Fig. 1

Recovery of vOTUs and their hypervariable regions (HVRs) from the paired short-read-only (SR) and short+long-read (SLR) assemblies from CB. Both assemblies were conducted with identical sequencing depth (i.e., identical number of nucleotides; see Table S2). A–C Comparison for sample CB17: A Number of vOTUs recovered from SR (lavender) and SLR (light green) assemblies by selecting vOTUs at lengths of ≥5, ≥10, ≥25, and ≥50 kb. B Number of unique and shared vOTUs between the two assembly types; viral contigs were clustered into a vOTU if sharing ≥95% nucleotide identity across ≥80% of their lengths. C HVR identification of all vOTUs including the unique and shared ones in both assemblies. More and longer HVRs were identified in vOTUs recovered from SLR assembly than SR assembly. D–F Comparison for sample CB18: D, E, and F display the same type of information as A, B, and C, respectively. G An example to show the genome matches of contig fragments recovered from SR assembly to the long contig (i.e., CB17_contig_337_pilon) recovered from SLR assembly. More examples of such comparisons are provided in Tables S3 and S4 for the vOTUs from the samples CB17 and CB18, respectively. SR assembly: only short reads from Illumina sequencing were used for assembly; SLR assembly: half depth of Illumina short reads and half depth of Nanopore long reads were used for hybrid and Pilon software-based assemblies (see Table S2 and “Methods”)