Velvet

Velvet assembles de novo genomic sequences from short-read sequencing technologies (Solexa, 454, SOLiD) by constructing and manipulating de Bruijn graphs of k-mers to reconstruct contigs for genomic analyses.

Key Features:

• De Bruijn graph utilization: Constructs and manipulates de Bruijn graphs based on k-mers to assemble short reads into contigs.
• Short-read sequencing support: Targets high-coverage datasets with very short reads (25–50 base pairs) from technologies including Solexa, 454, and SOLiD.
• Paired-end read integration: Incorporates paired-end read information to improve contiguity and accuracy of assemblies.
• Performance on simulated data: Produced contigs with N50 up to 50 kilobases for prokaryotic simulations and approximately 3 kilobases for simulated mammalian BACs.
• Performance on real data: On Solexa datasets without paired-end information generated contigs averaging ~8 kilobases for prokaryotes and ~2 kilobases for mammalian BACs.

Scientific Applications:

• Prokaryotic genome assembly: Generation of long contigs (simulated N50 up to 50 kb; real-data averages around 8 kb) from high-coverage short reads.
• Mammalian BAC assembly: Assembly of simulated mammalian BACs (N50 ~3 kb) and real-data BAC contigs averaging ~2 kb from short reads.
• Downstream genomic analyses: Provides assembled contigs for gene prediction, comparative genomics, and evolutionary studies.

Methodology:

Constructs and manipulates de Bruijn graphs from k-mers and incorporates paired-end read information for de novo assembly of short reads (25–50 bp) from Solexa, 454, and SOLiD.