CloudBurst

CloudBurst maps short reads from next-generation DNA sequencing technologies to reference genomes to provide scalable, parallel alignment for large-scale analyses such as SNP discovery, genotyping, and personal genomics.

Key Features:

• Parallelization with MapReduce: CloudBurst leverages the Hadoop implementation of MapReduce to parallelize execution across multiple compute nodes, enabling scaling with data size and computational resources.
• Performance Efficiency: In a 24-processor core configuration, CloudBurst achieves up to a 30-fold increase in speed compared to single-core execution of RMAP while producing an identical set of alignments.
• Scalability: Running time scales linearly with the number of reads mapped and exhibits near-linear speedup with added processors, with reported performance improvements exceeding 100-fold on a 96-core remote compute cloud, reducing processing from hours to minutes for millions of short reads.
• Flexibility in Alignment Reporting: Modeled after RMAP, CloudBurst can report either all possible alignments or the unambiguous best alignment per read and accommodate any number of mismatches or differences.

Scientific Applications:

• Reference Genome Mapping: Mapping next-generation sequence data to the human genome and other reference genomes.
• SNP Discovery: Identifying single nucleotide polymorphisms within large sequencing datasets.
• Genotyping: Determining genetic variants across individuals or populations.
• Personal Genomics: Facilitating analyses of individual genomic data for personalized medicine applications.

Methodology:

CloudBurst uses the Hadoop implementation of MapReduce to distribute read-mapping tasks across compute nodes and implements alignment reporting behavior modeled on RMAP.