Key points from article :
Using deep learning, MIT researchers identified compounds that kill drug-resistant MRSA bacteria.
They developed methods to understand how the AI makes its predictions, aiding future drug design.
From 12 million compounds, the AI picked 5 classes likely effective against MRSA.
These antibiotics disrupt a key bacterial function, different from existing drugs.
Researchers are refining these candidates and using the AI to find antibiotics for other bacteria.
Published in Nature, the study demonstrated the compounds' ability to kill MRSA in lab dishes and two mouse models with low toxicity to human cells, making them promising drug candidates.
