public void VennVcf_FxnlTest_HG19()
{
var outDir = TestPaths.LocalScratchDirectory;
var VcfPathRoot = _TestDataPath;
string VcfA = Path.Combine(VcfPathRoot, "control_S15.vcf");
string VcfB = Path.Combine(VcfPathRoot, "control_S18.vcf");
string OutputPath = Path.Combine(outDir, "Consensus.vcf");
string ExpectedPath = Path.Combine(VcfPathRoot, "ExpectedConsensus.vcf");
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = Path.Combine(outDir, "Consensus.vcf");
parameters.OutputDirectory = outDir;
parameters.DebugMode = true;
parameters.CommandLineArguments = new string[] { "testcase commandline" };
VennProcessor Venn = new VennProcessor(new string[] { VcfA, VcfB }, parameters);
Venn.DoPairwiseVenn();
TestHelper.CompareFiles(OutputPath, ExpectedPath);
}
public void VennVcf_EmptyInputTest()
{
var outDir = TestPaths.LocalTestDataDirectory;
var VcfPathRoot = _TestDataPath;
string VcfA = Path.Combine(VcfPathRoot, "Empty_S1.vcf");
string VcfB = Path.Combine(VcfPathRoot, "Empty_S2.vcf");
string OutputPath = Path.Combine(outDir, "EmptyConsensus.vcf");
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = Path.Combine(outDir, "EmptyConsensus.vcf");
parameters.OutputDirectory = outDir;
parameters.DebugMode = true;
VennProcessor Venn = new VennProcessor(new string[] { VcfA, VcfB }, parameters);
Venn.DoPairwiseVenn();
Assert.True(File.Exists(OutputPath));
var observedVariants = AlleleReader.GetAllVariantsInFile(OutputPath);
Assert.Equal(0, observedVariants.Count);
}
public void VennVcf_GtTest()
{
var outDir = TestPaths.LocalScratchDirectory;
var VcfPathRoot = _TestDataPath;
string VcfA = Path.Combine(VcfPathRoot, "gtTests_S15.vcf");
string VcfB = Path.Combine(VcfPathRoot, "gtTests_S18.vcf");
string OutputPath = Path.Combine(outDir, "gtConsensusOut.vcf");
string ExpectedPath = Path.Combine(VcfPathRoot, "gtConsensus.vcf");
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = OutputPath;
parameters.OutputDirectory = outDir;
VennProcessor Venn = new VennProcessor(new string[] { VcfA, VcfB }, parameters);
Venn.DoPairwiseVenn();
Assert.True(File.Exists(OutputPath));
var expectedVariants = AlleleReader.GetAllVariantsInFile(ExpectedPath);
var observedVariants = AlleleReader.GetAllVariantsInFile(OutputPath);
Assert.Equal(expectedVariants.Count, observedVariants.Count);
for (int i = 0; i < expectedVariants.Count; i++)
{
var ExpectedVariant = expectedVariants[i];
var OutputVariant = observedVariants[i];
Assert.Equal(ExpectedVariant.ToString(), OutputVariant.ToString());
}
}
public void VennVcf_CombineTwoPoolVariants_MergeRefCalls()
{
//this is from an issue where there were multiple co-located variants in one pool,
//and just ref in the other, at chr15 92604460. The consensus answer should be
// a single ref call (and not multiple ref calls!).
