AI-Guided Optimization of an Antimicrobial Peptide targeting Bacillus cereus: Enhancing Activity and Reducing Toxicity Through Machine Learning Tools

Abstract

The growing threat of antibiotic resistance has underscored the urgent need for alternative therapeutic strategies. Antimicrobial peptides (AMPs), naturally occurring molecules with broad-spectrum activity, present a promising solution but face limitations such as toxicity, instability, and delivery challenges. This study aimed to enhance the safety and efficacy of a low-performing AMP active against Bacillus cereus by utilizing artificial intelligence (AI) and machine learning (ML) tools. Four computational platforms - CAMPr4, ToxinPred3, AlphaFold, and ChatGPT - were used to predict antimicrobial potential, evaluate toxicity, model structure, and streamline peptide design. Starting with an AMP scoring 0.2895 in CAMPr4, sequence modifications were made to increase net charge and decrease toxicity. The final optimized peptide, named Cereus-Black93, achieved a predicted antimicrobial score of 0.93 and a significantly reduced toxicity score of 0.27. Structural modeling confirmed a stable alpha-helical conformation. These results demonstrate the potential of AI-driven approaches to accelerate the design of novel AMPs and pave the way for future in vitro validation and therapeutic development.