pCPPs-sADNN: predicting cell-penetrating peptides using self-attention based deep neural network.

Area of Science:

• Biotechnology
• Computational Biology
• Drug Delivery

Background:

• Cell-penetrating peptides (CPPs) are crucial for drug delivery and intracellular localization.
• Traditional laboratory methods for CPP identification are time-consuming and costly.
• Computational approaches offer a faster and more economical alternative.

Purpose of the Study:

• To develop an accurate and efficient computational model for predicting cell-penetrating peptides (CPPs).
• To leverage advanced machine learning techniques for enhanced CPP prediction.
• To overcome limitations of existing experimental and computational methods.

Main Methods:

• Developed the pCPPs-sADNN model integrating feature embeddings from Protein Text-to-Text Transfer Transformer and Evolutionary Scale Modeling, alongside Conjoint Triad Features.
• Employed Random Forest-based Recursive Feature Elimination for feature selection.
• Utilized Adaptive Synthetic Sampling Approach for class imbalance and trained a deep neural network with an attention mechanism.

Main Results:

• The pCPPs-sADNN model achieved a high training accuracy of 98.58% and an AUC of 0.99.
• On the test dataset, the model demonstrated strong performance with an accuracy of 96.84% and an AUC of 0.99.
• Feature fusion and deep learning with attention significantly improved prediction accuracy.

Conclusions:

• The pCPPs-sADNN model provides a rapid, cost-effective, and accurate computational solution for CPP prediction.
• This approach enhances drug delivery and intracellular localization strategies.
• The study highlights the potential of integrating diverse feature sets and deep learning for biological predictions.