Mining of food metagenomes reveals an unexplored diversity of dsDNA bacteriophages

Bacteriophages are key drivers of microbial ecology, co-existing and co-evolving with bacteria across diverse environments. Limitations in culturing, alongside advances in sequencing and bioinformatics, have driven the use of metagenomics to explore viral diversity. Viral-specific analysis of >3000 food metagenomes from cFMD produced the FVGC, comprising ~3400 metagenome-assembled viruses, most of which belong to novel Caudoviricetes lineages (n = 91), with only ~15% represented in IMG/VR v4. Together, these findings reveal extensive uncharacterized viral diversity in food systems. Beyond serving as a reference, the FVGC facilitates detailed investigation of virus–host interactions. Viral sequences were pervasive across microbial genomes, with several bacterial families exhibiting near-universal associations with viral elements. Bacterial antiviral defence systems were abundant and taxonomically diverse, dominated by restriction–modification systems, while CRISPR–Cas systems showed pronounced lineage-specific distributions; in contrast, viral anti-defence genes were detected at low frequency (<10% of MAVs). Host prediction linked MAVs to clinically relevant taxa, including expanded ESKAPE pathogens such as Klebsiella pneumoniae, Acinetobacter baumannii, Staphylococcus aureus, and Enterobacter spp., highlighting the ecological connectivity between food-associated viruses and clinically important bacteria. Antimicrobial resistance signals were scarce, suggesting minimal phage-mediated AMR dissemination in food environments. This new publicly available viral database represents a valuable resource for further exploration of viral diversity.