Doing more with less: Genomic quasi-G-primes differentiate septic from healthy patients

Abstract

Sepsis is a life-threatening state of disseminated infection, and treatment requires knowledge of the organism responsible. The gold standard for sepsis diagnosis is blood culture, which requires days of growth. Next-generation sequencing has been proposed as an alternative; however, existing methods may lack sensitivity. In this work, we explore the idea of genomic quasi-G-primes, which are short DNA sequences specific to a single species within a group of relevant species. We first validated the genomic quasi-G-prime classification in controlled Staphylococcus aureus sequencing experiments, and then applied the same approach to blood-derived sequencing data from septic and healthy patients, where genomic quasi-G-prime profiles distinguished disease states. Our method is highly space-efficient, permitting fast classification on modest hardware and enabling it to outperform existing taxonomic classification approaches in this task.