AI Helps Find an Antimicrobial Candidate Now Being Tested in the Lab
Published on · Joaquín Dati
Millions of proteins have no known function, and some could hold clues to new ways of fighting bacteria. Researchers used Google Antigravity, an AI-assisted workbench, to identify a candidate predicted to act against pathogens involved in respiratory infections. The molecule is now undergoing laboratory testing.
The search focused on antimicrobial peptides, small chains of amino acids that can act against microbes. These include bacteriocins, which bacteria produce to target competing bacteria. They offer a promising research direction as resistance to antimicrobial treatments rises.
Building a case from several clues
The team used ProtNLM, a model that predicts what a protein does by reading its amino acid sequence. They then compared the candidate with known proteins, examined its three-dimensional structure and investigated neighboring genes. Together, those clues supported the idea that it was a bacteriocin.
The researcher guided the questions and scientific strategy, while the AI handled tasks such as searching databases and bringing together findings from the literature. The outcome is a candidate for further investigation. Whether it performs as predicted remains a question for laboratory experiments.