Antibiotic resistance is a growing global concern as it poses a significant threat to public health. Antibiotics are essential in treating bacterial infections, but their misuse and overuse have led to the development of antibiotic-resistant bacteria. This can render many infections untreatable, leading to severe consequences for patients.
One of the most dangerous antibiotic-resistant bacteria identified by the World Health Organization is Acinetobacter baumannii. This pathogen is responsible for causing various infections, including pneumonia, meningitis, and wound infections, particularly in hospital settings where vulnerable patients are at risk.
A collaborative effort between McMaster University and the Massachusetts Institute of Technology, led by Dr. Jonathan Stokes, an assistant professor of biochemistry at McMaster, sought to combat A. baumannii infections using artificial intelligence (AI). They aimed to identify a novel antibacterial compound that could effectively treat these infections. Their groundbreaking discovery, a new compound named “abaucin,” was published in the journal Nature Chemical Biology.
Conventional methods for finding antibiotics to target A. baumannii have been challenging, time-consuming, and labor-intensive. Broad-spectrum antibiotics, which target a wide range of bacteria, are suboptimal as pathogens can quickly develop resistance to them. Moreover, most new antibiotics are mere analogs of existing classes, leading to limited long-term efficacy due to pre-existing resistance determinants.
The development of new antibiotics must focus on compounds with unique mechanisms of action. Such narrow-spectrum antibacterial compounds are likely to have prolonged utility due to low pre-existing clinical resistance.
AI, specifically machine learning methods, allows researchers to explore vast chemical compounds rapidly, thereby increasing the chances of finding promising antibacterial agents. Unlike traditional screening methods, which can only test a few million compounds for antibacterial activity, algorithmic approaches can screen through hundreds of millions to billions of molecules.
In their study, Stokes’s team screened around 7,500 molecules to identify those with antibacterial activity against A. baumannii. They employed a neural network trained on an inhibition dataset and utilized the Drug Repurposing Hub to predict novel compounds with inhibitory activity against A. baumannii.
This approach successfully led them to discover abaucin, a narrow-spectrum antibacterial compound. The narrow spectrum of activity of abaucin is advantageous as it can overcome intrinsic acquired resistance mechanisms. The researchers also demonstrated the effectiveness of abaucin in controlling an A. baumannii infection in a mouse wound model.
The increasing availability of high-quality datasets for training AI algorithms presents an opportunity for machine learning to revolutionize the discovery of novel antibacterial compounds. AI in medicine holds tremendous promise for the future, allowing us to combat antibiotic resistance effectively and safeguard public health.
As research in this field advances, we can look forward to a brighter future with more efficient and effective solutions for antibiotic-resistant infections.