MetaVelvet

MetaVelvet extends Velvet to perform de novo metagenome assembly from short sequence reads derived from complex microbial communities.

Key Features:

• Extension of Velvet: Adapts the Velvet single-genome assembler for metagenomic datasets.
• de Bruijn graph decomposition: Decomposes the de Bruijn graph constructed from mixed short reads into sub-graphs representing individual isolate genomes.
• Coverage- and connectivity-based partitioning: Uses coverage (abundance) differences and graph connectivity to partition the graph into species-specific components.
• Scaffold construction per sub-graph: Constructs scaffolds based on each decomposed sub-graph.
• Low-coverage genome reconstruction: Reconstructs low-coverage genome sequences into scaffolds.
• Improved assembly contiguity: Produces higher N50 scores and longer scaffolds compared to conventional single-genome assemblers on simulated and real datasets.
• Enhanced gene recovery: Increases the number of predicted genes recovered from assemblies of human gut microbial read data.

Scientific Applications:

• De novo metagenome assembly: Assembles metagenomes from short-read sequencing data to reconstruct genomes from mixed microbial communities.
• Reconstruction of low-abundance genomes: Recovers and scaffolds genomes present at low coverage within metagenomic samples.
• Human gut microbiome analysis: Produces longer scaffolds and increased gene predictions when applied to human gut microbial read data.
• Benchmarking with simulated datasets: Demonstrates improved assembly contiguity (N50) on simulated datasets relative to single-genome assemblers.

Methodology:

Construct a de Bruijn graph from mixed short reads, decompose the graph into sub-graphs using coverage (abundance) differences and graph connectivity, and construct scaffolds from each decomposed sub-graph.