Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 26 of the 29 references it cites were found in a registry or an index; 7 of the 15 sources its claims cite were checked for support.

Accelerating genomic workflows using NVIDIA Parabricks

Open access

Abstract

BACKGROUND: As genome sequencing becomes better integrated into scientific research, government policy, and personalized medicine, the primary challenge for researchers is shifting from generating raw data to analyzing these vast datasets. Although much work has been done to reduce compute times using various configurations of traditional CPU computing infrastructures, Graphics Processing Units (GPUs) offer opportunities to accelerate genomic workflows by orders of magnitude. Here we benchmark one GPU-accelerated software suite called NVIDIA Parabricks on Amazon Web Services (AWS), Google Cloud Platform (GCP), and an NVIDIA DGX cluster. We benchmarked six variant calling pipelines, including two germline callers (HaplotypeCaller and DeepVariant) and four somatic callers (Mutect2, Muse, LoFreq, SomaticSniper). RESULTS: We achieved up to 65 × acceleration with germline variant callers, bringing HaplotypeCaller runtimes down from 36 h to 33 min on AWS, 35 min on GCP, and 24 min on the NVIDIA DGX. Somatic callers exhibited more variation between the number of GPUs and computing platforms. On cloud platforms, GPU-accelerated germline callers resulted in cost savings compared with CPU runs, whereas some somatic callers were more expensive than CPU runs because their GPU acceleration was not sufficient to overcome the increased GPU cost. CONCLUSIONS: Germline variant callers scaled well with the number of GPUs across platforms, whereas somatic variant callers exhibited more variation in the number of GPUs with the fastest runtimes, suggesting that, at least with the version of Parabricks used here, these workflows are less GPU optimized and require benchmarking on the platform of choice before being deployed at production scales. Our study demonstrates that GPUs can be used to greatly accelerate genomic workflows, thus bringing closer to grasp urgent societal advances in the areas of biosurveillance and personalized medicine.

Overview

CoChat Summary

TL;DR

This paper benchmarks NVIDIA Parabricks GPU-accelerated variant calling against CPU workflows on AWS, GCP, and an NVIDIA DGX cluster. Germline callers accelerated strongly and often reduced cost, while somatic callers varied by algorithm, GPU count, and platform and sometimes cost more than CPU runs.

Key findings

  • Germline variant callers achieved up to 65× acceleration: HaplotypeCaller runtimes fell from 36 h on CPU to 33 min on AWS, 35 min on GCP, and 24 min on the NVIDIA DGX.
  • On AWS, GPU acceleration reached up to 65.1× for HaplotypeCaller, 30.7× for DeepVariant, 56.8× for Mutect2, 7.7× for SomaticSniper, 18.9× for Muse, and 3.7× for LoFreq; on GCP, up to 65.8×, 26.5×, 29.3×, 7.0×, 21.8×, and 4.5× respectively.
  • Germline callers scaled well with GPU number, but somatic callers showed more variation; for example, Mutect2 and SomaticSniper were sometimes faster with 4 GPUs than 8 GPUs.
  • GPU runs produced cost savings for most germline workflows, but poorly optimized somatic workflows could cost more: LoFreq CPU runs cost less than $9/sample on both platforms but as much as $30 with GPUs, and SomaticSniper CPU runs cost less than $14.5/sample but as much as $75 on AWS with 8 GPUs.

Methods

  • Benchmarked six variant calling pipelines: two germline callers (HaplotypeCaller v4.2.0.0 and DeepVariant v1.1.0) and four somatic callers (Mutect2 v4.2.0.0, SomaticSniper v1.0.5.0, Muse v2.0, and LoFreq v2.1), using Parabricks v3.7.0-1 for GPU runs and Snakemake v6.6.1 CPU workflows as baselines.
  • Used HG002 from Genome in a Bottle, down-sampled to 30× coverage with Samtools v1.9 and aligned to GRCh38; somatic benchmarking used synthetic tumor data generated with SomatoSim v1.0.0, adding 198 SNPs at variant allele frequencies from 0.001 to 0.4.
  • Compared CPU and GPU configurations on AWS (c6i.8xlarge CPU; p3 V100 and p4 A100 GPU machines), GCP (n2-standard-32 CPU; a2-highgpu A100 GPU machines), and an NVIDIA DGX SuperPOD with up to 8 A100 GPUs, testing 2, 4, and 8 GPU configurations where possible.

