Fung-AI: An AI/ML-driven pipeline for antifungal peptide discovery

Area of Science:

• Biotechnology
• Computational Biology
• Drug Discovery

Background:

• Emerging fungal pathogens pose significant threats to global health and food security.
• There is a critical need for novel antifungal agents to combat rising resistance and vulnerability.
• Current antifungal discovery methods can be time-consuming and costly.

Purpose of the Study:

• To develop an AI/ML-driven pipeline, Fung-AI, for the rapid discovery of de novo antifungal peptides.
• To generate and prioritize novel peptide sequences with antifungal and low hemolytic activity.
• To validate the efficacy of AI-generated peptides against key fungal pathogens.

Main Methods:

• Utilized a generative adversarial network (GAN) to create novel peptide sequences.
• Developed in silico antifungal and hemolytic classifiers for peptide prioritization.
• Experimentally validated selected peptides against Fusarium graminearum and Candida albicans.
• Assessed cytotoxicity of promising peptides using HepG2 cells.

Main Results:

• The Fung-AI pipeline generated ~10,000 candidate peptides.
• Five peptides exhibited mild antifungal activity against Fusarium graminearum (MIC: 250–500 µg/mL).
• Four of these peptides were also active against Candida albicans (MIC: 500 µg/mL).
• Two peptides showed low cytotoxicity in HepG2 cells (LC50 > 704.2 µg/mL).
• No significant activity was observed against Candida auris.

Conclusions:

• The Fung-AI pipeline serves as a proof-of-principle for AI-driven antifungal peptide discovery.
• AI-generated peptides show potential as scaffolds for therapeutic development.
• Pathogen-specific optimization is necessary for effective antifungal peptide design.
• This approach accelerates the identification of novel antifungal candidates.