A De Novo Robust Clustering Approach for Amplicon-Based Sequence Data.

Area of Science:

• Microbiology
• Bioinformatics
• Computational Biology

Background:

• Analyzing microbial communities relies on categorizing 16S rRNA gene sequences into operational taxonomic units (OTUs).
• Current clustering tools use fast, one-pass algorithms but employ a "crisp" approach, assigning each sequence to a single OTU.
• This crisp clustering often yields lower quality results, necessitating manual postprocessing by users.

Purpose of the Study:

• To develop a new clustering approach for 16S rRNA gene sequences that accounts for uncertainty.
• To improve the quality and presentation of operational taxonomic unit (OTU) results.
• To provide a membership degree for sequence-to-OTU assignments to aid postprocessing and quality evaluation.

Main Methods:

• A novel clustering algorithm was developed to handle uncertainty in 16S rRNA gene sequence categorization.
• The approach assigns a membership degree to each sequence for its potential OTUs.
• This probabilistic assignment aims to overcome limitations of traditional crisp clustering methods.

Main Results:

• The proposed method generates OTUs with associated membership degrees, offering a more nuanced view of sequence-OTU relationships.
• This approach is expected to improve the overall quality of OTU clustering compared to existing tools.
• The membership degrees facilitate automated quality assessment and simplify downstream data interpretation.

Conclusions:

• The developed clustering approach effectively addresses the challenge of uncertainty in 16S rRNA gene sequence analysis.
• Incorporating membership degrees enhances the utility of OTUs for microbial community studies.
• This method offers a significant advancement in computational tools for microbiome research.