Metagenomic strain-resolved DNA modification patterns link extrachromosomal genetic elements to host strains

DNA modification is central to microbial defense against extrachromosomal genetic elements (ECEs), consequently ECEs tend to adopt their host's modification patterns. Shared ECE-host modification patterns enable linking ECEs to their hosts, but modification detection tools are designed for single genomes and are ineffective at metagenome scale. Here, we present MODIFI, software for detecting DNA modifications in metagenomes. MODIFI assumes that each k-mer in a metagenome is mostly unmodified and calculates background signal levels for that k-mer from PacBio HiFi reads, eliminating the need for matched control experiments. MODIFI ECE-host linkages were validated using >1,000 isolate and mock microbiome datasets. Illustrating the approach, we identified 315 strain-resolved, non-redundant ECE-host linkages in environmental and human metagenomes. In infant gut microbiomes, a chromosomal inversion in Enterococcus faecalis alters host and associated plasmid methylation motifs simultaneously. Overall, MODIFI solves a major bottleneck in DNA modification analysis and provides a foundational tool for understanding microbial epigenomics.