var outDir = TestPaths.LocalScratchDirectory;
var vcfPathRoot = _TestDataPath;
string VcfPath_PoolA = Path.Combine(vcfPathRoot, "C64-Ct-4_S17.genome.vcf");
string VcfPath_PoolB = Path.Combine(vcfPathRoot, "C64-Ct-4_S18.genome.vcf");
string VcfPath_Consensus = Path.Combine(vcfPathRoot, "ExpectedConsensus2.vcf");
string OutputPath = Path.Combine(outDir, "Consensus2.vcf");
if (File.Exists(OutputPath))
{
File.Delete(OutputPath);
}
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.InputFiles = new string[] { VcfPath_PoolA, VcfPath_PoolB };
parameters.OutputDirectory = outDir;
parameters.ConsensusFileName = OutputPath;
VennProcessor venn = new VennProcessor(parameters.InputFiles, parameters);
venn.DoPairwiseVenn();
Assert.Equal(File.Exists(OutputPath), true);
var CombinedVariants = AlleleReader.GetAllVariantsInFile(OutputPath);
var ExpectedVariants = AlleleReader.GetAllVariantsInFile(VcfPath_Consensus);
Assert.Equal(ExpectedVariants.Count, CombinedVariants.Count);
int NumVariantsAtPos92604460 = 0;
for (int i = 0; i < ExpectedVariants.Count; i++)
{
var EVariant = ExpectedVariants[i];
var Variant = CombinedVariants[i];
if ((Variant.ReferencePosition == 92604460) &&
(Variant.Chromosome == "chr15"))
{
NumVariantsAtPos92604460++;
}
Assert.Equal(EVariant.ToString(), Variant.ToString());
}
Assert.Equal(NumVariantsAtPos92604460, 1);
}
public void VennVcf_FxnlTest()
{
var outDir = TestPaths.LocalScratchDirectory;
var VcfPathRoot = _TestDataPath;
string VcfA = Path.Combine(VcfPathRoot, "control_S15.vcf");
string VcfB = Path.Combine(VcfPathRoot, "control_S18.vcf");
string OutputPath = Path.Combine(outDir, "Consensus.vcf");
string ExpectedPath = Path.Combine(VcfPathRoot, "ExpectedConsensus.vcf");
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = Path.Combine(outDir, "Consensus.vcf");
parameters.OutputDirectory = outDir;
parameters.DebugMode = true;
VennProcessor Venn = new VennProcessor(new string[] { VcfA, VcfB }, parameters);
Venn.DoPairwiseVenn(false);
Assert.True(File.Exists(OutputPath));
using (VcfReader ReaderE = new VcfReader(ExpectedPath))
{
using (VcfReader ReaderO = new VcfReader(OutputPath))
{
VcfVariant ExpectedVariant = new VcfVariant();
VcfVariant OutputVariant = new VcfVariant();
while (true)
{
bool ExpectedVariantExists = ReaderE.GetNextVariant(ExpectedVariant);
bool OutputVariantExists = ReaderO.GetNextVariant(OutputVariant);
Assert.Equal(ExpectedVariantExists, OutputVariantExists);
if (!ExpectedVariantExists || !OutputVariantExists)
{
break;
}
Assert.Equal(ExpectedVariant.ToString(), OutputVariant.ToString());
}
}
}
}
public void VennVcf_FxnlTest_GRCH37()
{
var outDir = Path.Combine(TestPaths.LocalScratchDirectory, "GRCH37_fxnl");
TestHelper.RecreateDirectory(outDir);
string VcfA = Path.Combine(TestPaths.LocalTestDataDirectory, "GRCH37_S25.bam.genome.vcf");
string VcfB = Path.Combine(TestPaths.LocalTestDataDirectory, "GRCH37_S30.bam.genome.vcf");
