For the First Time, an Artificial Intelligence has Discovered a New Kind of Antibiotic

The discovery of antibiotics revolutionised medicine. We often forget that many diseases easily treatable today were frequently deadly a mere century ago. However, the golden age of antibiotic discovery has come to an end.

There has been a 30 year void since the last new class of antibiotic was introduced. Meanwhile, misuse of antibiotics amongst humans and animals alike has accelerated the development of antibiotic resistance, making diseases such as tuberculosis harder to treat.

Now, a new compound that can kill a range of antibiotic-resistant bacteria has been discovered, using AI to analyse large amounts of data on existing drugs. Once the AI was ‘trained’ to recognise antibacterial molecules, researchers set their algorithm to look for compounds that did not resemble existing antibiotics, making it less likely that bacteria would have developed resistance mechanisms against them.

The drug, called halicin, was originally conceived for the treatment of diabetes. Halicin works by disrupting the flow of protons across the cell membrane, and was able to kill many bacteria in which multi-drug resistance is an increasing concern, including C. difficile and M. tuberculosis.

The discovery of this antibiotic would not have been possible without the use of AI, due to the sheer volume of data analysed. While the compound is effective at killing pathogens, it also kills the normal bacteria present in the healthy gut, and so one of the researchers’ next goals is to use AI to identify more selective antibiotics. AI could also be used assist in the synthesis of entirely new antibiotic compounds from scratch.