> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.nvidia.com/clara/parabricks/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/clara/parabricks/_mcp/server. # 4.7.1-1 Release Notes **Highlights:** * Performance improvements in [deepvariant](/tool-reference/tools/deepvariant) with small model enabled. **The small model is now enabled by default, similar to Google DeepVariant.** * Added Roche SBX mode to [pangenome\_aware\_deepvariant](/tool-reference/tools/pangenome-aware-deepvariant) with the new `--sbx` flag for GPU-accelerated variant calling on Roche SBX-D and SBX-Fast sequencing data. * Performance improvements for [fq2bam](/tool-reference/tools/fq-2-bam-bwa-mem-gatk) and [fq2bam\_meth](/tool-reference/tools/fq-2-bam-meth). Added optional support for processing reads up to 1024 bp on GPU. * Performance improvements for single-end alignment with [giraffe](/tool-reference/tools/giraffe-vg-giraffe-gatk). * Added support for reading CRAM v3.1 files. * Updated version of [minimap2](/tool-reference/tools/minimap-2) to v2.31. * GPU-accelerated chaining in [minimap2](/tool-reference/tools/minimap-2) with the new `--chaining-on-gpu` flag. ## Improvements All tools that rely on Parabricks GPU-accelerated sort now have improved error handling. Running `pbrun --help` or `pbrun -h` now works without GPUs visible. ### Tool Updates [deepvariant](/tool-reference/tools/deepvariant), [deepvariant\_germline](/tool-reference/tools/deepvariant-germline), [pacbio\_germline](/tool-reference/tools/pacbio-germline), and [ont\_germline](/tool-reference/tools/ont-germline): * Improved performance with the small model across WGS, PacBio, and ONT modes. * The small model is now enabled by default. Use `--disable-small-model` to disable it. [pangenome\_aware\_deepvariant](/tool-reference/tools/pangenome-aware-deepvariant): * Added Roche SBX mode with `--sbx` for GPU-accelerated variant calling on Roche SBX-D and SBX-Fast short-read data. [giraffe](/tool-reference/tools/giraffe-vg-giraffe-gatk): * Improved performance of the GPU seed-extension (GBWT) stage. * Improved performance when producing final read alignments. * Improved performance of the GPU Dozeu alignment kernel. * Improved initialization time for single-end alignment. * Moved minimizer payload decoding to the GPU, reducing CPU load. * Additional performance improvements and reduced host-side memory usage during alignment. [minimap2](/tool-reference/tools/minimap-2): * Updated version of minimap2 to v2.31. * Added `--chaining-on-gpu` flag to use GPUs to accelerate chaining, a compute-intensive step in minimap2 alignment, reducing CPU load. * Improved performance of GPU-accelerated spliced alignment. [fq2bam](/tool-reference/tools/fq-2-bam-bwa-mem-gatk) and [fq2bam\_meth](/tool-reference/tools/fq-2-bam-meth): * Improved performance for both single-end and paired-end alignment. * Added partition mode (`--run-partition`), which runs one process per partition to more efficiently split work across multiple GPUs. The `--bwa-gpu-num-per-partition` option controls the number of GPUs each partition drives (default: 2). * Increased the maximum read size supported for processing on GPU to 1024 bp. The maximum read size to process on GPU can be controlled via `--max-read-size-gpu`, where valid values are 512 or 1024 bp and the default remains 512 bp. In either mode, larger reads will continue to go to CPU recovery. [fq2bam\_meth](/tool-reference/tools/fq-2-bam-meth): * Updated the default CPU threading configuration to values that are more optimal for end-to-end runtime: `--bwa-primary-cpus` now defaults to `16 * ` and `--bwa-cpu-thread-pool` now defaults to `1`. Both options can still be set manually to override the defaults. [rna\_fq2bam](/tool-reference/tools/rna-fq-2-bam): * Performance