List <string> OutputVcfs = new List <string>()
{
Path.Combine(outDir, "Consensus.vcf"),
Path.Combine(outDir, "GRCH37_S25.bam_and_GRCH37_S30.bam.vcf"),
Path.Combine(outDir, "GRCH37_S25.bam_not_GRCH37_S30.bam.vcf"),
Path.Combine(outDir, "GRCH37_S30.bam_and_GRCH37_S25.bam.vcf"),
Path.Combine(outDir, "GRCH37_S30.bam_not_GRCH37_S25.bam.vcf")
};
List <string> ExpectedVcfs = new List <string>()
{
Path.Combine(TestPaths.LocalTestDataDirectory, "Expected_GRCH37_Consensus.vcf"),
Path.Combine(TestPaths.LocalTestDataDirectory, "Expected_GRCH37_S25.bam_and_GRCH37_S30.bam.vcf"),
Path.Combine(TestPaths.LocalTestDataDirectory, "Expected_GRCH37_S25.bam_not_GRCH37_S30.bam.vcf"),
Path.Combine(TestPaths.LocalTestDataDirectory, "Expected_GRCH37_S30.bam_and_GRCH37_S25.bam.vcf"),
Path.Combine(TestPaths.LocalTestDataDirectory, "Expected_GRCH37_S30.bam_not_GRCH37_S25.bam.vcf")
};
VennVcfOptions parameters = new VennVcfOptions();
parameters.ConsensusFileName = Path.Combine(outDir, "Consensus.vcf");
parameters.OutputDirectory = outDir;
parameters.DebugMode = true;
parameters.VariantCallingParams.AmpliconBiasFilterThreshold = null;
VennProcessor Venn = new VennProcessor(new string[] { VcfA, VcfB }, parameters);
Venn.DoPairwiseVenn();
for (int i = 0; i < 5; i++)
{
TestHelper.CompareFiles(OutputVcfs[i], ExpectedVcfs[i]);
}
}
public void VennVcf_CombineTwoPoolVariants_RulesEandF_Tests()
{
//Rule "E" test (ie an Alt+ref call converges to a REf, and we also had a ref call following it)
//E if we end up with multiple REF calls for the same loci, combine those .VCF lines into one ref call.
//Rule "F" test (ie various alt calls all ended up as no-call. we dont want multiple no call lines in the vcf.)
//F if we end up with multiple NOCALL calls for the same loci, leave those .VCF lines separate
var outDir = TestPaths.LocalScratchDirectory;
var VcfPathRoot = _TestDataPath;
string VcfPath_PoolA = Path.Combine(VcfPathRoot, "RulesEandF_S1.genome.vcf");
string VcfPath_PoolB = Path.Combine(VcfPathRoot, "RulesEandF_S2.genome.vcf");
string OutputPath = Path.Combine(outDir, "outEandF.vcf");
if (File.Exists(OutputPath))
{
File.Delete(OutputPath);
}
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = OutputPath;
parameters.OutputDirectory = outDir;
VennProcessor VennVcf = new VennProcessor(
new string[] { VcfPath_PoolA, VcfPath_PoolB }, parameters);
VennVcf.DoPairwiseVenn();
Assert.Equal(File.Exists(OutputPath), true);
var PoolAVariants = AlleleReader.GetAllVariantsInFile(VcfPath_PoolA);
var PoolBVariants = AlleleReader.GetAllVariantsInFile(VcfPath_PoolB);
var CombinedVariants = AlleleReader.GetAllVariantsInFile(OutputPath);
//Rule "E" test (ie an Alt+ref call converges to a REf, and we also had a ref call following it)
//E if we end up with multiple REF calls for the same loci, combine those .VCF lines into one ref call.
