GASAL2

GASAL2 accelerates DNA and RNA sequence alignment on NVIDIA GPUs using CUDA to support high-throughput next-generation sequencing analyses.

Key Features:

• Accelerated alignment kernels: Specialized CUDA kernels implement local, global, and semi-global pairwise sequence alignments with optional traceback.
• Sequence packing performance: GPU sequence packing achieves up to 750 times faster throughput compared to NVBIO.
• Comparative performance: Demonstrated performance improvements over CPU SIMD implementations such as SeqAn and Parasail and over NVIDIA's NVBIO library.
• Asynchronous execution: Alignment functions are asynchronous and non-blocking, enabling full overlap of CPU and GPU execution.
• Hardware and software compatibility: Tested on NVIDIA GPUs with compute capabilities 2.0, 3.5, and 5.0 and demonstrated on a Geforce GTX 1080 Ti; requires Linux and CUDA toolkit version 8 or higher.
• Integration with aligners: Used to accelerate BWA-MEM, achieving a 20× speedup in local alignment and a 1.3× overall application speedup versus CPU execution on a dual-socket hyper-threaded Intel Xeon system with up to 12 threads.

Scientific Applications:

• High-throughput NGS data analysis: Accelerates alignment steps for large-scale next-generation sequencing datasets.
• Genomic research: Enables rapid pairwise alignments for genomic analyses requiring local, global, or semi-global alignment modes.
• Personalized medicine workflows: Supports workflows that require fast alignment of reads for clinical and translational applications.
• Alignment pipeline acceleration: Provides high-performance APIs for incorporation into bioinformatics pipelines and aligners.

Methodology:

Implements CUDA-based GPU kernels for local, global, and semi-global pairwise alignment with optional traceback, performs GPU sequence packing, and exposes asynchronous non-blocking alignment functions to enable CPU–GPU overlap.