As nearly 5 million people around the world lose their lives due to antibiotic resistance every year, finding new antibiotic solutions has become a top priority, and generative AI has made breakthrough progress in the field of antibiotic research and development.

According to a report from Stanford Medicine on March 28, researchers at Stanford Medicine and McMaster University have made important progress in this area. They have developed an AI model called SyntheMol that can help chemists synthesize drugs in the laboratory and Create a recipe.

The model creatively designed six new drugs against the drug-resistant bacterial strain Acinetobacter baumannii, one of the leading pathogens responsible for antibiotic resistance-related deaths. The model not only created new drug structures but also provided chemical Detailed recipes for synthesis enable chemists to synthesize these drugs in the laboratory.

According to Kyle Swanson, a PhD student in computational science at Stanford:

Nearly 10^60 drug-like molecules may exist in chemical space, and the 100 million compounds currently known are just the tip of the iceberg. The SyntheMol model successfully generated approximately 25,000 possible antibiotics and their synthetic formulations in less than 9 hours by utilizing more than 130,000 molecular building blocks and a series of validated chemical reactions.

What is unique about generative AI is its "illusion", which is the ability to create completely new molecules that have never appeared in nature. However, to ensure that these molecules could be synthesized in the laboratory, the researchers imposed necessary restrictions and guidance on SyntheMol's activity.

The researchers successfully synthesized 58 of 70 potential compounds through collaboration with the Ukrainian chemical company Enamine, and found in laboratory tests that six of them were effective in killing drug-resistant A. baumannii strains. These new compounds were also effective in killing drug-resistant A. baumannii strains. Other multidrug-resistant infectious bacteria have antibacterial activity.

Dr. James Zou, associate professor of biomedical data science and co-senior author of the study, emphasized the public health need for rapid development of new antibiotics. He said:

Although there are a large number of potentially effective drug molecules in nature that have yet to be discovered and tested, through the power of AI, researchers are now able to design entirely new molecules.