Development of reference-based model for improved analysis of bacterial community

Abstract

Probiotic bacteria play a vital role in maintaining gut microbial homeostasis and are widely used in various commercial products. Although 16S rRNA amplicon-based next-generation sequencing (NGS) is commonly used to analyze probiotic products, biases can arise from various 16S rRNA amplification regions, sequencing platforms, and library kits. In this study, a reference-based bias correction model was developed to correct sequencing biases. The model was validated using eight mock communities and 12 commercial products, which were analyzed across multiple NGS platforms and various 16S rRNA regions. Specific primer-probe assays were developed for accurate bacterial quantification, and their specificity was validated and used in conjunction with droplet digital PCR (ddPCR) to establish initial bacterial ratios within communities. Analysis of the mock communities revealed platform- and region-specific biases, with specific species consistently over- or under-represented. Similarly, commercial product analyses have shown biased outcomes owing to varying sequencing protocols. The correction model, based on PCR efficiencies from the reference communities, successfully corrected biased ratios across different amplification regions and platforms to achieve results that closely matched the proportions predicted by ddPCR. The model effectively corrected the biases arising from the different polymerases. Notably, partial references containing approximately 40 % of the species achieved correction results that were comparable to those of the complete references. This approach demonstrates the potential for improving microbiome analysis accuracy within predictable ranges, and could serve as a model for addressing sequencing bias in metagenomic research.