Painting Peptides with Antimicrobial Potency through Deep Reinforcement Learning

Abstract

In the post-antibiotic era, antimicrobial peptides (AMPs) serve as ideal drug candidates for their lower likelihood of inducing resistance. Computational models offer an efficient way to design novel AMPs. However, current optimization and generation approaches are tailored for different application scenarios. To address this challenge, we propose a novel AMP design model named AMPainter. Based on deep reinforcement learning, AMPainter integrates both optimization and generation tasks in a unified framework. We apply AM-Painter to three types of peptides, including known AMPs, signal peptides (SPs), and random sequences. AMPainter outperforms ten related models in enhancing the activity of known AMPs, and evolves effective AMPs from membrane-active SPs with a success rate of 80%. Furthermore, several de novo designed AMPs from random sequences are validated along with their evolutionary paths. Therefore, AMPainter contributes to paint antimicrobial potency to diverse peptides, assisting in expanding the AMP sequence space and discovering novel antimicrobial agents.