var VariantA_1 = PoolAVariants[0];
Assert.Equal(VariantA_1.Genotype, Genotype.HomozygousRef);
Assert.Equal(VariantA_1.Frequency, 1.0 - 0.0021, 4); //note Vf here is the ref freq
Assert.Equal(VariantA_1.VariantQscore, 100);
Assert.Equal(VariantA_1.Filters.Count, 0);
Assert.Equal(VariantA_1.ReferencePosition, 25378561);
var VariantA_2 = PoolAVariants[1];
Assert.Equal(VariantA_2.ReferencePosition, 25378562);
var VariantB_1 = PoolBVariants[0];
Assert.Equal(VariantB_1.Genotype, Genotype.HeterozygousAltRef); //"0/1");
Assert.Equal(VariantB_1.Frequency, 0.0173, 4);
Assert.Equal(VariantB_1.VariantQscore, 100);
Assert.Equal(VariantB_1.Filters.Count, 0);
Assert.Equal(VariantB_1.ReferencePosition, 25378561);
var VariantB_2 = PoolBVariants[1];
Assert.Equal(VariantB_2.Genotype, Genotype.HomozygousRef);
Assert.Equal(VariantB_2.Frequency, 0.9827, 4); //note Vf here is the ref freq
Assert.Equal(VariantB_2.VariantQscore, 100);
Assert.Equal(VariantB_2.Filters.Count, 0);
Assert.Equal(VariantB_2.ReferencePosition, 25378561);
var Consensus_1 = CombinedVariants[0];
Assert.Equal(Consensus_1.Genotype, Genotype.HomozygousRef);
Assert.Equal(Consensus_1.Frequency, 0.9907, 4); //slightly improved from .008. //note Vf here is the ref freq
Assert.Equal(Consensus_1.VariantQscore, 100);
Assert.Equal(Consensus_1.Filters.Count, 0); //<-low VF tag will NOT added by post-processing b/c is ref call
Assert.Equal(Consensus_1.ReferencePosition, 25378561);
var Consensus_2 = CombinedVariants[1];
Assert.Equal(Consensus_2.ReferencePosition, 25378562);
//Rule "F" test (ie various alt calls all ended up as no-call.
//F if we end up with multiple NOCALL calls for the same loci, leave those .VCF lines separate
VariantA_1 = PoolAVariants[1];
Assert.Equal(VariantA_1.Genotype, Genotype.HeterozygousAltRef); //"0/1");
Assert.Equal(VariantA_1.Frequency, 0.0776, 4);
Assert.Equal(VariantA_1.VariantQscore, 100);
Assert.Equal(VariantA_1.Filters.Count, 0);
Assert.Equal(VariantA_1.ReferencePosition, 25378562);
VariantA_2 = PoolAVariants[2];
Assert.Equal(VariantA_2.Genotype, Genotype.HeterozygousAltRef); //"0/1");
Assert.Equal(VariantA_2.Frequency, 0.0776, 4);
Assert.Equal(VariantA_2.VariantQscore, 100);
Assert.Equal(VariantA_2.Filters.Count, 0);
Assert.Equal(VariantA_2.ReferencePosition, 25378562);
var VariantA_3 = PoolAVariants[3];
Assert.Equal(VariantA_3.Genotype, Genotype.HeterozygousAltRef); //"0/1");
Assert.Equal(VariantA_3.Frequency, 0.0776, 4);
Assert.Equal(VariantA_3.VariantQscore, 100);
Assert.Equal(VariantA_3.Filters.Count, 0);
Assert.Equal(VariantA_3.ReferencePosition, 25378562);
VariantB_1 = PoolBVariants[2];
Assert.Equal(VariantB_1.Genotype, Genotype.HomozygousRef);
Assert.Equal(VariantB_1.Frequency, 0.9989, 4);
Assert.Equal(VariantB_1.VariantQscore, 100);
Assert.Equal(VariantB_1.Filters.Count, 0);
Assert.Equal(VariantB_1.ReferencePosition, 25378562);
VariantB_2 = PoolBVariants[3];
Assert.Equal(VariantB_2.ReferencePosition, 25378563);
Consensus_1 = CombinedVariants[1];
Assert.Equal(Consensus_1.ReferencePosition, 25378562);
Assert.Equal(Consensus_1.Genotype, Genotype.AltLikeNoCall);
Assert.Equal(Consensus_1.Frequency, 0.0069, 4);
Assert.Equal(Consensus_1.VariantQscore, 0);
Assert.Equal(Consensus_1.Filters.Count, 1); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_1.Filters[0], FilterType.PoolBias); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_1.ReferenceAllele, "C");
Assert.Equal(Consensus_1.AlternateAllele, "T");
Consensus_2 = CombinedVariants[2];
Assert.Equal(Consensus_2.ReferencePosition, 25378562);
Assert.Equal(Consensus_2.Genotype, Genotype.AltLikeNoCall);
