EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC 4.0

Advancing ADMET prediction through multiscale fragment-aware pretraining with MSformer-ADMET

Huihui Liu iD, Bingjie Zhu, Shuyang Nie, Haoran Li, Yugang Lin, Tianyi Ma, Xin Shao iD, Qian Chen, Minjie Shen, Yanrong Zheng, Xiaohui Fan iD, Jie Liao

Briefings in Bioinformatics · 31.08.2025 · 26 (5)

DOI: 10.1093/bib/bbaf506 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Briefings in Bioinformatics, online erschienen am 29.09.2025. Verfasst von Huihui Liu und 11 weiteren Autor:innen. Beteiligte Einrichtung: Department of Pharmaceutical Sciences, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, 310058, China. Hauptthemen laut MEDLINE: Drug Discovery, Deep Learning, Pharmacokinetics, Software. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0). Gefördert von: Starlit South Lake Leading Elite Program, National Natural Science Foundation of China, Zhejiang Provincial Natural Science Foundation of China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties are critical determinants of the pharmacokinetic and safety profiles of drug candidates. Accurate and early-stage prediction of ADMET characteristics is essential for reducing late-stage attrition rates, lowering development costs, and accelerating the drug discovery process. Recent advances in deep learning have shown great promise in molecular property prediction, especially with the emergence of Transformer-based architectures that can effectively model long-range dependencies in molecular representations. However, most existing methods rely heavily on atom-level encodings (e.g. smiles or molecular graphs), which often lack structural interpretability and generalization across heterogeneous tasks. Previously, we developed a de novo and flexible molecular representation framework named MSformer (available at https://github.com/ZJUFanLab/MSformer), which demonstrated success in bioactivity prediction. We have now adapted and specialized this architecture for ADMET property prediction. This adapted implementation, designated as MSformer-ADMET, extends the framework’s capabilities to pharmacokinetic and toxicity endpoints while maintaining its flexible, fragmentation-based approach to molecular representation learning. MSformer-ADMET is fine-tuned on 22 tasks collected from the Therapeutics Data Commons (TDC), covering both classification and regression settings. Results demonstrate that MSformer-ADMET achieves superior performance across a wide range of ADMET endpoints, consistently outperforming conventional smiles-based and graph-based models. Notably, we further conducted interpretability analyses by leveraging the model’s attention distributions and fragment-to-atom mappings, allowing the identification of key structural fragments that are highly associated with molecular properties. This post hoc interpretability provides more transparent insights into the structure–property relationship. Collectively, results demonstrate that MSformer-ADMET is a highly effective and broadly applicable model for ADMET prediction.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Drug DiscoveryDeep LearningPharmacokineticsSoftwareHumans

Stichwörter der Autor:innen

ADMETdeep learningdrug discoverymeta structurenatural productstransformer

Autor:innen und Institutionen

  • Department of Pharmaceutical Sciences, College of Pharmaceutical Sciences, Zhejiang University, Hangzhou, 310058, China.
  • Zhejiang Key Laboratory of Chinese Medicine Modernization, Innovation Center of Yangtze River Delta, Zhejiang University, Jiaxing, 314100, China.
  • State Key Laboratory of Chinese Medicine Modernization, Innovation Center of Yangtze River Delta, Zhejiang University, Jiaxing, 314100, China.
  • Department of Pharmacy, Affiliated Jinhua Hospital, Zhejiang University School of Medicine, Jinhua, 321000, China.
  • Hangzhou TCM Hospital Affiliated to Zhejiang Chinese Medical University, Hangzhou, 310007, China.
  • Zhejiang Collaborative Innovation Center for the Brain Diseases with Integrative Medicine, Zhejiang Key Laboratory of Neuropsychopharmacology, Zhejiang Chinese Medical University, Hangzhou, 310053, China.

ORCID-Profile: Huihui Liu, Xin Shao, Xiaohui Fan

Förderung

  • Starlit South Lake Leading Elite Program
  • National Natural Science Foundation of China
  • Zhejiang Provincial Natural Science Foundation of China
  • 'Pioneer' and 'Leading Goose' R&D Program of Zhejiang
  • Noncommunicable Chronic Diseases-National Science and Technology Major Project
  • ‘Pioneer’ and ‘Leading Goose’ R&D Program of Zhejiang

Zitieren

APA

Liu, H., Zhu, B., Nie, S., Li, H., Lin, Y., Ma, T., Shao, X., Chen, Q., Shen, M., Zheng, Y., Fan, X., & Liao, J. (2025). Advancing ADMET prediction through multiscale fragment-aware pretraining with MSformer-ADMET. Briefings in Bioinformatics, 26(5). https://doi.org/10.1093/bib/bbaf506

Vancouver

Liu H, Zhu B, Nie S, Li H, Lin Y, Ma T, et al. Advancing ADMET prediction through multiscale fragment-aware pretraining with MSformer-ADMET. Briefings in Bioinformatics. 2025;26(5). doi:10.1093/bib/bbaf506

BibTeX
@article{Liu2025,
  author = {Liu, Huihui and Zhu, Bingjie and Nie, Shuyang and Li, Haoran and Lin, Yugang and Ma, Tianyi and Shao, Xin and Chen, Qian and Shen, Minjie and Zheng, Yanrong and Fan, Xiaohui and Liao, Jie},
  title = {Advancing ADMET prediction through multiscale fragment-aware pretraining with MSformer-ADMET},
  journal = {Briefings in Bioinformatics},
  year = {2025},
  volume = {26},
  number = {5},
  doi = {10.1093/bib/bbaf506}
}
Kostenlos starten →