Enhanced prediction of hemolytic activity in antimicrobial peptides using deep learning-based sequence analysis.

Area of Science:

• Biochemistry
• Computational Biology
• Pharmacology

Background:

• Antimicrobial peptides (AMPs) show broad-spectrum antimicrobial activity, offering potential for new therapeutics.
• Clinical use of AMPs is hindered by their hemolytic activity, causing red blood cell destruction.
• Developing AMPs with reduced hemolysis is critical for safe and effective therapeutic applications.

Purpose of the Study:

• To develop a deep learning model for predicting the hemolytic activity of antimicrobial peptides.
• To utilize convolutional neural networks (CNNs) for accurate hemolysis prediction in AMPs.
• To facilitate the design of safer AMPs for clinical use.

Main Methods:

• Peptide sequences were encoded using one-hot encoding.
• A convolutional neural network (CNN) architecture with convolutional and fully connected layers was employed.
• The model was trained and validated on six diverse datasets (HemoPI-1, HemoPI-2, HemoPI-3, RNN-Hem, Hlppredfuse, AMP-Combined).

Main Results:

• The CNN model achieved high predictive performance across multiple datasets.
• Matthew's correlation coefficients ranged from 0.5614 to 0.9274, demonstrating robust accuracy.
• The developed model outperformed existing methods for predicting AMP hemolytic activity.

Conclusions:

• The deep learning model effectively predicts AMP hemolytic activity.
• This predictive capability aids in designing AMPs with minimized toxicity.
• The research supports the advancement of AMPs as safer therapeutic agents against bacterial infections.