GeFaST: An improved method for OTU assignment by generalising Swarm's fastidious clustering approach.

Area of Science:

• Genomics
• Bioinformatics
• Microbial Ecology

Background:

• High-throughput sequencing generates massive genomic data, enabling deeper understanding of microbial communities.
• Clustering 16S rRNA gene sequences into operational taxonomic units (OTUs) is crucial for analyzing microbial diversity and structure.
• Swarm's clustering strategy improved upon existing de novo methods but had limitations, particularly with its fastidious option.

Purpose of the Study:

• To introduce GeFaST, an exact, alignment-based de novo clustering tool.
• To generalize Swarm's fastidious clustering option for broader applicability and improved performance.
• To evaluate GeFaST's clustering quality and speed against existing tools.

Main Methods:

• GeFaST implements a generalized version of Swarm's fastidious clustering strategy.
• The tool allows for arbitrary clustering thresholds and adjustable greediness.
• Evaluations were performed on both mock-community and natural microbial data.

Main Results:

• GeFaST achieved higher clustering quality and performance compared to Swarm and other de novo tools, especially at small to medium thresholds.
• Clustering with GeFaST was significantly faster, ranging from 6 to 197 times the speed of Swarm.
• Memory usage varied, with Swarm using less memory for non-fastidious clustering but GeFaST being more memory-efficient for fastidious clustering.

Conclusions:

• GeFaST expands the utility of Swarm's clustering approach by generalizing the fastidious option.
• The tool offers enhanced clustering quality and performance, particularly for the fastidious clustering mode.
• GeFaST is poised to facilitate the application of fastidious clustering strategies to larger datasets and higher thresholds.