EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

GTAM: A Molecular Pretraining Model with Geometric Triangle Awareness

Xiaoyang Hou iD, Tian Zhu, Milong Ren, Bo Duan, Chunming Zhang, Dongbo Bu iD, Shiwei Sun iD

Bioinformatics · 2024

DOI: 10.1093/bioinformatics/btae524 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioinformatics, 2024. Verfasst von Xiaoyang Hou und 6 weiteren Autor:innen. Beteiligte Einrichtung: Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, Beijing, 100049, China. Hauptthemen laut MEDLINE: Deep Learning. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, National Key Research and Development Program of China, Chongqing Municipal Key Special Project for Technological Innovation and Application Development.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Motivation Molecular representation learning is pivotal for advancing deep learning applications in quantum chemistry and drug discovery. Existing methods for molecular representation learning often fall short of fully capturing the intricate interactions within chemical bonds of 2D topological graphs and the multifaceted effects of 3D geometric conformations. Results To overcome these challenges, we present a novel contrastive learning strategy for molecular representation learning, named Geometric Triangle Awareness Model (GTAM). This method integrates innovative molecular encoders for both 2D graphs and 3D conformations, enabling the accurate capture of geometric dependencies among edges in graph-based molecular structures. Furthermore, GTAM is bolstered by the development of two contrastive training objectives designed to facilitate the direct transfer of edge information between 2D topological graphs and 3D geometric conformations, enhancing the functionality of the molecular encoders. Through extensive evaluations on a range of 2D and 3D downstream tasks, our model has demonstrated superior performance over existing approaches. Availability The test code and data of GTAM are available online at https://github.com/StellaHxy/GTAM. Supplementary information Comprehensive implementation details, datasets, and additional experiments can be found in Supplementary Materials A-C.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningMolecular ConformationModels, MolecularDrug Discovery

Autor:innen und Institutionen

  • Key Laboratory of Intelligent Information Processing, Institute of Computing Technology, Beijing, 100049, China.
  • School of Computer Science and Technology, University of Chinese Academy of Sciences, Beijing, 100049, China.
  • Western Institute of Computing Technology, Chongqing, 400000, China.
  • Phil Rivers Technology, Beijing, 100190, China.
  • Central China Artificial Intelligence Research Institute, Henan Academy of Sciences, Zhengzhou, 450046, China.

ORCID-Profile: Xiaoyang Hou, Dongbo Bu, Shiwei Sun

Förderung

  • National Natural Science Foundation of China
  • National Key Research and Development Program of China
  • Chongqing Municipal Key Special Project for Technological Innovation and Application Development

Zitieren

APA

Hou, X., Zhu, T., Ren, M., Duan, B., Zhang, C., Bu, D., & Sun, S. (2024). GTAM: A Molecular Pretraining Model with Geometric Triangle Awareness. Bioinformatics. https://doi.org/10.1093/bioinformatics/btae524

Vancouver

Hou X, Zhu T, Ren M, Duan B, Zhang C, Bu D, et al. GTAM: A Molecular Pretraining Model with Geometric Triangle Awareness. Bioinformatics. 2024. doi:10.1093/bioinformatics/btae524

BibTeX
@article{Hou2024,
  author = {Hou, Xiaoyang and Zhu, Tian and Ren, Milong and Duan, Bo and Zhang, Chunming and Bu, Dongbo and Sun, Shiwei},
  title = {GTAM: A Molecular Pretraining Model with Geometric Triangle Awareness},
  journal = {Bioinformatics},
  year = {2024},
  doi = {10.1093/bioinformatics/btae524}
}
Kostenlos starten →