CLAST: CUDA implemented large-scale alignment search tool.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Metagenomics relies on nucleotide sequence similarity searching against databases.
• Next-generation sequencing generates vast amounts of data, often from unsequenced microbes.
• Existing tools struggle with large datasets and identifying weak similarities in novel microbial genomes.

Purpose of the Study:

• To develop a rapid sequence similarity search tool for large-scale metagenomic analyses.
• To address the limitations of current tools in handling massive next-generation sequencing data.
• To enable accurate detection of weak sequence similarities in microbial communities.

Main Methods:

• Development of CLAST (CUDA implemented large-scale alignment search tool).
• Utilizes NVIDIA Fermi architecture graphics processing units for high-speed computation.
• Supports both global and local alignment, with global alignment as default.

Main Results:

• CLAST is significantly faster than BLAST (~80.8x) and BLAT (~9.6x).
• Achieves high accuracy in assigning reads to taxonomic and functional groups using distant nucleotide sequences.
• Requires minimal main memory (<2 GB) and no preprocessed sequence databases.

Conclusions:

• CLAST offers high speed and sensitivity comparable to established tools like Bowtie 2, BLAST, BLAT, and FR-HIT.
• Eliminates the need for extensive database preprocessing or specialized hardware.
• Represents a powerful and practical approach for analyzing massive next-generation sequencing data in metagenomics.