RVFScan predicts virulence factor genes and hypervirulence of the clinical metagenome

Abstract

Abstract Bacterial pathogenicity often involves various virulence factors (VFs). Detecting virulence factor genes (VFGs) is critical for the precise treatment and prognostic management of bacterial infections. However, there is a lack of rapid and accurate methods for VFG identification from the metagenomes of clinical samples. We developed RVFScan (Read-based Virulence Factors’ Scanner), a novel user-friendly online tool that integrates a comprehensive VFG database with corresponding similarity matrix-based criteria for VFG prediction and annotation using metagenomic data without assembly. RVFScan outperformed previous assembly-based and read-based VFG predictors with 97% sensitivity, 98% specificity and 98% accuracy. To investigate the application of RVFScan, we performed the first large-scale analysis of 2425 clinical metagenomic datasets, obtaining species-specific VFG profiles and VF-phenotype associations for 24 important pathogens. By combining genomic comparisons and network analysis, we identified 53 VFGs with significantly higher abundances in hypervirulent Klebsiella pneumoniae (hvKp) than in classical K. pneumoniae (cKp). Furthermore, a cohort of 1256 samples suspected of Klebsiella pneumoniae infection demonstrated that RVFScan could accurately identify hvKp (90% sensitivity, 100% specificity, and 98.73% accuracy; Cohen’s kappa, 0.94; 90% of hvKp samples were consistent with clinical diagnosis). RVFScan could be applied to assembly free metagenomic reads to detect VFGs in low-biomass and high-complexity clinical samples, enabling the rapid identification and symptomatic treatment of hvKp infection and could be applied to other hypervirulent pathogens.