AI4AMP: an Antimicrobial Peptide Predictor Using Physicochemical Property-Based Encoding Method and Deep Learning.

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Antimicrobial peptides (AMPs) are crucial innate immune components with potential as antibiotic alternatives.
• The rise of antibiotic resistance necessitates the development of novel antimicrobial agents.
• Current wet-lab screening methods for discovering new AMPs are inefficient and costly.

Purpose of the Study:

• To develop a computational model for accelerating the discovery of novel antimicrobial peptides (AMPs).
• To create a user-friendly web server for predicting antimicrobial activity and screening proteomes.

Main Methods:

• Collected an up-to-date dataset of antimicrobial peptides and unbiased negative examples.
• Investigated protein-encoding methods and a deep learning model for AMP prediction.
• Trained and validated the deep learning model using external testing data.

Main Results:

• The trained deep learning model achieved 90% precision in predicting antimicrobial potential.
• The model demonstrated superior performance compared to existing methods.
• A web server, AI4AMP, was implemented for practical application of the prediction model.

Conclusions:

• The developed deep learning model and AI4AMP web server offer an efficient approach to identify novel AMP candidates.
• This tool can significantly accelerate the discovery of new antimicrobials to combat antibiotic resistance.
• AI4AMP provides a valuable resource for researchers and drug developers in the field of antimicrobial discovery.