public void CallThroughAnEmptyNbhd()
{
var originalVcfVariant = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "T", 1000, 156);
var originalVcfVariant2 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 124, "A", "T", 1000, 156);
var vs1 = new VariantSite(originalVcfVariant);
var vs2 = new VariantSite(originalVcfVariant2);
var caller = new VariantCaller(new VariantCallingParameters(), new BamFilterParameters());
//since there is an alt at position 124 ( a call of 156 alt / 1000 total, that means 844 original ref calls.
//Of which we said, 100 will get sucked up. So that leaves 744 / 1000 calls for a reference.
//So, we can still make a confident ref call.
var nbhd = new VcfNeighborhood(new VariantCallingParameters(), "chr1", vs1, vs2, "");
nbhd.SetRangeOfInterest();
caller.CallMNVs(nbhd);
caller.CallRefs(nbhd);
var acceptedMNVs = nbhd.CalledVariants;
var acceptedRefs = nbhd.CalledRefs;
Assert.Equal(0, acceptedMNVs.Count);
Assert.Equal(2, acceptedRefs.Count);
Assert.Equal(Genotype.HomozygousRef, acceptedRefs[123].Genotype);
Assert.Equal(Genotype.HomozygousRef, acceptedRefs[124].Genotype);
Assert.Equal(123, acceptedRefs[123].Coordinate);
Assert.Equal(124, acceptedRefs[124].Coordinate);
}
private List <VcfNeighborhood> GetNeighborhoods(int expectedNumberOfThreads)
{
var neighborhoods = new List <VcfNeighborhood>();
for (var i = 0; i < expectedNumberOfThreads; i++)
{
var neighborhood = new VcfNeighborhood(new VariantCallingParameters(), "chr1", new VariantSite(120), new VariantSite(121), "T")
{
VcfVariantSites = new List <VariantSite>
{
new VariantSite(123)
{
ReferenceName = "chr1",
OriginalAlleleFromVcf = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "T", 1000, 156)
//orignally at index 0
},
}
};
neighborhoods.Add(neighborhood);
}
return(neighborhoods);
}
public void WriteHeader()
{
//WriteHeader should write the original header and add a line about phaser used right before the column headers
var writer = InitializeWriter(false);
writer.WriteHeader();
writer.Dispose();
Assert.Throws <Exception>(() => writer.WriteHeader());
Assert.Throws <Exception>(() => writer.Write(new List <CalledAllele> {
PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "G", 1000, 156)
}));
writer.Dispose();
Assert.True(File.Exists(_outputFile));
var fileLines = File.ReadAllLines(_outputFile);
Assert.Equal(_origHeader[0], fileLines[0]);
Assert.Equal(_origHeader[1], fileLines[1]);
Assert.NotEqual(_origHeader[2], fileLines[2]);
Assert.True(fileLines[2].StartsWith("##VariantPhaser=Scylla"));
Assert.Equal(_origHeader[2], fileLines[4]);
}
public void VarCallsBecomeRefsAndNulls()
{
var originalVcfVariant = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "T", 1000, 156);
var originalVcfVariant2 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 124, "A", "T", 1000, 156);
var vs1 = new VariantSite(originalVcfVariant);
var vs2 = new VariantSite(originalVcfVariant2);
var caller = new VariantCaller(new VariantCallingParameters(), new BamFilterParameters());
//since there is an alt at position 124 ( a call of 156 alt / 1000 total, that means 844 original ref calls.
//Of which we said, 100 will get sucked up. So that leaves 744 / 1000 calls for a reference.
