Skip to content

3.12.0 - Njoatsosvágge

Latest

Choose a tag to compare

@FriederikeHanssen FriederikeHanssen released this 11 Nov 12:17
786a11b

3.12.0 - Njoatsosvágge

Njoatsosvágge is a long valley crossing the northern part of Sarek National Park.

Added

Addition of:

  • Parabricks MutectCaller with --tools parabricks_mutectcaller, a GPU-accelerated somatic variant caller for tumor-only and tumor/normal samples. Its calls are filtered with GATK FilterMutectCalls, including read-orientation priors and contamination estimation.
  • --force_call_alleles and --force_call_alleles_tbi to force-call alleles from a VCF with Mutect2 and Parabricks MutectCaller, based on work by @jchchiu.
  • --contamination_resource and --contamination_resource_tbi to give GATK GetPileupSummaries a set of common biallelic SNPs instead of --germline_resource.
  • --mosdepth_fast_mode to control Mosdepth fast mode for whole-genome coverage. Fast mode stays enabled by default.

Bug fixes

  • Fix --intervals handling: the BED is now sorted in reference order, which fixes CalculateContamination failing with Invalid interval. The run now fails early for contigs missing from the reference, start > end, and intervals past the contig end, before any process starts when --fasta_fai is provided. BED files with track, browser or # header lines no longer fail the run.
  • Fail when --no_intervals is combined with --intervals, which it would ignore, and restore --intervals false to generate intervals from the FASTA.
  • Fix fgbio crashes during UMI consensus calling with bwa-mem2 by running samtools fixmate first.
  • Fix the DeepSomatic VCF merge failing when an interval yields no records, such as *_alt contigs.
  • Fix Join mismatch in post-variant-calling for --tools tiddit with --filter_vcfs, --normalize_vcfs or --concatenate_vcfs.
  • Fix MSIsensorPro analysing only the first tumor/normal pair when --msisensorpro_scan is provided, and publish its _all and _unstable outputs.
  • Fix csv/variantcalled.csv pointing at a directory instead of the published VCFs, and consensus VCFs intermittently being listed in the wrong folder.
  • Pass the Mutect2 and Parabricks MutectCaller stats file to FilterMutectCalls.
  • Fix validation for --step annotate.

Improvements

  • Update Parabricks to 4.8.0 and FastQC to 0.13.0.
  • Add bedtools 2.31.1 for interval sorting.
  • Mutect2 and Parabricks MutectCaller share one GATK filtering subworkflow.
  • Clarify the documentation for annotation caches and software versions, the <patient>_<sample> read group name required for BAM/CRAM input to somatic callers, and MSI analysis.
  • Update the pipeline template to nf-core/tools 4.2.1.

Important changes

  • The process MUTECT2_PAIRED is now MUTECT2. Custom configs using withName: 'MUTECT2_PAIRED' need to select MUTECT2 instead.
  • Mutect2 calls are now always filtered with FilterMutectCalls, also without --germline_resource. Downstream files of such runs are named after *.mutect2.filtered.vcf.gz.
  • With --joint_mutect2, tumors of one patient with different contamination values in the samplesheet no longer split the patient into several joint calls.
  • CNVKit target and antitarget BED names now carry the .sorted suffix of the sorted intervals file.
  • csv/variantcalled.csv now points at the published filtered, normalized, concatenated and Varlociraptor VCFs.

New parameters

  • --contamination_resource
  • --contamination_resource_tbi
  • --force_call_alleles
  • --force_call_alleles_tbi
  • --mosdepth_fast_mode

New Contributors

Full Changelog: 3.11.0...3.12.0