SUMMARY Sepsis is a leading cause of death in hospitalized patients and involves a dysregulated host inflammatory response to infection. Despite decades of clinical trials, no new, effective treatments have been identified, and mortality rates remain unacceptably high. Several factors have contributed to this impasse, starting with the complex challenge of accurately detecting the microbial pathogens precipitating sepsis. Heterogeneity across patient populations is a second key barrier to the development of effective treatment strategies, given that sepsis can be caused by a broad diversity of microbes, and immune responses vary widely between patients. In addition, while disease trajectories and interventional needs differ dramatically between individuals, no prognostic tools exist that account for both the host response and pathogen, the two primary drivers of sepsis pathobiology. Here we seek to address key gaps in our understanding and treatment of sepsis by leveraging blood and plasma specimens collected from 1563 patients enrolled across 60 medical centers in the CLOVERS trial, the largest study of septic shock to date. Aim 1 seeks to identify known and novel pathogens responsible for septic shock using metagenomic sequencing of host and microbe, a novel culture-independent method for studying sepsis. Aim 2 will combine host gene expression and microbial metagenomic data to discover novel sepsis subphenotypes and evaluate their associations with clinical outcomes. Aim 3 will use machine learning to develop a multiomic classifier that predicts sepsis mortality by incorporating host transcriptomic, microbial metagenomic, and clinical data. Together, these aims will accelerate our understanding of sepsis pathogenesis, and could redefine our diagnostic, therapeutic, and prognostic strategies, ultimately improving patient outcomes.