FPGA-based accelerator for adaptive banded event alignment in nanopore sequencing data analysis

Area of Science:

• Genomics
• Bioinformatics
• Computer Engineering

Background:

• Adaptive Banded Event Alignment (ABEA) is crucial for Nanopore sequencing analysis, including sequence polishing and DNA methylation detection.
• Current ABEA implementations often rely on CPUs and GPUs.
• Field-Programmable Gate Arrays (FPGAs) offer potential for improved performance, cost-effectiveness, and energy efficiency in computational tasks.

Purpose of the Study:

• To develop and evaluate an efficient FPGA-based accelerator for the ABEA algorithm.
• To demonstrate the performance and energy benefits of using FPGAs for intensive genome analysis tasks.

Main Methods:

• Designed an FPGA-based accelerator leveraging the parallelism and sequential access patterns of the ABEA algorithm.
• Implemented the accelerator targeting a Xilinx VU9P FPGA.
• Compared the performance and energy consumption against CPU-based (Nanopolish) and GPU-based implementations.

Main Results:

• The FPGA-based ABEA accelerator achieved an average throughput speedup of 10.05x over CPU-only.
• It demonstrated a 1.81x speedup over state-of-the-art GPU acceleration while consuming only 7.2% of the energy.
• Achieved a 10.11x speedup compared to a previous FPGA accelerator.

Conclusions:

• Cutting-edge FPGAs can significantly benefit intensive genome analysis.
• FPGA acceleration offers substantial improvements in both performance and energy efficiency for bioinformatics algorithms like ABEA.