Assert.Equal(Consensus_2.Frequency, 0.0069, 4);
Assert.Equal(Consensus_2.VariantQscore, 0);
Assert.Equal(Consensus_1.Filters.Count, 1); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_1.Filters[0], FilterType.PoolBias); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_2.ReferenceAllele, "C");
Assert.Equal(Consensus_2.AlternateAllele, "TT");
var Consensus_3 = CombinedVariants[3];
Assert.Equal(Consensus_3.ReferencePosition, 25378562);
Assert.Equal(Consensus_3.Genotype, Genotype.AltLikeNoCall);
Assert.Equal(Consensus_3.Frequency, 0.0069, 4);
Assert.Equal(Consensus_3.VariantQscore, 0);
Assert.Equal(Consensus_1.Filters.Count, 1); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_1.Filters[0], FilterType.PoolBias); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_3.ReferenceAllele, "CC");
Assert.Equal(Consensus_3.AlternateAllele, "T");
var Consensus_4 = CombinedVariants[4];
Assert.Equal(Consensus_4.ReferencePosition, 25378563);
if (File.Exists(OutputPath))
{
File.Delete(OutputPath);
}
}
public void VennVcf_CombineTwoPoolVariants_ProbePoolBias_Tests()
{
//this is from an issue anita had where a variant was in one pool at 1%, the other at 0%, and showed up as 6% in the combined pool.
var outDir = TestPaths.LocalScratchDirectory;
var VcfPathRoot = _TestDataPath;
string VcfPath_PoolA = Path.Combine(VcfPathRoot, "small_S14.genome.vcf");
string VcfPath_PoolB = Path.Combine(VcfPathRoot, "small_S17.genome.vcf");
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = Path.Combine(outDir, "Consensus.vcf");
parameters.OutputDirectory = outDir;
if (File.Exists(parameters.ConsensusFileName))
{
File.Delete(parameters.ConsensusFileName);
}
VennProcessor Venn = new VennProcessor(new string[] { VcfPath_PoolA, VcfPath_PoolB }, parameters);
Venn.DoPairwiseVenn();
Assert.Equal(File.Exists(parameters.ConsensusFileName), true);
var CombinedVariants = AlleleReader.GetAllVariantsInFile(parameters.ConsensusFileName);
var AandBVariants = AlleleReader.GetAllVariantsInFile(Path.Combine(outDir, "small_S14_and_S17.vcf"));
var BandAVariants = AlleleReader.GetAllVariantsInFile(Path.Combine(outDir, "small_S17_and_S14.vcf"));
var AnotBVariants = AlleleReader.GetAllVariantsInFile(Path.Combine(outDir, "small_S14_not_S17.vcf"));
var BnotAVariants = AlleleReader.GetAllVariantsInFile(Path.Combine(outDir, "small_S17_not_S14.vcf"));
//poolA
//chr1 115258743 . A . 100 PASS DP=35354 GT:GQ:AD:VF:NL:SB 0/0:100:30256:0.1442:20:-100.0000
//chr1 115258743 . AC TT 100 PASS DP=35354 GT:GQ:AD:VF:NL:SB 0/1:100:30720,4634:0.1311:20:-100.0000
//chr1 115258744 . C . 100 PASS DP=35253 GT:GQ:AD:VF:NL:SB 0/0:100:30277:0.1412:20:-100.0000
//chr1 115258745 . C . 100 PASS DP=35160 GT:GQ:AD:VF:NL:SB 0/0:100:35130:0.0009:20:-100.0000
//poolB
//chr1 115258743 . AC TT 100 PASS DP=49612 GT:GQ:AD:VF:NL:SB 0/1:100:44202,5410:0.1090:20:-100.0000
//chr1 115258743 . A T 100 PASS DP=49612 GT:GQ:AD:VF:NL:SB 0/1:100:43362,670:0.0135:20:-46.0807
//chr1 115258744 . C T 24 PASS DP=49902 GT:GQ:AD:VF:NL:SB 0/1:24:43905,560:0.0112:20:-8.3857
//when we had bug:
//chr1 115258743 . AC TT 100.00 PASS DP=84966 GT:GQ:AD:VF:NL: 0/1:100:74922,10044:0.1182:20:-100:-100.0000:100
//chr1 115258743 . A T 100.00 PB;LowVF DP=49612 GT:GQ:AD:VF:NL: ./.:100:43362,670:0.0135:20:-46.0807:0.0000:100
//chr1 115258743 . A . 100.00 PASS DP=35354 GT:GQ:AD:VF:NL: 0/0:100:30256:0.1442:20:-100.0000:-100.0000:100
//chr1 115258744 . C T 100.00 PB DP=85155 GT:GQ:AD:VF:NL: 0/1:100:74182,5536:0.0650:20:-
//(issue#1) at 743 we had a A->. in only one pool. It should be marked as BIAS and not PASS.
