Fix an edge case bug in PalindromeArtifactReadTransformer

[davidbenjamin]

Closes #5036.

@meganshand This fixes the bug you found.

[codecov-io]

Codecov Report

Merging #5038 into master will increase coverage by 0.088%.
The diff coverage is 66.667%.

@@              Coverage Diff               @@
##             master     #5038       +/-   ##
==============================================
+ Coverage     86.38%   86.469%   +0.088%     
- Complexity    28640     29108      +468     
==============================================
  Files          1782      1782               
  Lines        132603    134161     +1558     
  Branches      14761     15067      +306     
==============================================
+ Hits         114543    116007     +1464     
- Misses        12740     12805       +65     
- Partials       5320      5349       +29

Impacted Files	Coverage Δ	Complexity Δ
...itute/hellbender/tools/walkers/mutect/Mutect2.java	89.286% <0%> (-3.571%)	18 <1> (ø)
...formers/PalindromeArtifactClipReadTransformer.java	97.143% <80%> (-2.857%)	21 <0> (+4)
...iscovery/TestUtilsForAssemblyBasedSVDiscovery.java	88.889% <0%> (-6.633%)	12% <0%> (-1%)
...dataSources/vcf/VcfFuncotationFactoryUnitTest.java	97.656% <0%> (-0.43%)	40% <0%> (+11%)
.../AssemblyContigAlignmentsConfigPickerUnitTest.java	99.415% <0%> (-0.19%)	34% <0%> (+14%)
...utils/variant/GATKVariantContextUtilsUnitTest.java	87.194% <0%> (-0.088%)	316% <0%> (+156%)
...ct/CreateSomaticPanelOfNormalsIntegrationTest.java	100% <0%> (ø)	6% <0%> (+3%)	⬆️
.../validation/RemoveNearbyIndelsIntegrationTest.java	100% <0%> (ø)	5% <0%> (+2%)	⬆️
...walkers/validation/ConcordanceIntegrationTest.java	100% <0%> (ø)	10% <0%> (+4%)	⬆️
...dation/AnnotateVcfWithBamDepthIntegrationTest.java	100% <0%> (ø)	12% <0%> (+5%)	⬆️
... and 10 more

[meganshand]

Thank you! In a perfect world it would be great to have a new test, but I did run this branch on my data and it works. 👍

[davidbenjamin]

@meganshand Do you have a public mitochondrial bam that I could use for a test? If not, what if I took just the reads at the beginning of the contig, excluding any variants, and anonymized the header? That would be enough.

Actually, if you do have an entire mitochondrial bam that's public, I wouldn't mind adding a mitochondrial integration test to the repo. . .

[meganshand]

@davidbenjamin I do have a public bam! This is just chrM for NA12878: /humgen/gsa-hpprojects/dev/mshand/SpecOps/Mitochondria/NA12878/NA12878_chrMOnly.bam

The coverage should be quite high, so if it's too big I can downsample it if you want.

And my mitochondria only reference would probably also be useful: /humgen/gsa-hpprojects/dev/mshand/SpecOps/Mitochondria/MitochondriaOnlyFastas/Homo_sapiens_assembly38.mt_only.fasta (the fai and dict are there too).

[davidbenjamin]

@meganshand I will include testing for this bug as part of an overall mitochondria integration test using your NA12878 bam in a separate PR.