The team used machine learning to search for antimicrobial peptides with potential antibiotic properties in over 63,000 publicly available metagenomes and nearly 88,000 high-quality microbial genomes. The data was merged into the AMPSphere database, which allowed the scientists to mine a vast representation of microbial diversity on Earth. The team synthesized an antibiotic peptide from the database to tackle a dangerous bug causing skin lesions in mice, inhibiting bacterial growth with just one shot.
Key takeaways:
- A team of scientists used AI to analyze large databases of genetic material, uncovering nearly one million potential antibiotics.
- Out of 100 synthesized antibiotics, 63 were found to effectively fight off infections in a test tube, with one performing particularly well in a mouse model of skin disease.
- The researchers used machine learning to search for antimicrobial peptides with potential antibiotic properties in over 63,000 publicly available metagenomes and nearly 88,000 high-quality microbial genomes.
- The team's findings have been compiled into the AMPSphere database, which is open for public exploration and could potentially accelerate the discovery of new antibiotics.