//(issue#2) at 744 we had a C->T at 6% when it should be at ~0%, and called as a ref.
var FunnyResult0 = CombinedVariants[3];
Assert.Equal(FunnyResult0.Frequency, 0.8558, 4);
Assert.Equal(FunnyResult0.Filters.Count, 1);
Assert.Equal(FunnyResult0.Filters[0], FilterType.PoolBias);
Assert.Equal(FunnyResult0.ReferenceAllele, "A");
Assert.Equal(FunnyResult0.AlternateAllele, ".");
//this used to be a reference as a pass, even though it was only called in one pool.
var FunnyResult = CombinedVariants[6];
Assert.Equal(FunnyResult.ReferencePosition, 115258744);
Assert.Equal(FunnyResult.Frequency, 0.8711, 4);
Assert.Equal(FunnyResult.Filters.Count, 0);
Assert.Equal(FunnyResult.ReferenceAllele, "C");
Assert.Equal(FunnyResult.AlternateAllele, ".");
//when we had the bug, this used to get called at 6%.
//now, check the Venn functionality:
Assert.Equal(2, AandBVariants.Count());
Assert.Equal(2, BandAVariants.Count());
Assert.Equal(2, AnotBVariants.Count());
Assert.Equal(0, BnotAVariants.Count());
Assert.Equal(115258743, AandBVariants[0].ReferencePosition);
Assert.Equal("AC", AandBVariants[0].ReferenceAllele);
Assert.Equal("TT", AandBVariants[0].AlternateAllele);
Assert.Equal(115258747, AandBVariants[1].ReferencePosition);
Assert.Equal("C", AandBVariants[1].ReferenceAllele);
Assert.Equal("T", AandBVariants[1].AlternateAllele);
Assert.Equal(115258743, BandAVariants[0].ReferencePosition);
Assert.Equal("AC", BandAVariants[0].ReferenceAllele);
Assert.Equal("TT", BandAVariants[0].AlternateAllele);
Assert.Equal(115258747, BandAVariants[1].ReferencePosition);
Assert.Equal("C", BandAVariants[1].ReferenceAllele);
Assert.Equal("T", BandAVariants[1].AlternateAllele);
Assert.Equal(115258743, AnotBVariants[0].ReferencePosition);
Assert.Equal("A", AnotBVariants[0].ReferenceAllele);
Assert.Equal("T", AnotBVariants[0].AlternateAllele);
Assert.Equal(115258744, AnotBVariants[1].ReferencePosition);
Assert.Equal("C", AnotBVariants[1].ReferenceAllele);
Assert.Equal("T", AnotBVariants[1].AlternateAllele);
}
public void VennVcf_CombineTwoPoolVariants_RulesEandF_Tests()
{
//Rule "E" test (ie an Alt+ref call converges to a REf, and we also had a ref call following it)
//E if we end up with multiple REF calls for the same loci, combine those .VCF lines into one ref call.
