Related Experiment Video
Updated: Aug 9, 2026
Construction of Cyclic Cell-Penetrating Peptides for Enhanced Penetration of Biological Barriers
Published on: September 19, 2022
PepOSX-AI: CPP - an interpretable transformer-based deep learning model for prediction of cell-penetrating peptides
Haowen Chen1, Weiwei He2, Kaiyan Feng3
1School of Food Science and Engineering, South China University of Technology, Guangzhou, 510640, China; Guangdong Provincial Key Laboratory of Large Animal Models for Biomedicine, School of Pharmacy and Food Engineering, Wuyi University, Jiangmen, 529020, China.
A new deep-learning model, PepOSX-AI: CPP, accelerates the discovery of cell-penetrating peptides (CPPs) for drug delivery. This AI tool enhances transmembrane delivery efficiency and bioavailability, overcoming limitations of traditional screening methods.
Area of Science:
- Biotechnology
- Computational Biology
- Drug Delivery Systems
Background:
- Cell-penetrating peptides (CPPs) facilitate the delivery of molecules across cell membranes, showing promise for enhancing drug bioavailability.
- Conventional methods for identifying CPPs are inefficient due to reliance on extensive cell and animal experiments.
Purpose of the Study:
- To develop an efficient computational tool for predicting and identifying cell-penetrating peptides (CPPs).
- To overcome the limitations of traditional, time-consuming CPP screening methods.
Main Methods:
- A deep-learning model, PepOSX-AI: CPP, was developed using the Transformer architecture and six physicochemical descriptors.
- Model parameters were optimized through automated hyperparameter tuning.
- A self-attention mechanism was employed to interpret sequence dependencies and identify key features for CPP activity.
Main Results:
- The PepOSX-AI: CPP model achieved a high Area Under the Curve (AUC) of 0.914.
- The model demonstrated 91.00% accuracy on an independent test set, outperforming existing prediction models.
- Attention patterns provided insights into sequence features critical for CPP function.
Conclusions:
- PepOSX-AI: CPP offers a novel, accurate, and interpretable computational approach for CPP screening.
- The developed online platform facilitates the practical application of this AI tool for CPP discovery.
- This work provides a valuable computational resource and theoretical foundation for advancing CPP-based drug delivery research.
