AI-driven antibiotic discovery targets microbial dark matter

Nearly a century ago, the breakthrough discovery of antibiotics such as penicillin transformed the field of medicine by utilizing the natural antibacterial properties of microbes. A recent study conducted by researchers at the Perelman School of Medicine at the University of Pennsylvania proposes that the discovery of natural-product antibiotics is about to enter a new, accelerated phase, driven by artificial intelligence (AI).

Published in the journal Cell, the study outlines how the researchers employed a form of AI known as machine learning to identify potential antibiotics within an extensive dataset of the recorded genomes of tens of thousands of bacteria and other primitive organisms. This unprecedented analysis resulted in nearly one million potential antibiotic compounds, with dozens showing promising activity against pathogenic bacteria in initial tests.

"AI in antibiotic discovery has now become a reality and has significantly sped up our ability to identify new candidate drugs. What once took years can now be achieved in hours using computers," said César de la Fuente, PhD, a Presidential Assistant Professor involved in Psychiatry, Microbiology, Chemistry, Chemical and Biomolecular Engineering, and Bioengineering, and co-senior author of the study.

Nature has long served as a fertile ground for finding new medicines, particularly antibiotics. Bacteria, which are abundant on our planet, have developed numerous antibacterial defenses, often in the form of short proteins known as peptides that can disrupt bacterial cell membranes and other vital structures. While the initial discovery of penicillin and other natural-product antibiotics revolutionized medicine, the escalating threat of antibiotic resistance has underscored the urgent necessity for new antimicrobial compounds.

In recent years, de la Fuente and his team have been at the forefront of AI-driven searches for antimicrobials. They have identified preclinical candidates from the genomes of contemporary humans, extinct Neanderthals and Denisovans, woolly mammoths, and hundreds of other organisms. One of the primary aims of the lab is to mine the world's biological data for useful molecules, including antibiotics.

For this latest study, the researchers utilized a machine learning platform to examine multiple public databases containing microbial genomic data. The analysis covered 87,920 genomes from specific microbes as well as 63,410 mixes of microbial genomes—known as "metagenomes"—from environmental samples. This comprehensive search spanned a variety of habitats across the globe.

This extensive exploration succeeded in identifying 863,498 candidate antimicrobial peptides, over 90 percent of which were previously undocumented. To validate these findings, the researchers synthesized 100 of these peptides and tested them against 11 pathogenic bacterial strains, including antibiotic-resistant strains of Escherichia coli and Staphylococcus aureus.

"Our initial screening revealed that 63 of these 100 candidates completely halted the growth of at least one of the pathogens tested, and often multiple strains," de la Fuente said. "In some instances, these molecules were effective against bacteria at very low doses."

Encouraging results were also observed in preclinical animal models, where some of the potent compounds successfully prevented infections. Further analysis indicated that many of these candidate molecules destroy bacteria by disrupting their outer protective membranes, effectively causing them to burst.

The identified compounds originated from microbes inhabiting a wide variety of environments, including human saliva, pig intestines, soil, plants, corals, and numerous other terrestrial and marine organisms. This validates the researchers' broad approach to exploring global biological data.

Overall, the findings highlight the power of AI in discovering new antibiotics, providing multiple new leads for antibiotic developers, and heralding the beginning of a promising new era in antibiotic discovery. The research team has made their repository of putative antimicrobial sequences, termed AMPSphere, open access and freely available for further exploration.