**Background:** The gut microbiome (GM) in early life is critical for immune system maturation, and early-life GM dysbiosis is linked to chronic inflammatory diseases. While most research has focused on bacteria, viruses—primarily bacteriophages—are prominent GM members that can modulate bacterial composition and directly interact with the human immune system. However, large-scale studies of the infant gut virome have been sparse, and a large proportion of viral sequences (viral 'dark matter') remain uncharacterized. Recent advances by the ICTV now allow defining new viral taxa from sequence data alone, enabling systematic exploration of viral diversity.
**Methods:** The study was embedded in the COPSAC2010 prospective mother–child cohort (700 Danish children). Fecal samples were collected at age 1 year from 647 infants. Viromes were extracted via viral particle enrichment, short (30 min) multiple-displacement amplification (sMDA), and Illumina HiSeq X sequencing (average 3 Gbp per sample). After assembly (SPAdes) and species-level de-duplication (95% ANI), contigs were clustered by protein content and manually curated to remove bacterial contaminants. The final set of 10,021 viral OTUs (vOTUs) was pooled with 7,705 reference phage species. Viral orthologue gene clusters (VOGs) were defined de novo, and an aggregate protein similarity (APS) tree was constructed. The tree was cut at distances corresponding to viral genera (0.250), subfamilies (0.125), families (0.04), and orders (0.025), reproducing known taxonomies (e.g., Herelleviridae, Crassvirales). Host predictions used 317,968 CRISPR spacers from metagenome-assembled genomes (MAGs), 11 million spacers from CRISPRopenDB, and WIsH, merged by last common ancestor. Phage lifestyle (temperate vs. virulent) was determined by screening complete vOTUs for integrase genes.
**Key Results:** The 10,021 vOTUs fell within 248 curated VFCs: 16 known families (2,497 vOTUs) and 232 previously undescribed VFCs (7,524 vOTUs). 56% of vOTUs were complete or near-complete (83% of ssDNA, 46% of dsDNA). 70% of infant gut vOTUs were absent from three major adult gut virus databases (GVD, GPD, MGV). The infant gut virome was dominated by temperate phages, contrasting with the virulent-dominated adult gut. The ten major caudoviral VFCs included four known families (Skunaviridae, Salasmaviridae, β-Crassviridae, Flandersviridae) and six novel candidate families (e.g., 'Sisseviridae' present in 80% of infants, 'Amandaviridae', 'Jeppeviridae', 'Alberteviridae', 'Evaviridae', 'Hannahviridae'). Crassvirales, abundant in adults, were surpassed by other VOCs in infants. ssDNA viruses (microviruses, anelloviruses, inoviruses) accounted for 60% of mean relative abundance (MRA) after genome-size normalization. Temperate VFCs were significantly more prevalent (P=0.048) and genetically diverse (P=0.021) than virulent VFCs, but virulent VFCs had significantly higher abundance relative to prevalence (neutral community model residuals, P=2.1×10⁻⁵). Bacterial host genera were predicted for 63% of vOTUs; Bacteroides, Faecalibacterium, and Bifidobacterium were the top three hosts. Phage MRA correlated strongly with host bacterial MRA (Spearman's ρ=0.76, P<1.45×10⁻¹⁷).
**Clinical Implications:** This study provides the most comprehensive taxonomic atlas of the infant gut virome to date, resolving the viral 'dark matter' to only 7% unaccounted sequences. The dominance of temperate phages in infants versus virulent phages in adults suggests a developmental shift in phage ecology that may parallel immune system maturation. The identification of hundreds of novel viral clades and their predicted bacterial hosts enables future mechanistic studies on how the virome influences GM composition, immune programming, and risk of chronic inflammatory diseases. The publicly available interactive resource (http://copsac.com/earlyvir/f1y/fig1.svg) allows researchers to browse viral genomes, gene content, and taxonomy, facilitating translational viromics research.