Massively parallel read mapping on GPUs with the q-group index and PEANUT

Area of Science:

• Bioinformatics
• Computational Biology
• High-Performance Computing

Background:

• Accurate and efficient read mapping is crucial for genomic data analysis.
• Existing read mappers often face challenges with large datasets and computational demands.
• Graphics Processing Units (GPUs) offer potential for accelerating sequence alignment.

Purpose of the Study:

• Introduce a novel data structure, the q-group index, optimized for GPU architectures.
• Present PEANUT, a highly parallel GPU-based read mapper built upon the q-group index.
• Evaluate PEANUT's performance against state-of-the-art read mappers.

Main Methods:

• Developed the q-group index for efficient querying and building on GPUs.
• Implemented PEANUT, a parallel read mapper leveraging the q-group index.
• Conducted benchmark experiments comparing PEANUT with existing read mappers.

Main Results:

• PEANUT demonstrates significant speed improvements over other state-of-the-art read mappers.
• The q-group index enables a small memory footprint for GPU-based mapping.
• PEANUT maintains or slightly improves precision, recall, and sensitivity metrics.

Conclusions:

• PEANUT represents a significant advancement in GPU-accelerated read mapping.
• The q-group index is an effective data structure for parallel sequence alignment on GPUs.
• PEANUT offers a compelling solution for fast and accurate genomic data analysis.