Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Parallel DNA Sequence Alignment on High-Performance Systems with CUDA and MPI

Linus Zwaka

TL;DR

The paper tackles the computational bottleneck of global sequence alignment by implementing Needleman-Wunsch on a hybrid CUDA-MPI platform, introducing a per-cell parallel CUDA kernel for single alignments and using MPI to distribute many alignments across a cluster. It employs a center-star heuristic for multiple sequence alignment and uses a per-cell dependency grid with spin-wait synchronization to minimize idle time, achieving substantial speedups over CPU baselines. Strong and weak scaling analyses demonstrate effective GPU and distributed-resource utilization for large-scale alignments, with the overall time for pairwise alignments scaling as $O(n^2 k^2)$. The work highlights the potential of high-performance computing to advance sequence alignment workflows with minimal changes to the core DP algorithm, paving the way for genome-scale analyses on HPC systems.

Abstract

Sequence alignment is a cornerstone of bioinformatics, widely used to identify similarities between DNA, RNA, and protein sequences and studying evolutionary relationships and functional properties. The Needleman-Wunsch algorithm remains a robust and accurate method for global sequence alignment. However, its computational complexity, O(mn), poses significant challenges when processing large-scale datasets or performing multiple sequence alignments. To address these limitations, a hybrid implementation of the Needleman-Wunsch algorithm that leverages CUDA for parallel execution on GPUs and MPI for distributed computation across multiple nodes on a supercomputer is proposed. CUDA efficiently offloads computationally intensive tasks to GPU cores, while MPI enables communication and workload distribution across nodes to handle large-scale alignments. This work details the implementation and performance evaluation of the Needleman-Wunsch algorithm in a massively parallel computing environment. Experimental results demonstrate significant acceleration of the alignment process compared to traditional CPU-based implementations, particularly for large input sizes and multiple sequence alignments. In summary, the combination of CUDA and MPI effectively overcomes the computational bottlenecks inherent to the Needleman-Wunsch algorithm without requiring substantial modifications to the underlying algorithm, highlighting the potential of high-performance computing in advancing sequence alignment workflows.

Parallel DNA Sequence Alignment on High-Performance Systems with CUDA and MPI

TL;DR

The paper tackles the computational bottleneck of global sequence alignment by implementing Needleman-Wunsch on a hybrid CUDA-MPI platform, introducing a per-cell parallel CUDA kernel for single alignments and using MPI to distribute many alignments across a cluster. It employs a center-star heuristic for multiple sequence alignment and uses a per-cell dependency grid with spin-wait synchronization to minimize idle time, achieving substantial speedups over CPU baselines. Strong and weak scaling analyses demonstrate effective GPU and distributed-resource utilization for large-scale alignments, with the overall time for pairwise alignments scaling as . The work highlights the potential of high-performance computing to advance sequence alignment workflows with minimal changes to the core DP algorithm, paving the way for genome-scale analyses on HPC systems.

Abstract

Sequence alignment is a cornerstone of bioinformatics, widely used to identify similarities between DNA, RNA, and protein sequences and studying evolutionary relationships and functional properties. The Needleman-Wunsch algorithm remains a robust and accurate method for global sequence alignment. However, its computational complexity, O(mn), poses significant challenges when processing large-scale datasets or performing multiple sequence alignments. To address these limitations, a hybrid implementation of the Needleman-Wunsch algorithm that leverages CUDA for parallel execution on GPUs and MPI for distributed computation across multiple nodes on a supercomputer is proposed. CUDA efficiently offloads computationally intensive tasks to GPU cores, while MPI enables communication and workload distribution across nodes to handle large-scale alignments. This work details the implementation and performance evaluation of the Needleman-Wunsch algorithm in a massively parallel computing environment. Experimental results demonstrate significant acceleration of the alignment process compared to traditional CPU-based implementations, particularly for large input sizes and multiple sequence alignments. In summary, the combination of CUDA and MPI effectively overcomes the computational bottlenecks inherent to the Needleman-Wunsch algorithm without requiring substantial modifications to the underlying algorithm, highlighting the potential of high-performance computing in advancing sequence alignment workflows.
Paper Structure (13 sections, 2 equations, 5 figures)

This paper contains 13 sections, 2 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Serial Needleman-Wunsch Algorithm
  3. Grid Construction
  4. Populate Initial Data
  5. Scoring
  6. Backtracking
  7. Parallel Implementation
  8. Single Alignment
  9. Multiple Alignment
  10. Analysis
  11. Strong Scaling
  12. Weak Scaling
  13. Conclusion

Figures (5)

  • Figure 1: An example two-dimensional alignment grid with primary data as described by the Needleman-Wunsch algorithm
  • Figure 2: A continuation of the alignment shown in Figure 1 after applying the scoring algorithm.
  • Figure 3: Strong scaling comparison of a COL1A1 sequence alignment between mouse and human
  • Figure 4: Strong scaling comparison of a THAP11 sequence alignment between mouse and human
  • Figure 5: Weak scaling comparison of execution times as problem size increases