Review on GPU accelerated methods for genome-wide SNP-SNP interactions.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Modern biobanks offer vast genetic data from large populations, enabling complex disease pathway discovery.
• Analyzing millions of single nucleotide polymorphisms (SNPs) for genome-wide SNP-SNP interactions (epistasis) presents significant computational and statistical challenges.
• Efficient epistasis detection is vital for unlocking the potential of large-scale genomic datasets.

Purpose of the Study:

• To systematically review GPU-accelerated methods for exhaustive epistasis detection.
• To detail the statistical models and computational strategies used in these GPU-based approaches.
• To assess the performance enhancements offered by GPU implementations compared to traditional CPU methods.

Main Methods:

• Systematic review of published GPU-accelerated algorithms for epistasis detection.
• Analysis of statistical models employed for identifying SNP-SNP interactions.
• Evaluation of computational strategies for improving time and memory efficiency on GPUs.
• Comparison of GPU performance against CPU-based approaches.

Main Results:

• GPU-based parallel computing enables high-throughput and cost-effective genomic analysis.
• GPU implementations demonstrate substantial speedups for epistasis detection compared to CPU methods.
• Refined algorithms and GPU acceleration significantly improve time and memory efficiency.
• GPU solutions are effective in addressing the data scale challenges in biobank analysis.

Conclusions:

• GPU-accelerated methods show significant promise for advancing genome-wide epistasis detection.
• These methods are essential for harnessing the full potential of large-scale biobank data.
• Continued innovation in algorithm design and hardware optimization is necessary for future genomic research challenges.