In a significant breakthrough, researchers from MIT and McMaster University have harnessed the power of artificial intelligence (AI) to identify a potential new antibiotic capable of combating drug-resistant infections caused by Acinetobacter baumannii. This bacterium, commonly found in hospital settings, poses a serious threat to public health due to its ability to develop resistance to multiple antibiotics.
A machine-learning system was used by the researchers to sift through a large library of roughly 7,000 medicinal molecules. The goal was to identify a chemical that could efficiently target and kill A. baumannii, a prominent bacteria that causes a variety of drug-resistant illnesses such as pneumonia and meningitis. The AI algorithm was trained to analyse the chemical structures of these chemicals and identify their ability to prevent bacterial growth.
The results of this groundbreaking study, published in Nature Chemical Biology, hold significant promise in the field of antibiotic discovery. The researchers, led by senior authors Jonathan Stokes and James Collins, were able to identify a new antibiotic with remarkable efficacy against A. baumannii. This discovery brings hope in the battle against drug-resistant bacteria and provides a potential solution to combat the growing threat of infections caused by A. baumannii.
Drug-resistant microorganisms have emerged as a major worldwide health hazard. While antibiotic development has slowed, infections have developed and acquired resistance mechanisms, leaving existing therapies ineffective. This perilous scenario needs novel technologies, such as artificial intelligence (AI), to hasten the identification of novel antibiotics with distinct chemical structures.
Previously, the researchers proved the efficacy of machine learning in identifying chemicals that suppress the growth of E. coli, another antibiotic-resistant bacterium. Based on their previous results, the researchers turned their attention to A. baumannii, one of the most difficult bacterial threats in terms of antibiotic resistance.
To train the AI algorithm, the researchers exposed A. baumannii to thousands of different chemical compounds and analysed their impact on bacterial growth. By feeding the structure of each molecule into the model and indicating whether it inhibited growth, the algorithm learned to recognise chemical features associated with growth inhibition.
Once the model was trained, it analysed a separate set of 6,680 compounds that it had not encountered before. This analysis, completed in under two hours, yielded a selection of several hundred potential hits. From these, the researchers selected 240 compounds for experimental testing in the laboratory. They focused on compounds with distinct structures, different from existing antibiotics or molecules used in the training data.
The narrow spectrum of killing ability exhibited by the compound is highly desirable in an antibiotic, as it minimises the risk of rapid bacterial resistance development. Moreover, the compound’s specificity offers the potential to spare beneficial bacteria residing in the human gut, which play a crucial role in suppressing opportunistic infections.
The development of this novel antibiotic has enormous potential in the fight against drug-resistant A. baumannii infections. It emphasises the critical role that AI can play in hastening the hunt for viable medicines. Researchers can considerably broaden the scope of antibiotic discovery and develop targeted medicines to tackle the growing menace of drug-resistant infections by leveraging machine learning algorithms and enormous libraries of chemical compounds.