Bacteriophages are the most abundant organisms on Earth, making it difficult to select which ones have the potential for therapeutic or other commercial applications. Current methods rely on non-iterative in vitro elaboration for product development, requiring >6 months per phage product.
In the current project, we aim to generate input data from experiments to build deep learning algorithms that will guide and support phage product development by
- revealing which phages compete best under which specific conditions and combinations,
- identifying mutations important for phage activity and survival under environmental conditions (pH, temperature), and
- demonstrating the population effects of phage infection in complex community structures.
Our group specialises in the AI-prediction of phage-host interactions in two-components systems (phage, bacteria, lysis/no lysis). In this project, we will use a top-down approach to characterise the resulting community composition and mutational spectra from in vitro assays and/or clinical samples to measure: 1. competition between multiple phages; 2. phage-interaction in complex communities to simulate microbiome and polymicrobial infection settings; and 3. exposition of phage to multiple environmental conditions using genomic re-sequencing and/or cycling temperature capillary electrophoresis (CTCE) to identify regions of mutations associated with specific stimuli.