How researchers used AI to find an antibiotic against a superbug

Superbugs are bacteria that are resistant to various types of antibiotics. Each year, these drug-resistant bacteria infect more than 2 million people in the US and kill at least 23,000, according to the US Centers for Disease Control and Prevention (CDC). .

What is Acinetobacter baumannii?

The study (‘Deep Learning Guided Discovery of a Targeted Antibiotic Acinetobacter baumannii‘) published in the journal Nature Chemical Biology on May 25 dealt with the bacterium Acinetobacter baumannii and included the participation of McMaster University in Canada and the Massachusetts Institute of Technology (MIT) in the US.

In 2017, the World Health Organization (WHO) identified the bacterium as one of the most dangerous antibiotic-resistant bacteria in the world. “Notoriously difficult to eradicate, A. baumannii it can cause pneumonia, meningitis, and infect wounds, all of which can lead to death,” according to McMaster University. “A. baumanni is generally found in hospital settings, where it can survive on surfaces for long periods of time,” she said.

The WHO superbug list highlighted bacteria that have built-in abilities to find new ways to resist treatment and can pass on genetic material that allows other bacteria to become resistant to drugs as well.

How do bacteria become resistant to drugs?

Antibiotics are drugs used to prevent and treat bacterial infections. Antibiotic resistance occurs when bacteria change in response to the use of these drugs, says the WHO. Ultimately, this threatens the ability of drugs to treat common infectious diseases.

“Where antibiotics for human or animal use can be purchased without a prescription, the emergence and spread of resistance worsens,” he says, warning against overconsumption of drugs without the recommendation of medical professionals for the treatment of common illnesses.

The WHO lists infections such as pneumonia, tuberculosis and foodborne illnesses as increasingly difficult to treat with existing drugs due to increasing antibacterial resistance.

How did the researchers use AI in this case?

Narrowing down the right antibacterial chemicals against bacteria can be a long and difficult process. This is where algorithms come in because the concept of AI is based on the process of machines receiving large amounts of data and training themselves to identify patterns and solutions based on them.

According to MIT, researchers first exposed A. baumannii grown in a laboratory dish to about 7,500 different chemical compounds, to see which ones might help stop the growth of the bacteria.

So they fed the structure of each molecule in the machine learning model. They also told the model whether or not each structure could prevent bacterial growth. This allowed the algorithm to learn chemical features associated with growth inhibition.

Once the model was trained, the researchers used it to analyze a set of 6,680 compounds. This analysis took less than two hours and returned a few hundred results. Of these, the researchers chose 240 to test experimentally in the lab, focusing on compounds with structures that were different from existing antibiotics.

Those tests turned up nine antibiotics, including one that was very potent and effective at killing A. baumannii. This has been called abaucin.

“Using AI, we can quickly explore vast regions of chemical space, significantly increasing the chances of discovering fundamentally new antibacterial molecules,” said Jonathan Stokes, the paper’s lead author and an assistant professor in McMaster’s Department of Biomedicine and Biochemistry.