Terrill Dicki
Oct 14, 2025 09:45
NVIDIA Parabricks v4.6 improves genomic analysis with GPU-accelerated tools, including support for DeepVariant and DeepSomatic 1.9, enhancing variant calling accuracy and speed.
NVIDIA has unveiled the latest update to its genomics software suite, Parabricks v4.6, designed to offer enhanced capabilities for genomic analysis through GPU acceleration. The new version brings notable improvements, particularly in variant calling accuracy, a critical step in genomic analysis, according to [NVIDIA](https://developer.nvidia.com/blog/improve-variant-calling-accuracy-with-nvidia-parabricks/).
Key Features of Parabricks v4.6
The latest release includes support for Google’s DeepVariant and DeepSomatic 1.9, with a pangenome-aware mode that improves accuracy across diverse genetic populations. This mode significantly reduces errors, providing a more comprehensive understanding of genetic variations.
Parabricks v4.6 also introduces support for long-read and whole exome sequencing in DeepSomatic, along with STAR quantMode for gene-level read counts. These enhancements allow researchers to leverage the speed of GPU acceleration to uncover biological insights faster.
Advancements in Variant Calling
Variant calling identifies differences between sample and reference genomes, offering insights into diseases and potential treatments. DeepVariant and DeepSomatic, developed by Google, are deep learning-based tools that support this process. They offer significant improvements in identifying both inherited and somatic variants.
The integration of pangenome-aware DeepVariant in Parabricks v4.6 reduces reference bias, improving the accuracy of variant detection across diverse populations. This is crucial for understanding genetic diversity and its impact on health.
Giraffe and DeepVariant Integration
The software tool Giraffe, developed by the University of California, Santa Cruz, enhances read alignment to pangenome graphs. It improves variant-calling accuracy by mapping genomic sequences to a reference pangenome, rather than a traditional linear reference.
Combining Giraffe with the pangenome-aware mode in DeepVariant, now available in Parabricks, enhances accuracy and speed, providing a 14x speedup compared to CPU-only solutions.
STAR Improvements
Parabricks v4.6 also includes improvements to STAR, a tool for RNA-sequencing alignment. The latest release introduces quantMode GeneCounts, significantly accelerating the process with nearly 8x faster speedups on NVIDIA RTX PRO 6000 GPUs.
These advancements make Parabricks v4.6 a powerful tool for genomic researchers, enabling faster and more accurate analysis across a broad range of applications.
Image source: Shutterstock
Credit: Source link
