Machine learning finds a novel antibiotic able to kill superbugs in mice

The hunt for new antibiotics has gotten a modern upgrade. Searching for new antibacterial drugs has traditionally meant digging through soil for microbes that harbor bactericidal properties that could be isolated for drugs. In a new study, scientists report using machine learning to identify a new molecule that could one day prove effective against pathogens. The structurally distinct molecule, called halicin, was tested in mice and treated the gastrointestinal bug known as C. diff, which is responsible for more than a third of fatalities due to antibiotic resistance. The researchers also used machine learning to find eight other antibiotic compounds, which were also distinct from other drugs that are currently available.