Open-Source Sequence Clustering Methods Improve the State Of the Art.

Area of Science:

• Microbial ecology
• Bioinformatics
• Computational biology

Background:

• Sequence clustering into operational taxonomic units (OTUs) is crucial for analyzing amplicon-based microbial communities.
• Accurate and efficient clustering algorithms are needed for massive next-generation sequencing datasets.

Purpose of the Study:

• To benchmark the performance of state-of-the-art open-source sequence clustering tools against established methods.
• To evaluate the impact of quality filtering on species abundance and diversity estimation.

Main Methods:

• Comparison of OTUCLUST, Swarm, SUMACLUST, and SortMeRNA against UCLUST, USEARCH, and mothur's hierarchical clustering.
• Utilized simulated, mock, and environmental microbial communities for analysis.
• Assessed sensitivity, selectivity, alpha and beta diversity, and taxonomic composition.

Main Results:

• New open-source tools (Swarm, SUMACLUST, SortMeRNA) reported up to 60% fewer spurious OTUs compared to UCLUST.
• Stringent quality filtering, as used in UPARSE, can lead to significant underestimation of species diversity.
• Recent clustering algorithms improve accuracy and preserve diversity without aggressive filtering.

Conclusions:

• Open-source tools like Swarm, SUMACLUST, and SortMeRNA offer improved accuracy and diversity preservation in microbial community analysis.
• These tools are scalable, multithreaded, and suitable for large-scale projects like the Earth Microbiome Project.
• Avoiding aggressive filtering is key to obtaining biologically accurate results in sequence clustering.