Discovery of antimicrobial peptides in the global microbiome with machine learning - Graphical abstract & Highlights

Aug 5, 2026

Upload by Maeve

Machine-learning-based modeling is used to discover antimicrobial peptides (AMPs) from the global microbiome and to build AMPSphere, a comprehensive open-access catalog. The resource is generated from 63,410 metagenomes and 87,920 prokaryotic genomes, yielding 863,498 nonredundant peptides with limited matches to existing databases. Predictions are experimentally validated by synthesizing and testing 100 AMPs against drug-resistant pathogens and gut commensals, with 79 active peptides and 63 targeting pathogens. Active AMPs disrupt bacterial membranes in vitro and in vivo.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages35

Reading Time1 h 28 min

AccessFree to download

Highlights

Machine learning predictions for AMPs

AMPSphere open-access resource

Experimental validation and activity results

Summary

Resource Discovery of antimicrobial peptides in the global microbiome with machine learning Graphical abstract Highlights Authors C´elio Dias Santos-J´unior, Marcelo D.T. Torres, Yiqian Duan, ..., Jaime Huerta-Cepas, Cesar de la Fuente-Nunez, Luis Pedro Coelho Correspondence [cfuente@upenn.edu](cfuente@upenn.edu) (C.d.l.F.-N.), [luispedro@big-data-biology.org](luispedro@big-data-biology.org) (L.P.C.) In brief A machine-learning-based approach predicts nearly one million new antibiotics from the global microbiome, with 79 out of 100 tested peptides being active in vitro and several showing efﬁcacy comparable to a clinical antibiotic in a mouse preclinical model of infection. d Machine learning predicts nearly 1 million new antibiotics in the global microbiome d Out of 100 tested peptides, 79 were active in vitro; 63 of these targeted pathogens d Some peptides may originate from longer sequences through genomic fragmentation d The AMPSphere is an open-access resource to accelerate antibiotic discovery Santos-J´unior et al., 2024, Cell 187, 3761–3778 July 11, 2024 ª 2024 The Authors. Published by Elsevier Inc. [https://doi.org/10.1016/j.cell.2024.05.013](https://doi.org/10.1016/j.cell.2024.05.013) ll ll OPEN ACCESS Resource Discovery of antimicrobial peptides in the global microbiome with machine learning C´elio Dias Santos-J´unior,1,2,16 Marcelo D.T. Torres,3,4,5,6,16 Yiqian Duan,1 ´Alvaro Rodr´ıguez del R´ıo,7 Thomas S.B. Schmidt,8,9 Hui Chong,1 Anthony Fullam,8 Michael Kuhn,8 Chengkai Zhu,1 Amy Houseman,1 Jelena Somborski,1 Anna Vines,1 Xing-Ming Zhao,1,12,13,14 Peer Bork,8,10,11 Jaime Huerta-Cepas,7 Cesar de la Fuente-Nunez,3,4,5,6,* and Luis Pedro Coelho1,15,17,* 1Institute of Science and Technology for Brain-Inspired Intelligence-ISTBI, Fudan University, Shanghai 200433, China 2Laboratory of Microbial Processes & Biodiversity-LMPB, Department of Hydrobiology, Universidade Federal de S˜ao Carlos – UFSCar, S˜ao Carlos, S˜ao Paulo 13565-905, Brazil 3Machine Biology Group, Departments of Psychiatry and Microbiology, Institute for Biomedical Informatics, Institute for Translational Medicine and Therapeutics, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA 4Departments of Bioengineering and Chemical and Biomolecular Engineering, School of Engineering and Applied Science, University of Pennsylvania, Philadelphia, PA, USA 5Department of Chemistry, School of Arts and Sciences, University of Pennsylvania, Philadelphia, PA, USA 6Penn Institute for Computational Science, University of Pennsylvania, Philadelphia, PA, USA 7Centro de Biotecnologay Gen´omica de Plantas, Universidad Polit´ecnica de Madrid (UPM) -Instituto Nacional de Investigaci´ony Tecnologa Agraria y Alimentaria (INIA-CSIC), Campus de Montegancedo-UPM, Pozuelo de Alarc´on, 28223 Madrid, Spain 8Structural and Computational Biology Unit, European Molecular Biology Laboratory, Heidelberg, Germany 9APC Microbiome & School of Medicine, University College Cork, Cork, Ireland 10Max Delbr¨uck Centre for Molecular Medicine, Berlin, Germany 11Department of Bioinformatics, Biocenter, University of W¨urzburg, W¨urzburg, Germany 12Department of Neurology, Zhongshan Hospital, Fudan University, Shanghai, China 13State Key Laboratory of Medical Neurobiology, Institutes of Brain Science, Fudan University, Shanghai, China 14MOE Key Laboratory of Computational Neuroscience and Brain-Inspired Intelligence and MOE Frontiers Center for Brain Science, Fudan University, Shanghai, China 15Centre for Microbiome Research, School of Biomedical Sciences, Queensland University of Technology, Translational Research Institute, Woolloongabba, QLD, Australia 16These authors contributed equally 17Lead contact *Correspondence: [cfuente@upenn.edu](cfuente@upenn.edu) (C.d.l.F.-N.), [luispedro@big-data-biology.org](luispedro@big-data-biology.org) (L.P.C.) [https://doi.org/10.1016/j.cell.2024.0

FAQs

What does the AMPSphere resource provide?

AMPSphere is a comprehensive, open-access catalog of antimicrobial peptides predicted from global microbiome data, including evolutionary insights into peptide origins and habitat-dependent AMP production.

How were the machine-learning predictions validated?

Researchers synthesized 100 predicted AMPs and tested them against clinically relevant drug-resistant pathogens and human gut commensals in vitro and in vivo.

What were the key experimental outcomes?

Out of 100 tested peptides, 79 were active in vitro, and 63 targeted pathogens; the active AMPs showed antibacterial effects by disrupting bacterial membranes.

Aug 5, 2026

Upload by Maeve