Taxonomy Informed Clustering, an Optimized Method for Purer and More Informative Clusters in Diversity Analysis and...

Area of Science:

• Microbiology
• Bioinformatics
• Computational Biology

Background:

• 16S rRNA gene amplicon sequencing is standard for bacterial diversity analysis.
• Current methods often use greedy clustering algorithms based on similarity cutoffs, which can group phylogenetically diverse sequences.
• Taxonomic classifiers offer more precise sequence assignment.

Purpose of the Study:

• To introduce Taxonomy Informed Clustering (TIC), a novel pipeline for 16S rRNA amplicon data processing.
• To improve the accuracy and purity of sequence clustering in bacterial diversity studies.
• To provide an automated and efficient tool for taxonomic classification and diversity analysis.

Main Methods:

• Developed a pipeline integrating denoising, taxonomic classification, and the TIC clustering algorithm.
• TIC restricts clustering to sequences sharing the same taxonomic path, utilizing classifier-assigned taxonomy.
• Evaluated TIC against greedy algorithms (USEARCH, VSEARCH) using Living Tree Project data and IMNGS database Bifidobacteriaceae sequences.

Main Results:

• TIC demonstrates superior cluster purity and entropy compared to greedy clustering algorithms.
• Application of TIC to the IMNGS Bifidobacteriaceae dataset revealed thousands of novel molecular genera and species.
• The pipeline provides a straightforward and effective approach for bacterial diversity studies.

Conclusions:

• Taxonomy Informed Clustering (TIC) offers a significant advancement over traditional greedy clustering methods for 16S rRNA amplicon data.
• The TIC pipeline enables more accurate and sensitive identification of bacterial diversity, including novel taxa.
• This approach enhances the reliability of microbial community analyses.