Related Experiment Video
Updated: Oct 14, 2025
10:24
Next-generation Sequencing of 16S Ribosomal RNA Gene Amplicons
Published on: August 29, 2014
83.9K
Reprocessing 16S rRNA Gene Amplicon Sequencing Studies: (Meta)Data Issues, Robustness, and Reproducibility
Xiongbin Kang1,2, Dong Mei Deng1, Wim Crielaard1
1Department of Preventive Dentistry, Academic Centre for Dentistry Amsterdam (ACTA), University of Amsterdam and Vrije Universiteit Amsterdam, Amsterdam, Netherlands.
Frontiers in Cellular and Infection Microbiology
|November 8, 2021
Summary
Bioinformatics pipelines for analyzing microbial communities using 16S rRNA gene sequencing can yield different results. Ensuring consistent data and methods is crucial for reproducible microbiome research.
Area of Science:
- Microbial ecology
- Bioinformatics
- Genomics
Background:
- High-throughput sequencing is key for microbial ecology studies.
- Bioinformatics pipelines convert 16S rRNA gene data into operational taxonomic unit (OTU) tables for microbial community analysis.
- Assessing pipeline robustness is vital for reliable statistical outcomes.
Purpose of the Study:
- To evaluate the consistency of four mainstream bioinformatics pipelines (VSEARCH, USEARCH, mothur, UNOISE3) in analyzing oral microbiome datasets.
- To identify potential influences of pipeline choice and parameters on diversity metrics and statistical test results.
- To highlight the need for improved methods description and data deposition in published microbiome studies.
Main Methods:
- Searched for publicly available oral microbiome datasets (n=5) related to smoking, oral cancer, caries, diabetes, and periodontitis.
- Processed retrieved datasets using four distinct bioinformatics pipelines: VSEARCH, USEARCH, mothur, and UNOISE3.
- Analyzed alpha-diversity and beta-diversity, tested group differences, and compared results across pipelines, including rarefaction for normalization.
Main Results:
- Only 57% of deposited datasets included complete sequencing and metadata, with processing issues arising from read characteristics and tool defaults.
- While generally similar, P-values occasionally differed significantly between pipelines, impacting statistical conclusions.
- Rarefaction as a normalization method necessitates analyzing multiple subsamples for robust reproducibility.
Conclusions:
- Significant variations in P-values across pipelines underscore the impact of bioinformatics choices on microbiome study outcomes.
- Improvements in the description of bioinformatics methods and data deposition practices are essential for published research.
- Reproducibility in microbiome analysis, particularly when using rarefaction, requires careful consideration and potentially multiple subsample analyses.
