Separating metagenomic short reads into genomes via clustering

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Metagenomics enables simultaneous sequencing of all genomes in environmental samples, yielding complex datasets with unknown genome numbers and abundance ratios.
• Next-generation sequencing (NGS) technologies produce shorter reads, complicating species read separation.
• Existing methods like similarity-based and composition-based approaches have limitations with unknown species, short reads, and similar abundance levels.

Purpose of the Study:

• To develop a novel heuristic algorithm for separating short paired-end reads from different genomes in metagenomic data.
• To address limitations of existing methods, particularly for short reads and genomes with similar abundance ratios.

Main Methods:

• A two-phase heuristic algorithm utilizing l-mer frequencies and repeat information.
• Phase one clusters l-mers belonging to individual genomes.
• Phase two merges clusters based on repeat data to assign reads, handling short reads and sequencing errors.

Main Results:

• The algorithm successfully separates reads from different genomes, even with short reads and sequencing errors.
• It is effective for genomes with similar abundance levels and has been extended for arbitrary ratios.
• The software is available for free download.

Conclusions:

• The developed method demonstrates high precision and sensitivity in separating NGS reads.
• Genome separation is feasible when genome-specific repeats outweigh common repeats.
• This advancement aids in analyzing complex metagenomic datasets.