In a groundbreaking development against drug-resistant Staphylococcus aureus (MRSA) bacteria, researchers have harnessed the power of more transparent deep learning models to uncover a novel class of antibiotics, marking the first significant advance in antibiotic discovery in six decades.

Artificial intelligence (AI) has emerged as a transformative force in medicine, playing a pivotal role in enabling scientists to identify the latest antibiotics. The newfound compound, capable of combating a drug-resistant bacterium responsible for thousands of global fatalities annually, stands as a potential turning point in the battle against antibiotic resistance.

Professor James Collins, from the Massachusetts Institute of Technology (MIT) and a co-author of the study published in Nature, explained, “The insight here was that we could see what was being learned by the models to make their predictions that certain molecules would make for good antibiotics.”

The study’s goal was to demystify the workings of deep-learning models, commonly regarded as “black boxes.” These models, employing artificial neural networks, automatically learn and represent features from data without explicit programming. The team focused on MRSA, a bacterium causing infections ranging from mild skin issues to life-threatening conditions like pneumonia and bloodstream infections.

To train their deep learning model, researchers expanded datasets and evaluated approximately 39,000 compounds for antibiotic activity against MRSA. The resulting data, along with details about the chemical structures of the compounds, were input into the model, aiming to shed light on its inner workings.

Felix Wong, one of the lead authors, stated, “What we set out to do in this study was to open the black box.” The team refined the selection of potential drugs using three additional deep-learning models, which assessed the toxicity of compounds on three types of human cells.

The integration of toxicity predictions with antimicrobial activity data enabled the identification of compounds effective against microbes while minimizing harm to the human body. The researchers screened around 12 million commercially available compounds, identifying promising antibiotic candidates from five different classes.

In laboratory tests against MRSA, two compounds from the same class emerged as particularly promising. In experiments involving mouse models for MRSA skin and systemic infections, these compounds significantly reduced the MRSA population, showcasing the potential of AI-driven antibiotic discovery in combating drug-resistant bacteria.

By Impact Lab