Results and discussionThese two raw soy sauces showed differences in phage composition (121 viral operational taxonomic units (vOTUs) in NJ and 387 vOTUs in PJ), with a higher abundance of the family Siphoviridae (58.50%) in the NJ phage community and a higher abundance of Myoviridae (33.01%) in PJ. Auxiliary metabolic functional annotation analyses showed that phages in the raw soy sauces mostly encoded genes with unknown functions (accounting for 66.33% of COG profiles), but the NJ sample contained genes mostly annotated to conventional functions related to carbohydrate metabolism (0.74%) and lipid metabolism (0.84%), while the PJ sample presented a higher level of amino acid metabolism functions (0.12%). Thirty auxiliary metabolism genes (AMGs) were identified in phage genomes, which were associated with carbohydrate utilization, cysteine and methionine metabolism, and aspartic acid biosynthesis for the host. To identify phage-host interactions, 30 host genomes (affiliated with 22 genera) were also recruited from the metagenomic dataset. The phage-host interaction analysis revealed a wide range of phage hosts, for which a total of 57 phage contigs were associated with 17 host genomes, with Shewanella fodinae and Weissella cibaria infected by the most phages. This study provides a comprehensive understanding of the phage community composition, auxiliary metabolic functions, and interactions with hosts in two different types of raw soy sauce.