Limitations

  • The paper states that somatic findings used synthetic somatic data based on sites from a real patient, and some findings could be artifacts of the somatic variant sampling design.
  • The paper states that AWS/GCP CPU runtime differences likely reflect processor generation differences, unless unaccounted-for factors also influenced variation; it also notes that comparisons between 2 and 4 A100 GPU machines were imprecise because AWS CPU limits could not be controlled for all workflows, biasing some comparisons toward AWS.
  • The paper states that somatic callers were less GPU optimized with the Parabricks version tested and require benchmarking on the platform of choice before production deployment; the text also makes evident that this study benchmarks speed and cost, not variant-calling accuracy, and that DGX Parabricks runs were limited to a single DGX A100 system with 8 GPUs.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1186/s12859-023-05292-2

Cites

26 of 26 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Comparison of execution times of variant calling algorithms on CPU and GPU environments between AWS and GCP. A 32 vCPU machine with the latest processors was used for CPU benchmarking on both cloud platforms. Here we show results for varying numbers of NVIDIA Tesla V100 GPUs running the Parabricks bioinformatics suite for AWS, and NVIDIA Tesla A100 GPUs for GCP
Fig. 1 Comparison of execution times of variant calling algorithms on CPU and GPU environments between AWS and GCP. A 32 vCPU machine with the latest processors was used for CPU benchmarking on both cloud platforms. Here we show results for varying numbers of NVIDIA Tesla V100 GPUs running the Parabricks bioinformatics suite for AWS, and NVIDIA Tesla A100 GPUs for GCPCC BY · PubMed Central
Fig. 2 GPU benchmarking results for NVIDIA Tesla GPUs. On GCP and the DGX results are shown for A100 GPUs, whereas AWS results are shown for the V100 GPU runs
Fig. 2 GPU benchmarking results for NVIDIA Tesla GPUs. On GCP and the DGX results are shown for A100 GPUs, whereas AWS results are shown for the V100 GPU runsCC BY · PubMed Central
Fig. 3 Comparison of AWS (V100 GPU machine) versus GCP GPU cost savings per variant caller. Percentage of total cost savings shows higher cost savings using GPUs in algorithms optimized for GPU-acceleration, but losses when algorithms are not well optimized
Fig. 3 Comparison of AWS (V100 GPU machine) versus GCP GPU cost savings per variant caller. Percentage of total cost savings shows higher cost savings using GPUs in algorithms optimized for GPU-acceleration, but losses when algorithms are not well optimizedCC BY · PubMed Central
Fig. 4 Comparison of AWS V100 versus GCP A100 GPU cost ratio per variant caller. Cost ratio is the ratio between cost per hour and fold speed-up. Cost per fold-speedup shows the benefit of harnessing GPU over CPU in select algorithms, while other algorithms are more cost-efficient with CPUs when using the version of Parabricks that we benchmarked
Fig. 4 Comparison of AWS V100 versus GCP A100 GPU cost ratio per variant caller. Cost ratio is the ratio between cost per hour and fold speed-up. Cost per fold-speedup shows the benefit of harnessing GPU over CPU in select algorithms, while other algorithms are more cost-efficient with CPUs when using the version of Parabricks that we benchmarkedCC BY · PubMed Central
Fig. 5 Comparison of runtimes between V100 and A100 GPU machines on AWS
Fig. 5 Comparison of runtimes between V100 and A100 GPU machines on AWSCC BY · PubMed Central

Verification

Claims the paper makes

7 Valid · 2 Not valid · 6 Not checked · 57 Own results and interpretations

18 more claims

Create a free account to see all 72 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 16
  • It may state two separate claims as one. 40

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

1 of 1

Citations

Cites

Show all 29 references

Research works printed but not linked: 2

  1. Van der Auwera GA, O’Connor BD. Genomics in the cloud: using Docker, GATK, and WDL in Terra. O’Reilly Media; 2020.
  2. Li H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. ArXiv Prepr. ArXiv13033997; 2013.

Cited by

See all 62

Metadata

Authors
O’Connell, K. A., Yosufzai, Z. B., Campbell, R. A., Lobb, C. J., Engelken, H. T., Gorrell, L. M., Carlson, T. B., Catana, J. J., Mikdadi, D., Bonazzi, V. R., & Klenk, J. A.
Affiliations
Deloitte (United States) (US)
Corresponding author
  • Bonazzi, Vivien R.
  • Klenk, Juergen A.
Venue
BMC Bioinformatics
Published
2023-05-31
Volume
24
Issue
1
Pages
221-221
ISSN
1471-2105 (ISSN-L)
OpenAlex
W4378907379
Cited by
62
References
29
Citations by year
2026: 25 · 2025: 23 · 2024: 11 · 2023: 2
FWCI
12.49
Type
journal article
Language
en
Open access
Licence
Locations
  • BMC Bioinformatics (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
  • Social Sciences › Decision Sciences › Information Systems and Management › Scientific Computing and Data Management
Keywords
GPU acceleration, variant calling, HaplotypeCaller, DeepVariant, Mutect2, LoFreq, Amazon Web Services, Google Cloud Platform
MeSH terms
Computer Graphics, Software, Genomics, Workflow

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Accelerating genomic workflows using NVIDIA Parabricks. https://app.cochat.ai/papers/10.1186/s12859-023-05292-2
The paper
O’Connell, K. A., Yosufzai, Z. B., Campbell, R. A., Lobb, C. J., Engelken, H. T., Gorrell, L. M., Carlson, T. B., Catana, J. J., Mikdadi, D., Bonazzi, V. R., & Klenk, J. A. (2023). Accelerating genomic workflows using NVIDIA Parabricks. BMC Bioinformatics. https://doi.org/10.1186/s12859-023-05292-2

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.