A deep learning algorithm has discovered a drug that kills off a broad spectrum of deadly bacteria in the lab, including antibiotic-resistant strains. Computer-simulated tests show promise in eight other drugs.
To develop the algorithm researchers gave it a molecule and asked it 'Is it antibacterial?' and let the program learn on its own which molecular features are important.
The algorithm used a database of more than 1,000 FDA-approved drugs and natural compounds with well-known antibiotic or non-antibiotic properties and was tasked with finding antibiotics that are as chemically distant as possible from known antibiotics in order to eliminate strains that may prove ineffective.
The drug called halicin was discovered which had originally been developed as an anti-diabetes drug and had no obvious similarities to existing antibiotics.
Initial topical testing of the drug found that it eradicated skin infections that would not be treatable with antibiotics.
More Info about this Invention: