A comparative analysis of gut microbiome in dogs using short- and long-reads of 16S rRNA sequences reveals workflow-dependent biases.

Short-read Illumina sequencing of hypervariable regions of the 16S rRNA gene and long-read Oxford Nanopore Technologies (ONT) sequencing of the full-length 16S gene are increasingly used to profile microbial communities. However, differences in sequencing chemistry, read length, and taxonomic assignment methods raise concerns about the comparability of microbiome profiles and their impact on biological interpretation. Faecal samples from two groups of healthy dogs, 8 young (32 < age < 59 months) and 8 old (age > 109 months), were analysed. Taxonomic relative abundances obtained from Illumina and ONT sequencing were compared after nomenclature harmonisation. Agreement between workflows was assessed using Bland-Altman analysis on log2-transformed relative abundances. Although the mean bias at the genus level was small (0.229), limits of agreement indicated poor interchangeability between workflows. When restricted to taxa detected by both workflows, variability decreased but a larger negative bias emerged, suggesting abundance-dependent discrepancies. The effects of sequencing workflow on biologically relevant signals related to host age were evaluated by computing alpha and beta diversity metrics independently within each workflow. Alpha diversity differed between workflows, with ONT yielding higher Shannon diversity and richness values than Illumina. Beta diversity analyses indicated significant (p < 0.05) age-related shifts in community composition only in ONT, with differences in variance explained and effect sizes. Workflow comparisons revealed taxon-specific differences at phylum, family, and genus levels, affecting both moderately and highly abundant taxa. These findings indicate that sequencing workflow choice impacts microbiome profiling and downstream interpretation, underscoring the need for care in cross-workflow comparisons.