interatomic potentials for predicting molecular properties","volume":"6","author":"Fedik","year":"2022","journal-title":"Nat. Rev. Chem."},{"key":"B11","doi-asserted-by":"publisher","first-page":"bbad306","DOI":"10.1093\/bib\/bbad306","article-title":"Transfoxmol: predicting molecular property with focused attention","volume":"24","author":"Gao","year":"2023","journal-title":"Briefings Bioinforma."},{"key":"B12","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2003.03123","article-title":"Directional message passing for molecular graphs","author":"Gasteiger","year":"2020","journal-title":"arXiv Prepr. arXiv:2003.03123"},{"key":"B13","doi-asserted-by":"publisher","first-page":"D1100","DOI":"10.1093\/nar\/gkr777","article-title":"Chembl: a large-scale bioactivity database for drug discovery","volume":"40","author":"Gaulton","year":"2012","journal-title":"Nucleic acids Res."},{"key":"B14","doi-asserted-by":"crossref","first-page":"199","DOI":"10.1007\/978-3-030-40245-7_10","article-title":"Message passing neural networks","volume-title":"Machine learning meets quantum physics","author":"Gilmer","year":"2020"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2207.04869","article-title":"Graph-based molecular representation learning","author":"Guo","year":"2022","journal-title":"arXiv Prepr. arXiv:2207.04869"},{"key":"B16","doi-asserted-by":"publisher","first-page":"76","DOI":"10.1021\/ci00001a012","article-title":"The electrotopological state: structure information at the atomic level for molecular graphs","volume":"31","author":"Hall","year":"1991","journal-title":"J. Chem. Inf. Comput. Sci."},{"key":"B17","doi-asserted-by":"publisher","first-page":"5052","DOI":"10.1039\/d4sc90043j","article-title":"A focus on molecular representation learning for the prediction of chemical properties","volume":"15","author":"Harnik","year":"2024","journal-title":"Chem. Sci."},{"key":"B18","doi-asserted-by":"publisher","first-page":"40","DOI":"10.1038\/nbt.2786","article-title":"Clinical development success rates for investigational drugs","volume":"32","author":"Hay","year":"2014","journal-title":"Nat. Biotechnol."},{"key":"B19","doi-asserted-by":"publisher","first-page":"1668","DOI":"10.3390\/molecules27051668","article-title":"Accurate physical property predictions via deep learning","volume":"27","author":"Hou","year":"2022","journal-title":"Molecules"},{"key":"B20","doi-asserted-by":"publisher","first-page":"175","DOI":"10.1038\/nature08506","article-title":"Predicting new molecular targets for known drugs","volume":"462","author":"Keiser","year":"2009","journal-title":"Nature"},{"key":"B21","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1609.02907","article-title":"Semi-supervised classification with graph convolutional networks","author":"Kipf","year":"2016","journal-title":"arXiv Prepr. arXiv:1609.02907"},{"key":"B22","unstructured":"Rdkit: open-source cheminformatics software\n          \n          \n            \n              Landrum\n              G.\n            \n          \n          \n          2016"},{"key":"B23","doi-asserted-by":"publisher","first-page":"645","DOI":"10.1038\/s42256-022-00501-8","article-title":"An adaptive graph learning method for automated molecular interactions and properties predictions","volume":"4","author":"Li","year":"2022","journal-title":"Nat. Mach. Intell."},{"key":"B24","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2110.00987","article-title":"Motif-based graph self-supervised learning for molecular property prediction","author":"Liu","year":"2021","journal-title":"arXiv Prepr. arXiv:2110.07728"},{"key":"B25","doi-asserted-by":"publisher","first-page":"1169","DOI":"10.1038\/s42256-024-00900-z","article-title":"Pre-training with fractional denoising to enhance molecular property prediction","volume":"6","author":"Ni","year":"2024","journal-title":"Nat. Mach. Intell."},{"key":"B26","article-title":"MoleculeNet: a benchmark for molecular machine learning","author":"Ramsundar","year":"2019","journal-title":"Deep Learn. life Sci. Appl. deep Learn. genomics, Microsc. drug Discov. more (O\u2019Reilly Media)"},{"key":"B27","first-page":"12559","article-title":"Self-supervised graph transformer on large-scale molecular data","volume":"33","author":"Rong","year":"2020","journal-title":"Adv. Neural. Inf. Process Syst."},{"key":"B28","doi-asserted-by":"publisher","first-page":"1256","DOI":"10.1038\/s42256-022-00580-7","article-title":"Large-scale chemical language representations capture molecular structure and properties","volume":"4","author":"Ross","year":"2022","journal-title":"Nat. Mach. Intell."},{"key":"B29","doi-asserted-by":"publisher","first-page":"673","DOI":"10.1038\/s41586-023-05905-z","article-title":"Computational approaches streamlining drug discovery","volume":"616","author":"Sadybekov","year":"2023","journal-title":"Nature"},{"key":"B30","doi-asserted-by":"publisher","first-page":"1572","DOI":"10.1021\/acscentsci.9b00576","article-title":"Molecular transformer: a model for uncertainty-calibrated chemical reaction prediction","volume":"5","author":"Schwaller","year":"2019","journal-title":"ACS central Sci."