From AI to action: Antimicrobial peptides engineered by generative adversarial networks (GANs)-A novel approach to combat resistant Bacteria
Journal
Chemical Engineering Journal
Journal Volume
519
Start Page
164905
ISSN
1385-8947
Date Issued
2025-09-01
Author(s)
Lai, Chia-Wen
Lin, Chung-Yen
Tsai, Meng-Chen
Chen, Wei-Jung
Hsieh, Chia-Chun
Lin, Zi-Jing
Lai, Lee-Jene
Chen, Shu-Hwa
Abstract
With the growing threat of antibiotic-resistant bacteria, the need for innovative antibacterial treatments is more urgent than ever. Antimicrobial peptides (AMPs) are gaining attention due to their unique mechanisms of action. We utilized a Wasserstein generative adversarial network (WGAN) with a gradient penalty to expedite the discovery of new AMP candidates. Among the promising peptides identified, GAN-pep3 stood out with its impressive antibacterial activity against Gram-negative and Gram-positive bacteria. It achieved a low geometric mean minimum inhibitory concentration (MIC) of 0.90 μM and 4.21 μM, respectively. Although GAN-pep3 showed slight hemolytic activity, its low MIC provided a high therapeutic index, particularly against Gram-negative bacteria. This study highlights the potential of AI-designed AMPs as powerful antimicrobial agents with broad-spectrum activity. It instills hope for significant therapeutic applications in the fight against antibiotic-resistant bacteria.
Subjects
AI-designed
Antimicrobial peptides
Generative adversarial network
Resistant bacteria
Publisher
Elsevier BV
Type
journal article