4.1.4.1

Download release: gatk-4.1.4.1.zip
Docker image: https://hub.docker.com/r/broadinstitute/gatk/

Highlights of the 4.1.4.1 release:

• New experimental HaplotypeCaller assembly mode which improves phasing, reduces false positives, improves calling at complex sites, and has 15-20% speedup vs the current assembler. It is enabled with option --linked-de-bruijn-graph. This mode is still experimental and not recommended for production use yet.
• IndexFeatureFile improvements:
  • now cloud enabled
  • changed controversial F argument to I instead.
• Bug fixes and improvements in GenomicsDB, Mutect2, variant annotation, and more!

Full list of changes:

• New Tools

  • PrintReadsHeader: a new tool to print a BAM/SAM/CRAM header to a file (#6153)

• HaplotypeCaller

  • Experimental prototype of JunctionTree based haplotype finding. (#6034) #5925
  • Fix a genotyping bug were reference/alt likelihoods were capped differently. (#6196)

• Mutect2

  • Mutect2 now warns but does not fail when three or more reads have the same name. (#6240)
  • Fixed the random seed at the beginning of FilterMutectCalls (#6208)
  • GetSampleName and GetPileupSummaries in the M2 pipeline are no longer beta. (#6215)
  • Increase number of iterations in CalculateContamination to 30. (#6282)
  • Handled an edge case with high scatter count in M2 WDL. (#6216)
  • Use ArgumentsBuilder in M2 tests. (#6219)

• Joint Calling

  • Allele-specific VQSR convergence fix. (#6262)
  • Fix to Allele Fraction annotation bug in multisample vcfs. (#6251)
  • Fix RAW_MQ header inconsistencies after reblocking. (#6276)
  • Mark SNP/indel mode argument in GatherTranches as required so tranches are named properly. (#6273)

• CNV Calling

  • Fixed model parameter assignment typo in gCNV ploidy model (#6285)
  • Added docker option to the gcnv QC tasks. (#6185)
  • Added epsilons to overdispersion in gCNV models to avoid NaNs. (#6245) #4824 #6226 #6227
  • Assert that ELBO did not become NaN during each step of inference of gCNV. (#6186)
  • Added ability to override THEANO_FLAGS environment variable in gCNV tools. (#6244) #6235
  • Removed erroneous short argument names in R scripts for CNV plotting. (#6197)

• GenomicsDB

  • Allow GATK to configure annotation processing instead of hardcoding values in GenomicsDB GDB-39
  • High ploidy sites with many genotypes no longer causes an overflow error. GDB-54
  • Add missing libcurl in the native GenomicsDB library. #6122 GDB-66
  • No longer crashes when vcfbufferstream from htslib appears to be invalid. GDB-67
  • Propagated native GenomicsDB exceptions as java IOExceptions. GDB-68
  • Fix issue when using vid protobuf interface and there is more than 1 config. GDB-70
  • Cleanup GenomicsDB vid combine protobuf mapping overrides. #6190

• Miscellaneous Changes

  • Cloud-enable IndexFeatureFile and change input arg name from -F to -I. (#6246) #6161
  • WDL to run ReadsPipelineSpark on a multicore machine (#6213)
  • Replace TwoPassReadWalker with more general MultiplePassReadWalker. (#6154)
  • Abolish unfilled likelihoods and revamp VariantAnnotator. (#6172)
  • Improve exception message in ValidateVariants. (#6076)
  • Fix Syntax Warning when running GATK with python 3.8 (#6231)

• Developer / Testing

  • Report errors logs in github comment (#6247) 6234
  • Add .java-version to gitignore to support jenv users. (#6232)
  • Restart test JVM after every 100 test classes do reduce out of memory failures. (#6093)
  • Running the cloud tests on java 11 on travis. (#6210)

• Documentation

  • Clarify definition of PGT in VCF header (#6221)
  • docs for paired reads in Mutect2 somatic genotyping (#6264)
  • Fix some typos in the allele subsetting code. (#6265)

• Dependencies

  • Update picard to 2.21.2 (#6253)
  • Update disq to 0.3.4 (#6252)
  • update htsjdk to 2.21.0 (#6250)
  • Update to Genomicsdb 1.1.2.2 (#6206) (#6188)