//Rule "F" test (ie various alt calls all ended up as no-call. we dont want multiple no call lines in the vcf.)
//F if we end up with multiple NOCALL calls for the same loci, leave those .VCF lines separate
var outDir = TestPaths.LocalScratchDirectory;
var VcfPathRoot = _TestDataPath;
string VcfPath_PoolA = Path.Combine(VcfPathRoot, "RulesEandF_S1.genome.vcf");
string VcfPath_PoolB = Path.Combine(VcfPathRoot, "RulesEandF_S2.genome.vcf");
string OutputPath = Path.Combine(outDir, "outEandF.vcf");
if (File.Exists(OutputPath))
{
File.Delete(OutputPath);
}
VennVcfOptions parameters = new VennVcfOptions();
parameters.VariantCallingParams.MinimumFrequencyFilter = 0.03f;
parameters.VariantCallingParams.MinimumFrequency = 0.01f;
parameters.ConsensusFileName = OutputPath;
parameters.OutputDirectory = outDir;
VennProcessor VennVcf = new VennProcessor(
new string[] { VcfPath_PoolA, VcfPath_PoolB }, parameters);
VennVcf.DoPairwiseVenn(false);
Assert.Equal(File.Exists(OutputPath), true);
List <VcfVariant> PoolAVariants = VcfReader.GetAllVariantsInFile(VcfPath_PoolA);
List <VcfVariant> PoolBVariants = VcfReader.GetAllVariantsInFile(VcfPath_PoolB);
List <VcfVariant> CombinedVariants = VcfReader.GetAllVariantsInFile(OutputPath);
//Rule "E" test (ie an Alt+ref call converges to a REf, and we also had a ref call following it)
//E if we end up with multiple REF calls for the same loci, combine those .VCF lines into one ref call.
VcfVariant VariantA_1 = PoolAVariants[0];
Assert.Equal(VariantA_1.Genotypes[0]["GT"], "0/0");
Assert.Equal(VariantA_1.Genotypes[0]["VF"], "0.0021");
Assert.Equal(VariantA_1.Quality, 100);
Assert.Equal(VariantA_1.Filters, "PASS");
Assert.Equal(VariantA_1.ReferencePosition, 25378561);
VcfVariant VariantA_2 = PoolAVariants[1];
Assert.Equal(VariantA_2.ReferencePosition, 25378562);
VcfVariant VariantB_1 = PoolBVariants[0];
Assert.Equal(VariantB_1.Genotypes[0]["GT"], "0/1");
Assert.Equal(VariantB_1.Genotypes[0]["VF"], "0.0173");
Assert.Equal(VariantB_1.Quality, 100);
Assert.Equal(VariantB_1.Filters, "PASS");
Assert.Equal(VariantB_1.ReferencePosition, 25378561);
VcfVariant VariantB_2 = PoolBVariants[1];
Assert.Equal(VariantB_2.Genotypes[0]["GT"], "0/0");
Assert.Equal(VariantB_2.Genotypes[0]["VF"], "0.0021");
Assert.Equal(VariantB_2.Quality, 100);
Assert.Equal(VariantB_2.Filters, "PASS");
Assert.Equal(VariantB_2.ReferencePosition, 25378561);
VcfVariant Consensus_1 = CombinedVariants[0];
Assert.Equal(Consensus_1.Genotypes[0]["GT"], "0/0");
Assert.Equal(Consensus_1.Genotypes[0]["VF"], "0.009"); //slightly improved from .008
Assert.Equal(Consensus_1.Quality, 100);
Assert.Equal(Consensus_1.Filters, "PASS"); //<-low VF tag will NOT added by post-processing b/c is ref call
Assert.Equal(Consensus_1.ReferencePosition, 25378561);
VcfVariant Consensus_2 = CombinedVariants[1];
Assert.Equal(Consensus_2.ReferencePosition, 25378562);
//Rule "F" test (ie various alt calls all ended up as no-call.
