gatk-protected-1.0.0.0-alpha1.2.7.2

This release is the same as gatk-protected-1.0.0.0-alpha1.2.6 with b51299f615f06ea5263119f2a94485bd829be0c2, a315b95b4c1a4a8db4c5835dcdf99e239faff812, and 581311359c9c5e41b272b8386ae0e2796f9563c2 applied. All of these commits relate to fixes for FilterByOrientationBias.

gatk-protected-1.0.0.0-alpha1.2.6

Mutect 2:

• Using new SplitIntervals tool. WDL updated.

Other:

• New GATK being used (4.alpha.2-234-gc3a44c9-20170427.144324-1).
• New pairHMM interface support.

gatk-protected-1.0.0.0-alpha1.2.5

Mutect 2:

• First cut of WDL for running multiple and single pairs
• Tumor-only supported in WDL
• Base quality and mapping quality filters added and associated annotations
• Fixed STR annotation, so that filter (str_contraction) has been activated
• Additional new annotations added: median fragment length for reads supporting ref and alt ; median distance from end of read
• Automated testing of WDL
• Oncotator and Orientation Bias Filter tasks added and are optional. Note that the oncotator task requires a docker image. Picard jar is required for the Orientation Bias Filter (for now).
• Fix to Orientation Bias Filter to accept empty VCFs
• Orientation Bias Filter now puts filter annotation in the INFO field and in the FORMAT field

CNV/ACNV (v1):

• First cut of improved WDL
• Automated testing of WDL upgraded
• CNLoH caller has been renamed to BalancedSegmentCaller. The loss-of-heterozygosity calls should be considered experimental and are very fragile.

Known issue:

• Docker image is bigger

gatk-protected-1.0.0.0-alpha1.2.4

New features/bugfixes

-CNV plotting tools now determine plotting regions from a sequence dictionary.
-New CLIs for HMM-based segmentation (PerformJointSegmentation, PerformCopyRatioSegmentation, PerformAlleleFractionSegmentation) are still under active development and should not be used.

gatk-protected-1.0.0.0-alpha1.2.3

New features/bugfixes

-Added WDL scripts for running somatic CNV case/PoN workflows
-Added tool for genotyping sex from coverage data (TargetCoverageSexGenotyper)
-Added ACNV tools CreateAllelicPanelOfNormals and CalculatePulldownPhasePosteriors
-Somatic CNV/ACNV tools now take TSV target files as input, rather than BED files (the latter can be converted into the former using the tool ConvertBedToTargetFile)
-Fixed various workflow and plotting issues
-Removed dependency on HDF5-Java JNI Libraries

gatk-protected-1.0.0.0-alpha1.2.2

New features/bugfixes

-Fixed a compiler warning that prevented the previous version from building.

Note: This release was tagged for use by a specific research group at the Broad Institute and so not all of the tools and workflows are in a working state (e.g., GATK ACNV). Other users should use alpha1.2.1 if they are able to build it successfully and alpha1.2 otherwise.

gatk-protected-1.0.0.0-alpha1.2.1

New features/bugfixes

• New option -numIterSimPerFit 0 to disable refitting between sim-seg iterations. See #503 for more information. When used, this speeds up ACNV, since it will do fewer iterations of MCMC. Results do not seem to be impacted in any appreciable way.
• Fixed ACS conversion bug that was raising the segment mean to the power of two, twice.
• Made ACNV Plotting a bit more robust to code changes in other tools.
• Made the x-axes in ACNV Plotting round the genomic position a bit more sensibly.

gatk-protected-1.0.0.0-alpha1.2

New features/bugfixes

• New het pulldown tool introduced: GetBayesianHetCoverage This can produce a het pulldown with matched tumor-normal pair or with a tumor sample only
• Due to new het pulldown tool mentioned above, ACNV can be run without a matched normal sample.
• Better performance in ACNV
• More documentation
• Improved plotting for ACNV
• Plotting of each segment in ACNV
• Somatic WGS support. This has not been evaluated yet. Users should use the SparkGenomeReadCounts tool for coverage collection of WGS samples. See Performance Notes at the end of this post.
• Germline CNV toolchain. Performance will be improved soon, so users should be warned that this code will be changed in the near future.
• CLI tool for GC correction in WGS or capture data.
• CNLoH caller implemented using ACNV results
• Balanced segment calling using ACNV results is part of the CNLoH caller. In other words, produces calls as to whether a segment has a MAF of 0.5.
• Added new conversion as part of CNLoH caller that will produce files that can be ingested by TITAN.
• Moved the conversion of ACNV to CGA AllelicCapSeg file format to the CNLoH Caller and integrated the balanced caller with this process.
• Updated WDL for tumor normal case sample workflow.

Incomplete

• Somatic WGS has not been evaluated yet.
• Still using a development version of the GATK4
• GATK ACNV (GetHetCoverage and AllelicCNV) on WGS samples takes ~36 hours. This will be improved in next milestone. Running GetBayesianHetCoverage is unlikely to improve this runtime.

Performance Notes (WGS)

All numbers were using a bin size of 3k bases.

• SparkGenomeReadCounts on WGS data:
• SparkGenomeReadCounts: ~5.5 hours on Broad NFS and ~5 minutes on hdfs with 120 core spark cluster
• CalculateTargetCoverage does not leverage Spark (yet), so no improvement can be gained there. We have seen runtimes well over 12 hours.
• PoN creation steps (not including coverage collection) took under 1.25 hrs for 142 samples on Broad NFS
• GATK CNV case workflow on WGS sample takes ~6 hrs

Notes

• The previous GetHetCoverage tool has not been removed. By default, we will not be using it in workflow releases.
• GC correction has not yet been included in the workflows.

gatk-protected-1.0.0.0-alpha1.1

Many improvements including:

• Better performance in ACNV
• More documentation
• Percentiles in ACNV output distributions

Incomplete:

• Germline CNV tools are not evaluated and should be considered incomplete
• CN LOH caller not implemented
• CN LOH plotting not implemented
• Still using a development version of the GATK4

gatk-protected-1.0.0.0-alpha1-rc1

• Introducing ACNV, which integrates heterozygous SNP (germline) information. See documentation for details.
• Fast WGS coverage collection
• Exposing many more parameters in segmentation
• Improved documentation
• Added Dockerfile
• Generic MCMC library
• Removed sample name parameter

Incomplete:

• Plotting ACNV results still unfinished
• CN LOH caller not implemented
• CN LOH plotting not implemented
• Still using a development version of the GATK4