Researchers have discovered a potential new antibiotic against a difficult species of disease-causing bacteria, Acinetobacter baumannii, using machine learning. The method can speed up the identification and testing of new antibiotics, which is crucial amid the rise of antimicrobial resistance. Machine-learning models were trained to look for molecules with properties suitable for fighting specific bacteria. The system was applied to a database of 6,680 molecules, with a shortlist of 240 produced in just a few hours, with nine compounds inhibiting bacterial growth by 80% or more. The compound ultimately chosen, abaucin, is species-selective, providing hope for fighting antibiotic-resistant bacterial infections. The global cost of antimicrobial resistance could be more than $1tn this year.
Abaucin compromises the normal function of a protein called CCR2. It also disrupts lipoprotein trafficking in Acinetobacter baumannii. It appears to selectively target Acinetobacter baumannii, and only disrupts the growth of this Gram-negative bacteria, not others. The team plans to enhance the model by focusing on gathering more robust training data and designing and testing compounds chemically similar to abaucin, to produce more effective antibiotics.
The research was conducted by a team of scientists led by assistant professor of biochemistry, Jon Stokes of McMaster University, Ontario.
McMaster University was founded in 1887 and is located in Hamilton, Ontario, Canada. It is one of Canada’s top research-intensive universities. Its faculties include the Arts and Science, Business, Engineering, Health Sciences, and Humanities.
Jon Stokes is an assistant professor of biochemistry at McMaster University, Ontario. Stokes’ main area of interest revolves around understanding the function and behavior of neurotransmitter receptors, which are important for cellular signalling in the brain. Stokes has been similarly involved in investigating the mechanisms behind bacteria and how antimicrobial resistance begins.
