DMclust, a Density-based Modularity Method for Accurate OTU Picking of 16S rRNA Sequences

Area of Science:

• Bioinformatics
• Computational Biology
• Microbiology

Background:

• Clustering 16S rRNA sequences into operational taxonomic units (OTUs) is essential for metagenomic data analysis.
• Existing methods face challenges in balancing clustering accuracy with computational efficiency.

Purpose of the Study:

• To introduce DMclust, a novel density-based modularity clustering method for accurate OTU binning.
• To evaluate DMclust's performance against existing methods in metagenomic datasets.

Main Methods:

• DMclust employs a four-phase algorithm: identifying dense sequence groups, constructing a weighted network, applying modularity-based community detection for preclusters, and assigning remaining sequences.
• The method defines dense groups as 'n-sequence communities' within a distance threshold.

Main Results:

• DMclust demonstrates higher accurate clustering performance compared to widely used existing methods.
• The algorithm achieves this accuracy with acceptable memory usage.

Conclusions:

• DMclust offers an effective solution for OTU clustering in metagenomic analysis.
• The method provides a superior balance between clustering accuracy and computational efficiency.