Antimicrobial resistance poses a major global health challenge, highlighting the urgent need for new strategies for antibiotic discovery. This project aims to develop a data-driven framework for antimicrobial discovery by integrating genome-wide phenotyping of ribosome dynamics with computational modelling and machine learning. Using the 5PSeq methodology developed in the host laboratory (Pelechano lab, MTC, Karolinska Institutet / SciLifeLab), large-scale datasets will be generated capturing ribosome-dynamics signatures in bacteria exposed to diverse antimicrobial compounds. Because 5PSeq sequences co-translational mRNA degradation intermediates, it provides a genome-wide in vivo snapshot of ribosome behaviour, enabling the identification of translational phenotypes associated with antimicrobial perturbations. The project is structured around three interconnected aims: (1) generation of a large-scale phenotypic dataset linking a chemically diverse library of approximately 5,000 small molecules with genome-wide ribosome dynamics in Staphylococcus aureus and additional bacterial pathogens, in collaboration with the SciLifeLab Chemical Biology Consortium (CBCS); (2) extraction of quantitative ribosome-dynamics features and development of machine learning classification models to predict antimicrobial activity and mechanism of action from translational signatures; and (3) integration of these phenotypic datasets with AI-based approaches to prioritise novel candidate antimicrobial compounds for experimental validation. The NAISS allocation requested here will support the computational infrastructure needed during the first year of the project, covering 5PSeq data processing, ribosome occupancy quantification, feature extraction, and initial machine learning classification analyses. This allocation is complementary to existing Pelechano lab NAISS resources, which support related but distinct projects on transcript isoform sequencing and post-transcriptional regulation. Generative and transformer-based modelling approaches are planned for a subsequent allocation once the full phenotypic dataset is available. All workflows will be developed following FAIR data principles, with code and processed datasets deposited in public repositories upon publication.