Optimal 16S rRNA gene amplicon sequencing analysis for oral microbiota to avoid the potential bias introduced by trimming length, primer, and database

Abstract Background . 16S rRNA gene amplicon sequencing analysis is widely used to investigate the diversity and complexity of bacterial communities in the environment. However, the bacterial composition estimated from the experimental data can differ from the original composition. Such a bias occurs depending on methodological stages, including trimming length, selected amplification regions, and referenced databases. The optimal condition to minimize the bias for oral bacterial analysis remain unknown. Therefore, this study aimed to evaluate the possible bias in 16S rRNA gene amplicon analysis using three bacterial DNA samples, namely mock1 community, which comprised 15 bacteria from various environments, mock2 community, which comprised 6 major oral bacteria, and dental calculus obtained from 5 patients, along with different trimming lengths, three databases, and nine primers targeting different hypervariable regions. Results . Mock1 community analysis results at the genus level showed the highest similarity between the data using 300 bp paired-end (PE), primer targeting V3 region, and SILVA ribosomal RNA database (SILVA) and the theoretical value obtained from the bacterial species. Mock2 community analysis with 300 bp PE showed one of the highest similarities between the theoretical value and data using the V3–V4 region with the Human Oral Microbiome Database (HOMD) at the genus level and data using the V1–V2 region with HOMD at the species level. In the species analysis of the dental calculus samples with 300 bp PE, the Shannon index value was higher in the V1–V2 region with HOMD than that in other combinations of primers and databases. The composition of the relative bacterial abundance was more markedly influenced by the inter-individual variability in the samples than the selected amplified region and/or database. Conclusion . The optimal conditions for analyzing oral microbiota with the most negligible bias were determined to be a combination of 300 bp PE, the primer targeting the V1–V2 region, and the HOMD database. Notably, this is the first report for such analyses of modern Japanese dental calculus. Furthermore, the methods of this study will be a guide for setting the appropriate sequence analysis conditions for each environment.