Machine learning reveals the dynamic importance of accessory sequences for Salmonella outbreak clustering

The dataset comprises a collection of Salmonella genomes associated with 58 historical outbreaks curated from the GenomeTrakr database and peer-reviewed PubMed articles. The outbreaks are linked to a diverse set of lineages that can be subdivided into 10 different serovars. The genomic data is accompanied by detailed contextual information describing the collection date, geographical origin and isolation source of the outbreak cases. The dataset was compiled to train classifiers to predict outbreak cluster labels from genomic data and infer a parsimonious set of outbreak predictive markers from the resulting classifiers.