Bugs & Feature Requests
Known issues and work in progress


This page lists the issues tagged "bug" in the GATK4 development repository. If you have reported a bug and cannot find it in this list, let us know in the thread where you reported it. Note that we are no longer listing or processing issues in older versions of GATK (3.x and prior).


Showing open



MultiVariantWalkerGroupedOnStart doesn't handle reads correctly
GATK4 Funcotator generated annotation differs from dbSNP
HaplotypeCaller emitting a MNP in GVCF mode when --max-mnp-distance is 0
Add better error message when Java is not installed to the launch script
`-ERC` marked “beta-feature” in haplotypecaller 4.1.0.0 and later
Funcotator datasource configs should not honor spaces (and some other special characters) in the datasource name nor version.
Strange behavior in ParallelCopyGCSDirectoryIntoHDFSSpark with NIO ver > 0.66
Warnings should include info about their source
SortSam ShardedInput tests failing.
BaseRecalibratorSpark fails with "java.lang.IllegalArgumentException: Table1 1,3 not equal to 88,3"
Problem with BaseRecalibrator: java.lang.IllegalArgumentException: fromIndex(64) > toIndex(62)
Intel deflater + long reads data = intermittent corrupt bams
GenotypeGVCFs not outputting some variants it should with -stand-call-conf
Funcotator reproducibly crashes on specific WES VCF record produced by GATK4.1.0 (java 1.8.0_45)
Wrong center locations in BandPassActivityProfile
AnalyzeCovariates produces invalid csv when readgroups contain commas, resulting in a crash
GenotypeGvcfs mangles the input phase
HaplotypeCallerSpark crashes:8 parallel GATk jobs (8 samples) using 5 Spark cpus on each in a node with 40 cpus
HaplotypeCaller emitting wrong calls when using --max-mnp-distance
Error in smith waterman native library
Funcotator returns a nullpointer exception when trying to output compressed VCF
ConcurrentModificationException in ReadsPipelineSparkIntegrationTest
Variants with spanning deletions lack phase set information
ReferenceConfidenceModel using the wrong break condition in calcNIndelInformativeReads()
Treat "NaN" and "nan" as equivalents (VariantFiltration)
ReadPosRankSumTest doesn't work for two deletions with one base in between
Incorrect GQ Score = 0
Bug in HaplotypeCaller: lack of consistency in the overlap criteria applied to reads to be considered for PL and AD/DP annotations.
README: R setup for running tests may need OS X update
GermlineCNVCaller: Odd behavior of the log dumped ELBO values vs the real tracked ELBO values in tracker-tar