//So, we can still make a confident ref call.
var nbhd = new VcfNeighborhood(new VariantCallingParameters(), "chr1", vs1, vs2, "");
nbhd.SetRangeOfInterest();
nbhd.AddAcceptedPhasedVariant(
new CalledAllele(AlleleCategory.Snv)
{
Chromosome = "chr1", Coordinate = 123, Reference = "A", Alternate = "T", VariantQscore = 35, AlleleSupport = 10, TotalCoverage = 50
});
nbhd.UsedRefCountsLookup = new Dictionary <int, int>()
{
};
caller.CallMNVs(nbhd);
caller.CallRefs(nbhd);
var acceptedMNVs = nbhd.CalledVariants;
var acceptedRefs = nbhd.CalledRefs;
Assert.Equal(1, acceptedMNVs.Count);
Assert.Equal(1, acceptedMNVs[123].Count);
Assert.Equal(2, acceptedRefs.Count);
var vcfVariant2asRef = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }, { "DP", "1000" }, { "AD", "844" }
}
},
};
VcfMergerTests.CheckVariantsMatch(originalVcfVariant, acceptedMNVs[123][0]);
VcfMergerTests.CheckVariantsMatch(vcfVariant2asRef, acceptedRefs[124]);
// If one has been sucked up and there are refs remaining, we should output it as a ref.
nbhd.UsedRefCountsLookup = new Dictionary <int, int>()
{
{ 124, 100 }
};
caller.CallMNVs(nbhd);
caller.CallRefs(nbhd);
acceptedMNVs = nbhd.CalledVariants;
acceptedRefs = nbhd.CalledRefs;
Assert.Equal(1, acceptedMNVs.Count);
Assert.Equal(1, acceptedMNVs[123].Count);
Assert.Equal(2, acceptedRefs.Count);
vcfVariant2asRef = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }, { "DP", "1000" }, { "AD", "744" }
}
},
};
VcfMergerTests.CheckVariantsMatch(originalVcfVariant, acceptedMNVs[123][0]);
VcfMergerTests.CheckVariantsMatch(vcfVariant2asRef, acceptedRefs[124]);
// If one has been sucked up all the way
// we should output it as a null.
nbhd.UsedRefCountsLookup = new Dictionary <int, int>()
{
{ 124, 1000 }
};
caller.CallMNVs(nbhd);
caller.CallRefs(nbhd);
acceptedMNVs = nbhd.CalledVariants;
acceptedRefs = nbhd.CalledRefs;
Assert.Equal(1, acceptedMNVs.Count);
Assert.Equal(1, acceptedMNVs[123].Count);
Assert.Equal(2, acceptedRefs.Count);
var vcfVariant2asNull = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "./." }, { "DP", "1000" }, { "AD", "0" }
}
},
};
VcfMergerTests.CheckVariantsMatch(originalVcfVariant, acceptedMNVs[123][0]);
VcfMergerTests.CheckVariantsMatch(vcfVariant2asNull, acceptedRefs[124]);
}
public void CallAVariantInANewLocation()
{
//set up the original variants
var originalVcfVariant1 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "T", 1000, 156);
var originalVcfVariant2 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 124, "A", "T", 1000, 156);
var originalVcfVariant3 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 234, "A", "T", 1000, 156);
var originalVcfVariant4 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 234, "A", "T", 1000, 156);
var vs1 = new VariantSite(originalVcfVariant1);
var vs2 = new VariantSite(originalVcfVariant2);
var vs3 = new VariantSite(originalVcfVariant3);
var vs4 = new VariantSite(originalVcfVariant4);
var caller = new VariantCaller(new VariantCallingParameters(), new BamFilterParameters());
var nbhd = new VcfNeighborhood(new VariantCallingParameters(), "chr1", vs1, vs2, "");
nbhd.AddVariantSite(vs3, "RRRRR"); //note, we do not add vs4, that is not goig to get used for phasing. Sps it is a variant that failed filters.
