MTSv: rapid alignment-based taxonomic classification and high-confidence metagenomic analysis

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Growing sequence databases and high-throughput data challenge traditional alignment for metagenomic taxonomic classification.
• Exact matching methods are fast but lack accuracy, leading to false positives.
• Full alignment tools offer higher confidence but are computationally intensive.

Purpose of the Study:

• To develop a computationally efficient and accurate tool for alignment-based taxonomic assignment in metagenomic analysis.
• To address the limitations of existing methods in handling large datasets and divergent genomes.

Main Methods:

• Designed MTSv, incorporating an FM-index assisted q-gram filter and SIMD-accelerated Smith-Waterman algorithm.
• Implemented a strategy to cease alignment to a TaxID upon finding a high-quality match, enhancing efficiency.
• Ensured flexibility for memory or processor-constrained systems.

Main Results:

• MTSv achieves higher precision than popular exact matching approaches, though not matching their speed.
• The tool performs full alignments, enabling classification of reads with low similarity to reference sequences.
• Demonstrates potential for high-confidence pathogen detection with reduced off-target assignments.

Conclusions:

• MTSv provides a balanced solution for metagenomic taxonomic classification, prioritizing accuracy and confidence.
• The tool is suitable for identifying pathogens and analyzing microbial communities, especially in resource-limited environments.
• Offers a valuable alternative for researchers needing reliable taxonomic assignments from complex metagenomic data.