//F if we end up with multiple NOCALL calls for the same loci, leave those .VCF lines separate
VariantA_1 = PoolAVariants[1];
Assert.Equal(VariantA_1.Genotypes[0]["GT"], "0/1");
Assert.Equal(VariantA_1.Genotypes[0]["VF"], "0.0725");
Assert.Equal(VariantA_1.Quality, 100);
Assert.Equal(VariantA_1.Filters, "PASS");
Assert.Equal(VariantA_1.ReferencePosition, 25378562);
VariantA_2 = PoolAVariants[2];
Assert.Equal(VariantA_2.Genotypes[0]["GT"], "0/1");
Assert.Equal(VariantA_2.Genotypes[0]["VF"], "0.0725");
Assert.Equal(VariantA_2.Quality, 100);
Assert.Equal(VariantA_2.Filters, "PASS");
Assert.Equal(VariantA_2.ReferencePosition, 25378562);
VcfVariant VariantA_3 = PoolAVariants[3];
Assert.Equal(VariantA_3.Genotypes[0]["GT"], "0/1");
Assert.Equal(VariantA_3.Genotypes[0]["VF"], "0.0725");
Assert.Equal(VariantA_3.Quality, 100);
Assert.Equal(VariantA_3.Filters, "PASS");
Assert.Equal(VariantA_3.ReferencePosition, 25378562);
VariantB_1 = PoolBVariants[2];
Assert.Equal(VariantB_1.Genotypes[0]["GT"], "0/0");
Assert.Equal(VariantB_1.Genotypes[0]["VF"], "0.0024");
Assert.Equal(VariantB_1.Quality, 100);
Assert.Equal(VariantB_1.Filters, "PASS");
Assert.Equal(VariantB_1.ReferencePosition, 25378562);
VariantB_2 = PoolBVariants[3];
Assert.Equal(VariantB_2.ReferencePosition, 25378563);
Consensus_1 = CombinedVariants[1];
Assert.Equal(Consensus_1.ReferencePosition, 25378562);
Assert.Equal(Consensus_1.Genotypes[0]["GT"], "./.");
Assert.Equal(Consensus_1.Genotypes[0]["VF"], "0.007");
Assert.Equal(Consensus_1.Quality, 0);
Assert.Equal(Consensus_1.Filters, "PB"); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_1.ReferenceAllele, "C");
Assert.Equal(Consensus_1.VariantAlleles[0], "T");
Consensus_2 = CombinedVariants[2];
Assert.Equal(Consensus_2.ReferencePosition, 25378562);
Assert.Equal(Consensus_2.Genotypes[0]["GT"], "./.");
Assert.Equal(Consensus_2.Genotypes[0]["VF"], "0.007");
Assert.Equal(Consensus_2.Quality, 0);
Assert.Equal(Consensus_2.Filters, "PB"); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_2.ReferenceAllele, "C");
Assert.Equal(Consensus_2.VariantAlleles[0], "TT");
VcfVariant Consensus_3 = CombinedVariants[3];
Assert.Equal(Consensus_3.ReferencePosition, 25378562);
Assert.Equal(Consensus_3.Genotypes[0]["GT"], "./.");
Assert.Equal(Consensus_3.Genotypes[0]["VF"], "0.007");
Assert.Equal(Consensus_3.Quality, 0);
Assert.Equal(Consensus_3.Filters, "PB"); //<-low VF tag will also get added by post-processing
Assert.Equal(Consensus_3.ReferenceAllele, "CC");
Assert.Equal(Consensus_3.VariantAlleles[0], "T");
VcfVariant Consensus_4 = CombinedVariants[4];
Assert.Equal(Consensus_4.ReferencePosition, 25378563);
if (File.Exists(OutputPath))
{
File.Delete(OutputPath);
}
}