Abstract
As sequencing technologies become more affordable and genomic databases expand continuously, the reuse of publicly available sequencing data emerges as a powerful strategy for studying microbial pathogens. Indeed, raw sequencing reads generated for the study of a given organism often contain reads originating from the associated microbiota. This review explores how such off-target reads can be detected and used for the study of microbial pathogens. We present genomic data mining as a method to identify relevant sequencing runs from petabase-scale databases, highlighting recent methodological advances that allow efficient database querying. We then briefly outline methods designed to retrieve relevant data and associated metadata, and provide an overview of common downstream analysis pipelines. We discuss how such approaches have (i) expanded the known genetic diversity of microbial pathogens, (ii) enriched our understanding of their spatiotemporal distribution, and (iii) highlighted previously unrecognized ecological interactions involving microbial pathogens. However, these analyses often rely on the completeness and accuracy of accompanying metadata, which remain highly variable. We detail common pitfalls, including data contamination and metadata misannotations, and suggest strategies for result interpretation. Ultimately, while data mining cannot replace dedicated studies, it constitutes an essential and complementary tool for microbial pathogen research. Broader utility will depend on improved data standardization and systematic genomic monitoring across ecosystems.