In breakthrough, neural network ‘explains’ how it found a new antibiotic Premium

Scientists have used deep-learning to discover a new class of antibiotics, revolutionising drug development strategies, fundamentally by making their deep-learning model ‘explainable’.

What connects ChatGPT and antibiotics?

History has a curious answer: 1944. In that year, scientists proposed the first artificial neural network, a technology that later led to the birth of deep-learning and artificially intelligent systems like ChatGPT. The same year, biologists discovered streptomycin, the world’s first aminoglycoside antibiotic. It would soon revolutionise the treatment of life-threatening diseases like tuberculosis.

Today, we have a deeper connection between deep-learning and antibiotics. In a December 2023 paper in Nature, scientists have reported discovering a new class of antibiotics using a form of deep-learning that has been gaining more attention.

According to their paper, the last known structural class of antibiotics was reported in 2000. Their work has thus ended a decades-long wait for a new class.

The researchers were from the Broad Institute, Integrated Biosciences Inc., the Wyss Institute for Biologically Inspired Engineering, and the Leibniz Institute of Polymer Research.

Unlike previous approaches that used deep-learning to discover new drugs, the researchers said they were able to identify the chemical motifs – technically called substructures – their model used to check whether a given compound could be an antibiotic. This, they wrote, made their model “explainable”.

The team also demonstrated the efficacy of two compounds from this novel class of antibiotics in mice infected with methicillin-resistant Staphylococcus aureus (MRSA). MRSA infections were responsible for more than 100,000 human deaths in 2019.