Datasets and code for deployment-oriented evaluation of pathogen genome-based machine-learning models for invasive Salmonella risk prediction

Raw genome files are not re-hosted in this release because the corresponding public genome records are available from public repositories. Instead, the present dataset includes three main components: (1) cohort metadata tables defining the discovery, external validation, and related analysis cohorts; (2) Python scripts used to reproduce the main analytical workflows, including discovery analyses, external validation, leave-one-serovar-out transportability testing, deployment-oriented evaluation, and diagnostic analyses.The purpose of this dataset is not to duplicate all raw public sequence files, but to provide the study-level data objects and code required to trace sample provenance, reconstruct the main analytical workflow, and reproduce the core findings of the associated manuscript.

Derived outputs not separately deposited in this release can be regenerated from the metadata tables, provided intermediate feature files, analysis scripts, and the corresponding publicly available genome records. This dataset is intended to support interpretation, verification, and reasonable reproducibility of the study.