BAMSI

BAMSI enables scalable filtering and extraction of genomic data from large BAM files, facilitating targeted analysis of mapped reads such as the 1000 Genomes phase 3 dataset.

Key Features:

• Scalability and Flexibility: Operates independently of specific storage systems and supports deployment across private and public cloud platforms to process large genomic datasets.
• Storage-Agnostic Framework: Integrates with diverse data storage solutions without requiring storage-specific modifications.
• Parallel Data Utilization: Simultaneously utilizes multiple mirrors of datasets and deploys filtering workers close to data sources to accelerate processing.
• Horizontal Scalability: Supports concurrent distributed processing to scale horizontally for analyses such as structural variation within the 1000 Genomes dataset.
• Data Integration and Cost Efficiency: Produces substantially reduced filtered outputs that can be downloaded or integrated into distributed computing environments like Hadoop clusters and analyzed with Hive or Spark, enabling shared hosting of large datasets.

Scientific Applications:

• Raw and aligned read analyses: Enables analyses that require raw or aligned read datasets rather than curated variant calls.
• Structural variation studies: Facilitates structural variation analysis on large cohorts such as the 1000 Genomes phase 3 mapped reads.
• Distributed downstream analysis: Generates filtered datasets suitable for downstream analysis in Hadoop, Hive, or Spark for re-analyses and exploratory investigations.

Methodology:

Performs distributed filtering of BAM files by deploying filtering workers near data mirrors, operates in a storage-agnostic manner, and outputs reduced-size filtered datasets for download or integration into Hadoop clusters for analysis with Hive or Spark.