A team including researchers at Fudan University mined DNA sequenced directly from soil, water, and the human gut, over 63,000 metagenomes in all, training a machine learning model to spot small proteins that looked like they could kill bacteria.
The search turned up 863,498 candidate antimicrobial peptides, a catalog the team named AMPSphere, almost all of them never described before. When the researchers synthesized 100 of the predicted candidates and tested them directly, 79 killed target bacteria in the lab.
Most of this catalog exists only as sequence so far; the 79 validated hits, some also tested in mice, are an early slice of what the full set might contain. What the study demonstrated was scale: a resource this size turns antibiotic discovery from hunting one compound at a time into searching a library nearly a million candidates deep.
Read the original study
Santos-Júnior et al., Cell, 2024 · doi.org/10.1016/j.cell.2024.05.013
Almost a million candidates nobody had looked for.
Source
- Santos-Júnior, C. D., Torres, M. D. T., Duan, Y., Rodríguez Del Río, Á., Schmidt, T. S. B., Chong, H., Fullam, A., Kuhn, M., Zhu, C., Houseman, A., Somborski, J., Vines, A., Zhao, X.-M., Bork, P., Huerta-Cepas, J., de la Fuente-Nunez, C., & Coelho, L. P. (2024). Discovery of antimicrobial peptides in the global microbiome with machine learning. Cell, 187(14), 3761–3778.e16. https://doi.org/10.1016/j.cell.2024.05.013