Genome-based model for differentiating between infection and carriage Staphylococcus aureus

Abstract

ABSTRACT Staphylococcus aureus ( S. aureus ) is a clinically significant opportunistic pathogen, which can colonize multiple body sites in healthy individuals and cause various life-threatening diseases in both children and adults worldwide. The genetic backgrounds of S. aureus that cause infection versus asymptomatic carriage vary widely, but the potential genetic elements (k-mers) associated with S. aureus infection remain unknown, which leads to difficulties in differentiating infection isolates from harmless colonizers. Here, we address the disease-associated k-mers by using a comprehensive genome-wide association study (GWAS) to compare the genetic variation of S. aureus isolates from clinical infection sites (272 isolates) with nasal carriage (240 isolates). This study uncovers consensus evidence that certain k-mers are overrepresented in infection isolates compared with carriage isolates, indicating the presence of specific genetic elements associated with S. aureus infection. Moreover, the random forest (RF) model achieved a classification accuracy of 77% for predicting disease status (infection vs carriage), with 68% accuracy for a single highest-ranked k-mer, providing a simple target for identifying high-risk genotypes. Our findings suggest that the disease-causing S. aureus is a pathogenic subpopulation harboring unique genomic variation that promotes invasion and infection, providing novel targets for clinical interventions. IMPORTANCE Defining the disease-causing isolates is the first step toward disease control. However, the disease-associated genetic elements of Staphylococcus aureus remain unknown, which leads to difficulties in differentiating infection isolates from harmless carriage isolates. Our comprehensive genome-wide association study (GWAS) found consensus evidence that certain genetic elements are overrepresented among infection isolates than carriage isolates, suggesting that the enrichment of disease-associated elements may promote infection. Notably, a single k-mer predictor achieved a high classification accuracy, which forms the basis for early diagnostics and interventions.