HARU - Hardware Accelerated Read Until

Problem

Nanopore sequencers, particularly from Oxford Nanopore Technologies (ONT), enable long-read, high-throughput sequencing with real-time data streaming. This capability introduces Read Until, a selective sequencing method where uninterested reads are rejected during sequencing. However, real-time selective sequencing requires fast and efficient computation, which is challenging on resource-constrained devices. Most existing solutions rely on large HPC systems or high-end GPUs, which are not energy-efficient or portable.

Solution - HARU

HARU is a hardware-software co-design framework that accelerates the Read Until process using a low-cost, off-the-shelf MPSoC (Multiprocessor System-on-Chip) with an on-chip FPGA. It implements a memory-efficient subsequence dynamic time warping (sDTW) algorithm to align genomic data in real-time. HARU achieves significant speedup and energy efficiency by performing the compute-intensive alignment on an FPGA while running supporting tasks on ARM cores.

Citing HARU

@article{shih2023efficient,
  title={Efficient real-time selective genome sequencing on resource-constrained devices},
  author={Shih, Po Jui and Saadat, Hassaan and Parameswaran, Sri and Gamaarachchi, Hasindu},
  journal={GigaScience},
  volume={12},
  pages={giad046},
  year={2023},
  publisher={Oxford University Press}
}