},{"key":"B31","first-page":"2831","article-title":"Communicative representation learning on attributed molecular graphs","volume-title":"Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence, IJCAI-20 2831-2838","author":"Song","year":"2020"},{"key":"B32","doi-asserted-by":"publisher","first-page":"6259","DOI":"10.1021\/acs.jcim.4c00747","article-title":"Transformers for molecular property prediction: lessons learned from the past five years","volume":"64","author":"Sultan","year":"2024","journal-title":"J. Chem. Inf. Model."},{"key":"B33","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1710.10903","article-title":"Graph attention networks","author":"Veli\u010dkovi\u0107","year":"2017","journal-title":"arXiv Prepr. arXiv:1710.10903"},{"key":"B34","doi-asserted-by":"publisher","first-page":"279","DOI":"10.1038\/s42256-022-00447-x","article-title":"Molecular contrastive learning of representations via graph neural networks","volume":"4","author":"Wang","year":"2022","journal-title":"Nat. Mach. Intell."},{"key":"B35","doi-asserted-by":"publisher","first-page":"201","DOI":"10.1016\/j.neucom.2016.02.022","article-title":"Protein\u2013protein interaction sites prediction by ensembling svm and sample-weighted random forests","volume":"193","author":"Wei","year":"2016","journal-title":"Neurocomputing"},{"key":"B36","doi-asserted-by":"publisher","first-page":"97","DOI":"10.1021\/ci00062a008","article-title":"Smiles. 2. algorithm for generation of unique smiles notation","volume":"29","author":"Weininger","year":"1989","journal-title":"J. Chem. Inf. Comput. Sci."},{"key":"B37","doi-asserted-by":"publisher","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":"B38","doi-asserted-by":"publisher","first-page":"00826","DOI":"10.48550\/arXiv.1810.00826","article-title":"How powerful are graph neural networks?","author":"Xu","year":"2018","journal-title":"arXiv Prepr. arXiv:1810"},{"key":"B39","doi-asserted-by":"publisher","first-page":"2268","DOI":"10.1021\/acs.jcim.4c01371","article-title":"Large language models as tools for molecular toxicity prediction: ai insights into cardiotoxicity","volume":"65","author":"Yang","year":"2025","journal-title":"J. Chem. Inf. Model."},{"key":"B40","first-page":"25581","article-title":"Molecular representation learning via heterogeneous motif graph neural networks","volume-title":"International conference on machine learning","author":"Yu","year":"2022"},{"key":"B41","doi-asserted-by":"publisher","first-page":"1004","DOI":"10.1038\/s42256-022-00557-6","article-title":"Accurate prediction of molecular properties and drug targets using a self-supervised image representation learning framework","volume":"4","author":"Zeng","year":"2022","journal-title":"Nat. Mach. Intell."},{"key":"B42","doi-asserted-by":"publisher","first-page":"1216765","DOI":"10.3389\/fmolb.2023.1216765","article-title":"Smg-bert: integrating stereoscopic information and chemical representation for molecular property prediction","volume":"10","author":"Zhang","year":"2023","journal-title":"Front. Mol. Biosci."},{"key":"B43","doi-asserted-by":"publisher","first-page":"195912","DOI":"10.1007\/s11704-024-3784-y","article-title":"Metagin: a lightweight framework for molecular property prediction","volume":"19","author":"Zhang","year":"2025","journal-title":"Front. Comput. Sci."},{"key":"B44","first-page":"15870","article-title":"Motif-based graph self-supervised learning for molecular property prediction","volume":"34","author":"Zhang","year":"2021","journal-title":"Adv. Neural Inf. Process. Syst."},{"key":"B45","doi-asserted-by":"publisher","first-page":"1094","DOI":"10.1038\/s42256-024-00888-6","article-title":"Learning motif-based graphs for drug\u2013drug interaction prediction via local\u2013global self-attention","volume":"6","author":"Zhong","year":"2024","journal-title":"Nat. Mach. Intell."},{"key":"B46","doi-asserted-by":"crossref","DOI":"10.26434\/chemrxiv-2022-jjm0j-v4","article-title":"Uni-mol: a universal 3d molecular representation learning framework","author":"Zhou","year":"2023"}],"container-title":["Frontiers in Molecular Biosciences"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmolb.2025.1623620\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,7,30]],"date-time":"2025-07-30T04:13:39Z","timestamp":1753848819000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fmolb.2025.1623620\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,7,30]]},"references-count":46,"alternative-id":["10.3389\/fmolb.2025.1623620"],"URL":"https:\/\/doi.org\/10.3389\/fmolb.2025.1623620","relation":{},"ISSN":["2296-889X"],"issn-type":[{"value":"2296-889X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,7,30]]},"article-number":"1623620"}}