Pre-trained molecular representations enable antimicrobial discovery.

Journal: Nature communications
PMID:

Abstract

The rise in antimicrobial resistance poses a worldwide threat, reducing the efficacy of common antibiotics. Determining the antimicrobial activity of new chemical compounds through experimental methods remains time-consuming and costly. While compound-centric deep learning models promise to accelerate this search and prioritization process, current strategies require large amounts of custom training data. Here, we introduce a lightweight computational strategy for antimicrobial discovery that builds on MolE (Molecular representation through redundancy reduced Embedding), a self-supervised deep learning framework that leverages unlabeled chemical structures to learn task-independent molecular representations. By combining MolE representation learning with available, experimentally validated compound-bacteria activity data, we design a general predictive model that enables assessing compounds with respect to their antimicrobial potential. Our model correctly identifies recent growth-inhibitory compounds that are structurally distinct from current antibiotics. Using this approach, we discover de novo, and experimentally confirm, three human-targeted drugs as growth inhibitors of Staphylococcus aureus. This framework offers a viable, cost-effective strategy to accelerate antibiotic discovery.

Authors

  • Roberto Olayo-Alarcon
    Department of Statistics, Ludwig-Maximilians-Universität München, Munich, Germany. roberto.olayo@lmu.de.
  • Martin K Amstalden
    Department of Microbiology, Julius-Maximilians-Universität Würzburg, Würzburg, Germany.
  • Annamaria Zannoni
    Department of Molecular Infection Biology II, Institute of Molecular Infection Biology (IMIB), Julius-Maximilians-Universität Würzburg, Würzburg, Germany.
  • Medina Bajramovic
    Department of Statistics, Ludwig-Maximilians-Universität München, Munich, Germany.
  • Cynthia M Sharma
    Department of Molecular Infection Biology II, Institute of Molecular Infection Biology (IMIB), Julius-Maximilians-Universität Würzburg, Würzburg, Germany.
  • Ana Rita Brochado
    Department of Microbiology, Julius-Maximilians-Universität Würzburg, Würzburg, Germany.
  • Mina Rezaei
    Department of Statistics, LMU Munich, Munich, Germany. mina.rezaei@stat.uni-muenchen.de.
  • Christian L Müller
    Helmholtz AI, Helmholtz Zentrum München, Neuherberg, Germany.