Nanopore metagenomic sequencing links clinically relevant resistance determinants to pathogens

Metagenomic sequencing can detect pathogens and antimicrobial resistance genes directly from clinical samples without culture, but linking resistance genes to their bacterial hosts remains challenging. Here, we exploit DNA methylation patterns in nanopore sequencing data to associate plasmid-encoded resistance genes with their host bacteria in metagenomic samples. We developed a contig similarity score based on shared methylation motifs and validated this approach using mock metagenomic communities of clinically relevant carbapenem-resistant Enterobacterales, achieving 91% accuracy at the taxonomic species level. We then applied our framework to nanopore metagenomic data from patient rectal swabs collected during routine hospital screening. Comparison with established culture-based diagnostics and whole-genome sequencing confirmed that our approach correctly associated plasmid-as well as chromosomally encoded resistance genes--including all detected carbapenemases--with their pathogenic hosts while identifying additional clinically relevant resistance genes missed by routine testing. Our results demonstrate that nanopore metagenomics can provide actionable resistance-pathogen associations for clinical surveillance. Impact statementNanopore metagenomic sequencing data encode bacterial DNA methylation patterns that can accurately link antimicrobial resistance genes to their disease-causing hosts directly from patient samples, without the need for laboratory culture. This finding bridges an extant gap between metagenomics and rapid pathogen and resistance surveillance. Data summaryAll sequencing data after human content filtering have been deposited at the European Nucleotide Archive (ENA, BioProject accession PRJEB108076). All code is available at GitHub: https://github.com/harikaurel/Nanopore-AMR-Host-Association-Pipeline. All other supporting data are provided in the article and supplementary tables.