Comparison of different microbiome analysis pipelines to validate their reproducibility of gastric mucosal microbiome composition.

Microbiome analysis has become a crucial tool for basic and translational research due to its potential for translation into clinical practice. However, there is ongoing controversy regarding the comparability of different bioinformatic analysis platforms and a lack of recognized standards, which might have an impact on the translational potential of results. This study investigates how the performance of different microbiome analysis platforms impacts the final results of mucosal microbiome signatures. Across five independent research groups, we compared three distinct and frequently used microbiome analysis bioinformatic packages (DADA2, MOTHUR, and QIIME2) on the same subset of fastQ files. The source data set encompassed 16S rRNA gene raw sequencing data (V1-V2) from gastric biopsy samples of clinically well-defined gastric cancer (GC) patients ( = 40; with and without [] infection) and controls ( = 39, with and without infection). Independent of the applied protocol, status, microbial diversity and relative bacterial abundance were reproducible across all platforms, although differences in performance were detected. Furthermore, alignment of the filtered sequences to the old and new taxonomic databases (i.e., Ribosomal Database Project, Greengenes, and SILVA) had only a limited impact on the taxonomic assignment and thus on global analytical outcomes. Taken together, our results clearly demonstrate that different microbiome analysis approaches from independent expert groups generate comparable results when applied to the same data set. This is crucial for interpreting respective studies and underscores the broader applicability of microbiome analysis in clinical research, provided that robust pipelines are utilized and thoroughly documented to ensure reproducibility.IMPORTANCEMicrobiome analysis is one of the most important tools for basic and translational research due to its potential for translation into clinical practice. However, there is an ongoing controversy about the comparability of different bioinformatic analysis platforms and a lack of recognized standards. In this study, we investigate how the performance of different microbiome analysis platforms affects the final results of mucosal microbiome signatures. Five independent research groups used three different and commonly used bioinformatics packages for microbiome analysis on the same data set and compared the results. This data set included microbiome sequencing data from gastric biopsy samples of GC patients. Regardless of the protocol used, status, microbial diversity, and relative bacterial abundance were reproducible across all platforms. The results show that different microbiome analysis approaches provide comparable results. This is crucial for the interpretation of corresponding studies and underlines the broader applicability of microbiome analysis.