Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
June 23, 2026Journal of CheminformaticsOpen Access

DCPM-ADMET: fusion of dual-component pre-trained model and molecular fingerprints to enhance drug ADMET properties prediction

View Full Paper
Ask AI
Bookmark
Share

Authors

LZLeilei ZhangYZYuchen ZengYQYue Qi

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates improved ADMET property prediction in drug development, suggesting enhanced safety assessments.

Key Points

  • The main aim is to improve the accuracy of ADMET property predictions through an advanced pre-trained model.
  • Developed DCPM-ADMET using a dual-component architecture combining an XLNet-based module and a GRU component.
  • Incorporated ECFP fingerprints for local substructure representation and fine-tuned using a proprietary database of 465,470 entries.
  • Created a free online prediction tool featuring 133 endpoints.
  • DCPM-ADMET outperformed traditional methods and previous models in prediction accuracy on benchmark datasets.
  • The model features the highest number of ADMET endpoints (97) reported to date, with multiple regression tasks.
  • The online ADMET prediction tool has been made publicly available to aid in drug discovery.

Cite This Study

Zhang et al. (2026) studied this question.

synapsesocial.com/papers/6a3a2194111626ef22ab659bhttps://doi.org/10.1186/s13321-026-01244-z
View Full Paper
Ask AI
Bookmark
Share