improvement through code optimization. * Added throughput metrics. [haplotypecaller](/tool-reference/tools/haplotypecaller) and [mutectcaller](/tool-reference/tools/mutectcaller): * Added multi-stream support with the `--num-streams-per-gpu` option, which controls the number of streams to use per GPU (default: 1). [bamsort](/tool-reference/tools/bamsort), [deepsomatic](/tool-reference/tools/deepsomatic), [deepvariant](/tool-reference/tools/deepvariant), [haplotypecaller](/tool-reference/tools/haplotypecaller), [minimap2](/tool-reference/tools/minimap-2), [mutectcaller](/tool-reference/tools/mutectcaller), and [pangenome\_aware\_deepvariant](/tool-reference/tools/pangenome-aware-deepvariant): * Added support for reading CRAM v3.1 file. ## Bug Fixes [deepvariant](/tool-reference/tools/deepvariant), [deepsomatic](/tool-reference/tools/deepsomatic), and [pangenome\_aware\_deepvariant](/tool-reference/tools/pangenome-aware-deepvariant): * Updated the logic used to determine valid mapped mates for a read, now mirroring Google DeepVariant. * Added a better error message when header information is missing in the supplied input sample. * Fixed a bug causing print statements to appear in incorrect order. [pangenome\_aware\_deepvariant](/tool-reference/tools/pangenome-aware-deepvariant): * Properly parse user-supplied pangenome sample and pangenome reference names. [deepvariant](/tool-reference/tools/deepvariant), [deepsomatic](/tool-reference/tools/deepsomatic), [pangenome\_aware\_deepvariant](/tool-reference/tools/pangenome-aware-deepvariant), [haplotypecaller](/tool-reference/tools/haplotypecaller) and [mutectcaller](/tool-reference/tools/mutectcaller): * Fixed invalid `.tbi` index generation in certain edge cases. * Fixed an issue where a `.tbi` index could appear older than its corresponding data file. * Fixed a BGZF writer crash during `vcf.gz` writing on records larger than 64 KB. [haplotypecaller](/tool-reference/tools/haplotypecaller) and [mutectcaller](/tool-reference/tools/mutectcaller): * Fixed a potential race condition and also ensure bit-identical results across architectures. [mutectcaller](/tool-reference/tools/mutectcaller): * Fixed an incorrect error message for invalid tumor/normal sample names. [giraffe](/tool-reference/tools/giraffe-vg-giraffe-gatk): * Fixed a possible crash when processing batches that contain only short reads (below 29 bp) that produce no minimizers. * Fixed two potential out-of-bounds writes in the Dozeu alignment kernel. [fq2bam](/tool-reference/tools/fq-2-bam-bwa-mem-gatk) and [fq2bam\_meth](/tool-reference/tools/fq-2-bam-meth): * Fixed several GPU alignment edge cases involving empty chains, short-seed scoring, large-read scratch data, and reads with many seeds. * Fixed paired-end long-read CPU recovery mode to allow reads longer than `--max-read-size-gpu` to be processed on CPU. * Fixed a case that could result in a crash if all reads in a batch were below BWA-MEM's minimum seed length, which by default is 19 bp. [minimap2](/tool-reference/tools/minimap-2): * Fixed an out of bounds access caused by reads that skipped alignment in the `splice:sr` preset. * Fixed a bug that could cause junction values to not be used in splice-aware alignments. [fq2bam](/tool-reference/tools/fq-2-bam-bwa-mem-gatk), [fq2bam\_meth](/tool-reference/tools/fq-2-bam-meth), [minimap2](/tool-reference/tools/minimap-2), [rna\_fq2bam](/tool-reference/tools/rna-fq-2-bam), [bamsort](/tool-reference/tools/bamsort), and associated pipelines: * Fixed a bug in the GPU CRAM writer that caused `samtools index` to fail on the GPU generated CRAM file. For further information refer to the [Parabricks datasheet](https://resources.nvidia.com/en-us-hc-genomics/healthcare-genomics).