Deep Learning in Antimicrobial Peptide Prediction
Changhang Lin, Shuwen Xiong, Feifei Cui, Zilong Zhang, Hua Shi, Leyi Wei
Abstract
Antimicrobial peptides (AMPs) have garnered significant attention from researchers as effective alternatives to antibiotics. In recent years, deep learning has demonstrated unique advantages in AMP prediction, surpassing traditional machine learning methods and offering new avenues to address the issue of antibiotic resistance. This review introduces the research foundations of deep learning in AMP prediction, covering data set status, processing methods, and representation learning approaches. It particularly focuses on the application of basic models, language models, graph-related models, and other mixed and multimodal models for AMP prediction from the perspective of algorithmic models. Additionally, this review provides a comparative validation using classic deep learning models, offering guidance for subsequent research. Finally, it discusses the challenges and opportunities faced by deep learning algorithms in AMP prediction, particularly in terms of data balance, data augmentation, cyclic peptides, and interpretability, providing a comprehensive perspective and reference for further research in this field.