NVIDIA Clara™ Parabricks is a GPU-accelerated computational genomics application framework. It delivers high-throughput, high-accuracy DNA and RNA sequencing.


Accelerated Applications For Genomic Analysis

From FASTQ to Binary Alignment Map (BAM) to VCF, NVIDIA accelerates genomics analysis for somatic, germline and structural variant workflows, as well as improving sequencer speeds and consensus read base calling accuracy.

Clara Parabricks can be run on-premise or in the cloud

Robust HPC conducts training workshop that guides researchers on how to run their genomic analysis workloads with Clara Parabricks.

For R&D institutions that looking for a HPC cluster solution, Robust HPC provides the following:

On-Premise Solutions

Running on one NVIDIA DGX™ A100, Clara Parabricks analyzes up to 25,000 whole genomes per year (30X coverage) with Genomic Analysis Toolkit (GATK) from Broad Institute.

Throughput scales linearly beyond a single DGX A100.

Learn more about DGX A100 or DGX Station A100. Interested in DGX H100? Get in touch with us here.

In the Cloud

In the cloud, time is money.

Choose from a range of GPU instances capable of analyzing genomes in a matter of minutes. For easy deployment, we have managed service to help you get started quickly.

Contact us to learn more about our Managed HPC/AI Cloud.


How Clara Parabricks Is Making a Real-World Impact

Regeneron Genetics Center

See how Regeneron is using DeepVariant with Clara Parabricks to generate scalable, high-quality data that can be reproduced across the genomics community.


National Biobank of Thailand (NBT)

NBT reduced its whole genome sequencing data processing by four months and its individual user processing time from over 30 hours to just 1-2 hours with NVIDIA DGX A100 and Clara Parabricks.


Translational Genomics Research Institute (TGen)

With NVIDIA Clara Parabricks and NVIDIA V100 Tensor Core GPUs, TGen sped up genomic sequencing analysis by 6X to help determine personalized treatments for children with rare neurological disorders.


Washington University in Saint Louis (WUSTL)

Integrating GPU-based steps with NVIDIA Clara Parabricks enabled Tychele Turner of WUSTL to drastically reduce the runtime for de novo variant calling.



The Parabricks Advantage

On a single server, the software can process more than 60 whole genomes per day

No need to sacrifice accuracy to reduce turnaround time. A suite of germline and somatic callers is available.

Create customized, accelerated pipelines by configuring over 33 tools for any specific pipeline or application need.

Reduce computing costs up to 50 percent by running 30–60X faster than CPU-only solutions.


The following standalone tools can be used with the NVIDIA Clara Parabricks Pipelines software. Please click on the tool names for tool specific options.

Tool Details
FQ2BAM Align using bwa-mem, co-ordinate sort and mark duplicates, optionally you can run bqsr.
BQSR Generate a BQSR report on a bam file.
APPLYBQSR Apply a BQSR report on a bam file to generate new bam file.
HAPLOTYPECALLER GPU-HaplotypeCaller for calling germline variants.
SAMTOOLS MPILEUP Accelerated samtools mpileup to generate pileup from a bam file.
BCFTOOLS MPILEUP Accelerated bcftools mpileup to generate pileup from a bam file.
BCFTOOLS CALL Accelerated bcftools call variant caller.
SOMATICSNIPER Accelerated Somaticsniper for tumor-normal analysis.
SOMATICSNIPER WORKFLOW Somaticsniper workflow to generate VCF from BAM input files.
MANTA Structural variant (SV) and indel caller from mapped paired-end sequencing reads.
STRELKA SNP and indel caller from mapped paired-end sequencing reads.
STRELKA WORKFLOW Strelka workflow to generate VCF from BAM/CRAM input files.
MUTECTCALLER GPU-Mutect2 for tumor-normal analysis.
DEEPVARIANT GPU-DeepVariant for calling germline variants.
CNVKIT Accelerated copy number variant caller.
BAMMETRICS Collect WGS Metrics on a bam file.
COLLECT MULTIPLE METRICS Collect multiple classes of metrics for a bam file.
DBSNP Annotate variants based on a variant database.
CNNSCOREVARIANTS Filter variants using Convolutional Neural Network.
VQSR Build a recalibration model to score variant quality and apply a score cutoff to filter variants.
TRIO COMBINE GVCF Combine gVCF of 2 or 3 samples.
GLNEXUS Scalable gVCF merging and joint variant calling for population sequencing projects.
INDEX GVCF Index a VCF/gVCF file.
RNA FQ2BAM Mapping RNA reads to a reference, using a two-pass mode to get better alignments around novel splice junctions.
STAR-FUSION Uses the STAR aligner to identify candidate fusion transcripts.

NVIDIA Clara Parabricks Datasheet

Want to know more about the technology inside NVIDIA Clara Parabricks?


NVIDIA Clara Paravricks


Parallel File System for AI/Big Data




× Live Chat