Studying host-pathogen interaction via microscopy and Deep Learning: application to antimicrobial resistant bacteria and monoclonal antibodies discovery

This research activity aims to develop deep-learning techniques for analyzing biological images of the Opsonophagocytosis assay. This critical immune process involves antibodies marking pathogens for destruction by phagocytes. 

In our opsonophagocytosis assay protocol, we acquire images of cells infected with the pathogen Shigella and test monoclonal antibodies at multiple concentrations to determine when they begin to promote phagocytosis of the bacteria. We trained a convolutional neural network (CNN) to recognize positive and negative control mAbs that exhibit strong phagocytic effects or none at all. We then used the network to screen for new antibodies, leading to the identification of promising mAbs to treat shigellosis.

By integrating deep learning with opsonophagocytosis assays, we aim to streamline the mAb screening process and contribute to the development of effective treatments against AMR-related infections. Moreover, we hope to extend this methodology to screen for mAbs against multiple bacterial pathogens and apply it to other assays.