MEGAHIT

MEGAHIT assembles NGS metagenomic sequencing reads de novo using a succinct de Bruijn graph to produce assemblies of large and complex metagenomic datasets.

Key Features:

• De novo NGS metagenomic assembly: Performs de novo assembly of next-generation sequencing (NGS) metagenomic reads.
• Succinct de Bruijn graph: Uses a succinct de Bruijn graph approach to reduce memory footprint during assembly.
• Low memory operation: Operates with low memory usage while maintaining high performance.
• No preprocessing required: Assembles large datasets without requiring partitioning or normalization preprocessing.
• GPU-accelerated and CPU modes: Supports GPU-accelerated execution (demonstrated) and also runs without a GPU.
• Scalability demonstration: Successfully assembled a 252 gigabase pair (Gbps) soil metagenomics dataset in 44.1 hours with a GPU and in 99.6 hours without a GPU on a single computing node.
• Improved assembly quality: Produces assemblies reported as three times larger than previous methods, with improved contig N50 and average contig lengths.
• Increased read alignment: Achieved 55.8% of reads aligned to the assembly for the demonstrated dataset, representing a fourfold improvement over earlier techniques.
• Environmental sample suitability: Effective for extensive metagenomic data from environments such as soil.

Scientific Applications:

• Large-scale metagenome assembly: Assembly of large and complex metagenomic datasets, including soil metagenomes.
• Comparative assembly benchmarking: Comparative evaluation to improve assembly size, contig N50, and average contig lengths versus previous methods.
• Read recruitment enhancement: Increasing the fraction of reads that align to assembled metagenomes (e.g., 55.8% in the demonstrated dataset).

Methodology:

Performs de novo assembly of NGS metagenomic reads using a succinct de Bruijn graph approach and can run with GPU acceleration or in CPU-only mode.