Related Experiment Video
Updated: May 23, 2025
13:24
Integration of Wet and Dry Bench Processes Optimizes Targeted Next-generation Sequencing of Low-quality and Low-quantity Tumor Biopsies
Published on: April 11, 2016
11.8K
Benchmarking accelerated next-generation sequencing analysis pipelines
Pubudu Saneth Samarakoon1, Ghislain Fournous2, Lars T Hansen1
1Scientific Computing Services, Division for Research, Dissemination and Education, University of Oslo, Oslo, 0373, Norway.
Bioinformatics Advances
|May 21, 2025
Summary
Accelerated next-generation sequencing (NGS) platforms like DRAGEN and Parabricks significantly reduce analysis time compared to CPU-based methods. Performance varies, with Parabricks-H100 showing top speedups, but scalability and resource usage differ across platforms.
Area of Science:
- Genomics
- Bioinformatics
- Computational Biology
Background:
- Central processing unit (CPU)-based next-generation sequencing (NGS) analysis tools face limitations due to long runtimes, impacting clinical and research applications.
- Accelerated NGS platforms, such as DRAGEN and Parabricks, have been developed to address these runtime issues, reducing analysis from days to hours.
Purpose of the Study:
- To comprehensively evaluate the performance, computational resource usage, and speedup scalability of accelerated NGS platforms.
- To address the gap in independent assessments of accelerated NGS platforms by investigating their efficiency and scalability.
Main Methods:
- Comparative analysis of CPU-only NGS pipelines against accelerated platforms (DRAGEN, Parabricks) using various hardware configurations (L4, A100, H100).
- Assessment of mapping and variant calling performance, speedups, and computational resource utilization.
- Scalability analysis based on sequencing coverage and profiler analysis for performance insights.
Main Results:
- Accelerated pipelines showed shorter runtimes than CPU-only methods, with Parabricks-H100 achieving the highest speedups.
- DRAGEN excelled in mapping speed, while Parabricks (A100, H100) demonstrated superior speedups in variant calling.
- Mapping scalability analysis indicated positive trends for DRAGEN and Parabricks-H100, while other configurations showed limitations. Profiler analysis revealed optimization potential for Parabricks.
Conclusions:
- Accelerated NGS platforms offer significant runtime reductions for genomic analyses.
- Platform selection should consider specific needs regarding coverage, time constraints, and budget, informed by performance and cost comparisons.
- Further optimization of accelerated platforms like Parabricks is possible, potentially leading to enhanced efficiency and scalability.
