Deep Learning in Antimicrobial Peptide Prediction.

Area of Science:

• Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Antimicrobial peptides (AMPs) are crucial alternatives to conventional antibiotics.
• Deep learning (DL) methods offer superior performance in AMP prediction compared to traditional machine learning.
• Antibiotic resistance necessitates novel therapeutic strategies, highlighting the importance of AMPs.

Purpose of the Study:

• To review the foundational aspects of deep learning in antimicrobial peptide (AMP) prediction.
• To analyze various algorithmic models, including basic, language, graph-related, and multimodal approaches for AMP prediction.
• To provide a comparative validation of classic deep learning models and discuss future research challenges and opportunities.

Main Methods:

• Review of existing literature on deep learning applications in AMP prediction.
• Analysis of data set status, processing, and representation learning techniques.
• Focus on algorithmic models: basic, language, graph-related, mixed, and multimodal.

Main Results:

• Deep learning models demonstrate significant advantages over traditional methods for AMP prediction.
• Various DL architectures, including language and graph-based models, are effective for AMP identification.
• Comparative validation of classic DL models provides insights into their performance.

Conclusions:

• Deep learning is a powerful tool for accelerating the discovery of novel antimicrobial peptides (AMPs).
• Addressing challenges like data imbalance, augmentation, cyclic peptides, and interpretability is key for advancing DL in AMP research.
• This review offers a comprehensive reference for future research in AI-driven antimicrobial peptide discovery.