deGSM

deGSM constructs scalable de Bruijn graphs from high-throughput sequencing (HTS) data and assembled genomes to enable large-scale sequence analysis and genome reconstruction.

Key Features:

• Scalability: Designed to handle terabase-scale datasets, with demonstrations on GenBank contigs totaling 305 Gbp, scaffolds of 1.1 Tbp, and a Picea abies HTS dataset of 9.7 Tbp.
• Memory efficiency: Builds the Burrows-Wheeler Transform (BWT) of unipaths using constant RAM space.
• Parallel construction: Employs a parallel construction approach for improved efficiency and scalability.
• Unitig reconstruction: Reconstructs original unitigs from the BWT representation.
• Performance: Reports faster or comparable graph-construction speeds relative to existing state-of-the-art methods.
• Data support: Applicable to both high-throughput sequencing reads and assembled genome sequences.

Scientific Applications:

• Large-scale de Bruijn graph construction: Representation and organization of genome sequences for sequence analysis tasks at terabase scale.
• Genome assembly and analysis: Enabling reconstruction and analysis of genomes from HTS datasets and assembled contigs/scaffolds.
• Public dataset processing: Applicable to processing extensive public datasets such as GenBank contigs/scaffolds and large-species HTS datasets (e.g., Picea abies).

Methodology:

Uses a parallel construction approach to build the Burrows-Wheeler Transform (BWT) of de Bruijn-graph unipaths in constant RAM space and then reconstructs original unitigs from the BWT representation.