elPrep

elPrep optimizes the preparation of SAM and BAM files for variant calling pipelines by executing multiple preprocessing steps efficiently.

Key Features:

• Single-Pass Data Processing: Executes multiple preparation steps in a single pass through SAM/BAM data to avoid repeated reads.
• Multithreaded and In-Memory Execution: Runs as a multithreaded application operating primarily in memory to minimize file I/O.
• Merged Computations: Merges computations for filtering, sorting, marking duplicates, and reordering contigs to eliminate redundant processing.
• Performance Efficiency: Benchmarks report a five-step preparation pipeline on whole-exome BAM (NA12878) completes in ~15 minutes with elPrep versus ~1 hour 40 minutes with SAMtools and Picard, and whole-genome runtime reduced from ~24 hours to under 5 hours.
• Scalability and Cost-Effectiveness: Reduces aggregate computation time in large-scale clinical sequencing studies, decreasing overall analysis time and associated costs.

Scientific Applications:

• Variant Calling Pipelines: Preprocessing SAM/BAM files for downstream variant calling workflows.
• Large-Scale Genomic and Clinical Studies: Processing whole-exome and whole-genome sequencing datasets for population- and clinic-scale analyses.
• Disease Research and Personalized Medicine: Accelerating detection of genetic variation relevant to disease studies and individualized treatment decisions.

Methodology:

Implements single-pass, multithreaded, in-memory processing that merges filtering, sorting, duplicate marking, and contig reordering operations; development included benchmarking runtime and memory usage across C++, Go, and Java and selection of Go based on those benchmarks and maintainability considerations.