A Massively Parallel Sequence Similarity Search for Metagenomic Sequencing Data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Sequence similarity searches are crucial for metagenomic data analysis, enabling taxonomic and functional profiling.
• Current methods are time-consuming, hindering large-scale analyses of complex datasets like the human oral microbiome.

Purpose of the Study:

• To develop a highly efficient parallel tool, GHOST-MP (Genome-wide HOmology Search Tool on Massively Parallel system), for accelerating large-scale metagenomic sequence similarity searches.

Main Methods:

• GHOST-MP utilizes a fast search algorithm based on suffix arrays for query and database sequences.
• A hierarchical parallel search strategy is employed to enhance computational efficiency on massively parallel systems.
• The tool's performance was evaluated for parallel computing efficiency and search speed.

Main Results:

• GHOST-MP demonstrates scalability across over 10,000 CPU cores.
• The tool achieved over 80-fold acceleration compared to mpiBLAST on identical computational resources.
• Application to human oral metagenomic data revealed distinct functional gene profiles across different oral sites.

Conclusions:

• GHOST-MP significantly accelerates large-scale metagenomic sequence similarity searches.
• The developed tool enables efficient analysis of complex microbial communities.
• Distinct functional characteristics were identified within the human oral microbiome using GHOST-MP.