DEEP LEARNING IN THE DISCOVERY OF NOVEL ANTIMICROBIAL PEPTIDES (AMPS)
Keywords:
Deep Learning, Antimicrobial Peptides (AMPs), Convolutional Neural Networks (CNNs), In Silico Screening, Genomic Mining, Protein Language Models, De-extinction Pharmacology, Bioinformatic Pipelines.Abstract
The rise of multi-drug resistant "superbugs" has necessitated a shift away from traditional small-molecule antibiotics toward Antimicrobial Peptides (AMPs)—nature’s ancient defense system. However, the chemical space for these peptides is vast, with trillions of potential combinations. This review evaluates the 2024–2026 revolution in "Digital Mining," where Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) have moved from experimental curiosities to essential tools. By scanning massive genomic databases and "resurrecting" molecules from extinct species, these models are predicting potent antimicrobial candidates with high accuracy before a single drop of liquid is touched in a lab.
References
1. Bai, W., et al. (2026). "AMPBAN: A Deep Learning Framework Integrating Protein Sequence and Structural Features." bioRxiv.
2. Ma, Y., et al. (2025). "Exploring the Repository of De Novo-Designed Bifunctional AMPs through Deep Learning." eLife.
3. Szczurek, E. (2024). "DOG-AMP: Deep Optimised Generation of Antimicrobial Peptides." Helmholtz Munich AI for Health.
4. Yan, J., et al. (2024). "Mining the Extinct Proteome: APEX and the Discovery of Mammuthusin." Nature Biomedical Engineering.
5. Williams, G. & Richard, C. (2025). "Biologically Relevant Chemical Space: Mapping the Future of Chemoinformatics." Frontiers in Drug Discovery.
