HaplotypeCaller GGA mode crashes with certain spanning deletions

[gmagoon]

Using HaplotypeCaller with GENOTYPE_GIVEN_ALLELES ("GGA") mode, I came across a couple of cases that crashed, and I traced them to spanning deletions (of the type considered in #4963).

The first case involved the following spanning deletion in the --alleles input:

22	16137300	rs567136176	TAG	T
22	16137302	rs573978809	G	C

and it crashed with:

java.lang.IllegalStateException: Allele in genotype TAG* not in the variant context [G*, *, C]
	at htsjdk.variant.variantcontext.VariantContext.validateGenotypes(VariantContext.java:1360)
	at htsjdk.variant.variantcontext.VariantContext.validate(VariantContext.java:1298)
	at htsjdk.variant.variantcontext.VariantContext.<init>(VariantContext.java:401)
	at htsjdk.variant.variantcontext.VariantContextBuilder.make(VariantContextBuilder.java:494)
	at htsjdk.variant.variantcontext.VariantContextBuilder.make(VariantContextBuilder.java:488)
	at org.broadinstitute.hellbender.utils.variant.GATKVariantContextUtils.simpleMerge(GATKVariantContextUtils.java:864)
	at org.broadinstitute.hellbender.utils.variant.GATKVariantContextUtils.simpleMerge(GATKVariantContextUtils.java:646)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.AssemblyBasedCallerUtils.makeMergedVariantContext(AssemblyBasedCallerUtils.java:221)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.HaplotypeCallerGenotypingEngine.assignGenotypeLikelihoods(HaplotypeCallerGenotypingEngine.java:150)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.HaplotypeCallerEngine.callRegion(HaplotypeCallerEngine.java:599)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.HaplotypeCaller.apply(HaplotypeCaller.java:236)
	at org.broadinstitute.hellbender.engine.AssemblyRegionWalker.processReadShard(AssemblyRegionWalker.java:291)
	at org.broadinstitute.hellbender.engine.AssemblyRegionWalker.traverse(AssemblyRegionWalker.java:267)
	at org.broadinstitute.hellbender.engine.GATKTool.doWork(GATKTool.java:966)
	at org.broadinstitute.hellbender.cmdline.CommandLineProgram.runTool(CommandLineProgram.java:139)
	at org.broadinstitute.hellbender.cmdline.CommandLineProgram.instanceMainPostParseArgs(CommandLineProgram.java:192)
	at org.broadinstitute.hellbender.cmdline.CommandLineProgram.instanceMain(CommandLineProgram.java:211)
	at org.broadinstitute.hellbender.Main.runCommandLineProgram(Main.java:160)
	at org.broadinstitute.hellbender.Main.mainEntry(Main.java:203)
	at org.broadinstitute.hellbender.Main.main(Main.java:289)

The second case included the following --alleles input:

22	16464044	rs571268158	CCAGGTCT	C
22	16464051	rs569099729	T	C

and it crashed similarly, with:

java.lang.IllegalStateException: Allele in genotype CCAGGTCT* not in the variant context [T*, *, C]
	at htsjdk.variant.variantcontext.VariantContext.validateGenotypes(VariantContext.java:1360)
	at htsjdk.variant.variantcontext.VariantContext.validate(VariantContext.java:1298)
	at htsjdk.variant.variantcontext.VariantContext.<init>(VariantContext.java:401)
	at htsjdk.variant.variantcontext.VariantContextBuilder.make(VariantContextBuilder.java:494)
	at htsjdk.variant.variantcontext.VariantContextBuilder.make(VariantContextBuilder.java:488)
	at org.broadinstitute.hellbender.utils.variant.GATKVariantContextUtils.simpleMerge(GATKVariantContextUtils.java:864)
	at org.broadinstitute.hellbender.utils.variant.GATKVariantContextUtils.simpleMerge(GATKVariantContextUtils.java:646)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.AssemblyBasedCallerUtils.makeMergedVariantContext(AssemblyBasedCallerUtils.java:221)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.HaplotypeCallerGenotypingEngine.assignGenotypeLikelihoods(HaplotypeCallerGenotypingEngine.java:150)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.HaplotypeCallerEngine.callRegion(HaplotypeCallerEngine.java:599)
	at org.broadinstitute.hellbender.tools.walkers.haplotypecaller.HaplotypeCaller.apply(HaplotypeCaller.java:236)
	at org.broadinstitute.hellbender.engine.AssemblyRegionWalker.processReadShard(AssemblyRegionWalker.java:291)
	at org.broadinstitute.hellbender.engine.AssemblyRegionWalker.traverse(AssemblyRegionWalker.java:267)
	at org.broadinstitute.hellbender.engine.GATKTool.doWork(GATKTool.java:966)
	at org.broadinstitute.hellbender.cmdline.CommandLineProgram.runTool(CommandLineProgram.java:139)
	at org.broadinstitute.hellbender.cmdline.CommandLineProgram.instanceMainPostParseArgs(CommandLineProgram.java:192)
	at org.broadinstitute.hellbender.cmdline.CommandLineProgram.instanceMain(CommandLineProgram.java:211)
	at org.broadinstitute.hellbender.Main.runCommandLineProgram(Main.java:160)
	at org.broadinstitute.hellbender.Main.mainEntry(Main.java:203)
	at org.broadinstitute.hellbender.Main.main(Main.java:289)

Based on the discussion surrounding #4963 and the test VCF, I gather that cases like these are intended to work, without crashing.

I was trying to figure out what was unique in these problematic cases, compared to the spanning deletion in the aforementioned test VCF. I noticed that the problematic cases both have the SNP at the very last base of the spanning deletion. I'm just speculating here, but maybe the issue is related to some sort of "off-by-one" bug?

This is based on testing with version 4.0.9.0.
I also tried with 4.0.5.1, and it didn't crash, but rather displayed warnings of the type discussed in #4963:
00:02:10.995 WARN HaplotypeCallerEngine - Multiple valid VCF records detected in the alleles input file at site 22:16137302-16137302, only considering the first record
00:03:08.220 WARN HaplotypeCallerEngine - Multiple valid VCF records detected in the alleles input file at site 22:16464051-16464051, only considering the first record

[gmagoon]

sorry, this is duplicate of #5336
I posted twice due to confusion related to recent github issues