AI employed to unveil abaucin, a potent drug combating A. baumannii, a perilous infection-causing bacterium.

In a recent publication in the journal Nature Chemical Biology, researchers from McMaster University and the Massachusetts Institute of Technology have harnessed the power of artificial intelligence to identify a novel antibiotic capable of eradicating a highly dangerous superbug. This particular superbug, known as Acinetobacter baumannii, has been designated by the World Health Organization as a “critical” menace among its “priority pathogens,” a classification reserved for bacterial families posing the most significant risks to human well-being.

As per the World Health Organization (WHO), A baumannii exhibits inherent adaptability, allowing it to develop resistance to treatment and transfer genetic material that facilitates the emergence of drug-resistant bacteria. This bacterium poses a significant risk to healthcare facilities, nursing homes, and patients who rely on ventilators, blood catheters, or have post-surgical wounds.

A baumannii has the ability to survive for extended periods on environmental surfaces and shared equipment, making it prone to transmission through contaminated hands. Besides causing bloodstream infections, it can also lead to infections in the urinary tract and lungs. The Centers for Disease Control and Prevention (CDC) states that A baumannii can even colonize a patient without causing symptoms or active infections.

In the study published on Thursday, scientists employed an AI algorithm to screen numerous antibacterial compounds, aiming to predict new structural classes. Through this AI-driven screening, they successfully identified a novel antibacterial compound named abaucin.

Gary Liu, a graduate student from McMaster University involved in the research, explained that their objective was to utilize the abundance of available data to determine which chemicals possessed antibacterial properties. His responsibility was to train the AI model, which would effectively predict the antibacterial potential of new molecules.

Through this approach, they significantly enhanced the efficiency of the drug discovery process, enabling them to focus specifically on the molecules with the highest significance. Liu further emphasized that this approach streamlined the drug discovery pipeline.

Once the AI model was trained, scientists employed it to analyze 6,680 compounds that were previously unknown to the model. The analysis, which took approximately an hour and a half, yielded several hundred compounds. Subsequently, 240 of these compounds underwent laboratory testing, ultimately identifying nine potential antibiotics, including abaucin.

Subsequently, the researchers conducted experiments to evaluate the efficacy of the newly discovered molecule against A. baumannii in a mouse model of wound infection. The results demonstrated that the molecule effectively suppressed the infection.

Jonathan Stokes, an assistant professor at McMaster University’s Department of Biomedicine and Biochemistry and one of the study’s leaders, expressed the significance of this work in validating the advantages of machine learning in the search for novel antibiotics. He emphasized that through the application of AI, scientists can swiftly explore vast chemical spaces, significantly enhancing the likelihood of identifying entirely new antibacterial compounds.

Stokes further explained that relying solely on broad-spectrum antibiotics is not optimal due to pathogens’ capability to evolve and adapt to various treatment strategies. He highlighted that AI methods present an opportunity to expedite the discovery of new antibiotics at a reduced cost. This avenue of research holds great importance in the pursuit of innovative antibiotic drugs.