AI4S - AI-based feature learning from 5PSeq for AMR phenotyping
Abstract
Antimicrobial resistance (AMR) is a major global health challenge, creating a need for faster methods to characterise antibiotic susceptibility and mechanisms of drug action. We have developed 5PSeq-based approaches that measure co-translational RNA degradation and provide genome-wide readouts of ribosome dynamics. However, current analytical approaches rely largely on predefined features and capture only part of the information contained in these high-dimensional data.
In collaboration with AI4S, we will develop AI-based representation learning approaches that learn informative molecular features directly from genome-wide 5PSeq profiles. We will investigate models integrating positional, sequence and coverage information and evaluate their ability to identify antibiotic mode of action and resistance phenotypes across bacterial species and experimental conditions.
The goal is to establish a scalable AI framework for molecular phenotyping that can support improved AMR characterisation and data-driven antibiotic discovery.
