4.4.0-1 Release Notes
Highlights:
We are introducing a new tool, the giraffe pangenome graph-based aligner for short reads.
Read the section "How should I use giraffe in Parabricks?" in the tool reference documentation for advice on getting started and best practices.
Unified 'multi-architecture' Docker containers for both Grace Hopper and AMD64 architectures.
With Parabricks 4.4.0 we are releasing a GPU-accelerated giraffe tool. Giraffe is a graph-based short-read mapping tool that efficiently handles genetic diversity and structural variations across populations. VG Giraffe was originally developed by Dr. Benedict Paten's lab at the University of California, Santa Cruz (UCSC).
Tool Updates
Multi-architecture container:
Instead of pulling different containers for different architectures, you now need only pull one container. The
pullcommand will determine which architecture is needed.
BAM reader and writer:
Support for reading and writing BAM records with more than 65535 CIGAR operations.
Support for
--run-partitionin force-calling mode (--htvc-alleles).
Improvements to both the progress meter and error logging messages.
Support for additional arguments with the fq2bam sorting step:
--gpusort,--gpuwrite,--gpuwrite-deflate-algo,--use-gds, and more.
Fixed a crash which occurred if the HG tag value equaled -1 when passing
--mutect-bam-output.
Fixed a crash which occurred if the HG tag value equaled -1 when passing
--htvc-bam-output.
Fixed a crash in the wrapper when passing
--out-sam-unmapped Within_KeepPairs.
Fixed a deadlock issue that occurred with certain datasets during GPU-accelerated CRAM write operation. (
--gpuwrite).
Fixed a deadlock issue that occurred with certain datasets during GPU-accelerated CRAM write operation. (
--gpuwrite).Fixed an issue that caused some large reads to not be processed by the fastq reader.
Fixed a crash that occurred in certain edge cases, such as with chromosome MAP2K3_chr17_22578583_22605165, related to handling supported reads.
For further information see the Parabricks datasheet.