Benchmarking 16S rRNA gene sub-regions against full-length sequencing in vaginal microbiota profiling

The extent to which different 16 S rRNA variable regions yield vaginal microbiota profiles consistent with those obtained from full-length sequences during pregnancy remains unclear. Using full-length 16 S rRNA sequencing data from a Chinese pregnancy cohort, we extracted in silico ten commonly used sub-regions (V1-V2, V1-V3, V3, V4, V3-V4, V3-V5, V4-V5, V4-V6, V6-V8, and V6-V9) and compared them with V1-V9 (internal benchmark). The region-comparison analysis was conducted in a randomly selected sub-cohort of 104 pregnant women, and a secondary case-cohort analysis of 176 women examined whether region choice changed microbiota associations with preterm birth (PTB). Sub-regions differed from V1-V9 across alpha diversity, beta diversity, community state types, taxonomic assignment, species co-occurrence, and predicted functional profiles. V6-V9 best approximated V1-V9 for richness metrics, whereas V1-V3 was most consistent for evenness metrics and Bray-Curtis structure. V6-V8 and V6-V9 performed best for weighted and unweighted UniFrac distances, respectively, and V6-V9 showed the highest concordance for predicted pathway profiles. V1-V3 showed high agreement for major community state types and vaginal types. In the secondary PTB analysis, S_Dialister micraerophilus was associated with PTB in full-length V1-V9, with the direction of the association consistent across subregions. These findings indicate that no single short region exhibits the highest consistency with the full-length sequence in all respects. Region selection should therefore be aligned with the intended analytical endpoint in vaginal microbiota studies.