Fast single individual haplotyping method using GPGPU

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Next-generation sequencing (NGS) data analysis demands substantial computational resources.
• Existing bioinformatic tools often require significant processing power.

Purpose of the Study:

• To assess the utility of graphic processing units (GPUs) for accelerating NGS data computation.
• To evaluate the performance of a parallelized haplotyping algorithm on GPUs.

Main Methods:

• Parallelization of the Probabilistic Evolutionary Algorithm with Toggling for Haplotyping (PEATH) method using general-purpose computing on GPU (GPGPU), creating PEATH/G.
• Testing PEATH/G on NA12878 fosmid-sequencing and HuRef datasets using NVIDIA GeForce GTX 1660Ti and GTX 950 GPUs.

Main Results:

• PEATH/G demonstrated significant speedups, running 46.8x and 25.4x faster than PEATH on the tested datasets.
• Even an inexpensive GPU (GTX 950) provided a 13.3x speed increase for fosmid-sequencing data.
• The parallelized tool maintained accuracy while enhancing computational efficiency.

Conclusions:

• PEATH/G offers a practical solution for single-individual haplotyping, balancing accuracy and speed.
• General-purpose computing on GPU (GPGPU) effectively reduces the computational time for NGS analysis tools.