nbhd.SetRangeOfInterest();
//now stage one candidate MNV:
var newMNV = new CalledAllele(AlleleCategory.Snv)
{
Chromosome = "chr1", Coordinate = 129, Reference = "A", Alternate = "TT"
};
nbhd.AddAcceptedPhasedVariant(newMNV);
nbhd.UsedRefCountsLookup = new Dictionary <int, int>()
{
{ 124, 1000 }
};
caller.CallMNVs(nbhd);
caller.CallRefs(nbhd);
var acceptedMNVs = nbhd.CalledVariants;
var acceptedRefs = nbhd.CalledRefs;
var vcfVariant0asRef = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 123,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }
}
},
};
var vcfVariant3asRef = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 234,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }
}
},
};
var vcfVariant2asNull = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "./." }
}
},
};
Assert.Equal(1, acceptedMNVs.Count);
Assert.Equal(1, acceptedMNVs[129].Count);
Assert.Equal(3, acceptedRefs.Count);
VcfMergerTests.CheckVariantsMatch(vcfVariant0asRef, acceptedRefs[123]);
VcfMergerTests.CheckVariantsMatch(vcfVariant2asNull, acceptedRefs[124]);
VcfMergerTests.CheckVariantsMatch(newMNV, acceptedMNVs[129][0]);
VcfMergerTests.CheckVariantsMatch(vcfVariant3asRef, acceptedRefs[234]);
}
public void FilterHeader()
{
var outputFilePath = Path.Combine(UnitTestPaths.TestDataDirectory, "PhasedVcfFileWriterTests.vcf");
File.Delete(outputFilePath);
var context = new VcfWriterInputContext
{
CommandLine = new [] { "myCommandLine" },
SampleName = "mySample",
ReferenceName = "myReference",
ContigsByChr = new List <Tuple <string, long> >
{
new Tuple <string, long>("chr1", 10001),
new Tuple <string, long>("chrX", 500)
}
};
// Variant strand bias too high or coverage on only one strand
var config = new VcfWriterConfig
{
DepthFilterThreshold = 500,
VariantQualityFilterThreshold = 30,
FrequencyFilterThreshold = 0.007f,
ShouldOutputNoCallFraction = true,
ShouldOutputStrandBiasAndNoiseLevel = true,
EstimatedBaseCallQuality = 23,
PloidyModel = PloidyModel.Diploid,
};
//note, scylla has no SB or RMxN or R8 filters.
var variants = new List <CalledAllele>
{
PhasedVariantTestUtilities.CreateDummyAllele("chrX", 123, "A", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr10", 124, "A", "C", 1000, 156),
};
variants[0].Filters.AddRange(new List <FilterType> {
FilterType.RMxN, FilterType.LowDepth, FilterType.LowVariantFrequency
});
variants[1].Filters.AddRange(new List <FilterType> {
FilterType.IndelRepeatLength, FilterType.LowVariantQscore, FilterType.StrandBias
});
var originalHeader = new List <string>
{
"##fileformat=VCFv4.1",
"##fileDate=20160620",
"##source=Pisces 1.0.0.0",
"##Pisces_cmdline=\"-B KRAS_42_S1.bam -g -MinimumFrequency 0.01 -MinBaseCallQuality 21 -MaxVariantQScore 100 -MinCoverage 300 -MaxAcceptableStrandBiasFilter 0.5 -MinVariantQScore 20 -VariantQualityFilter 20 -gVCF true -CallMNVs True -out \\myout",
"##reference=WholeGenomeFASTA",
"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Total Depth\">",
"##FILTER=<ID=q20,Description=\"Quality score less than 20\">",
"##FILTER=<ID=SB,Description=\"Variant strand bias too high\">",
"##FILTER=<ID=R8,Description=\"Indel repeat greater than or equal to 8\">",
"##FILTER=<ID=R5x9,Description=\"Repeats of part or all of the variant allele (max repeat length 5) in the reference greater than or equal to 9\">",
"##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">",
"##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">",
"#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT HD700n560_miseq1_S7.bam"
};
var writer = new PhasedVcfWriter(outputFilePath, config, new VcfWriterInputContext(), originalHeader, null);
writer.WriteHeader();
writer.Write(variants);
writer.Dispose();
VcfReader reader = new VcfReader(outputFilePath);
List <string> writtenHeader = reader.HeaderLines;
reader.Dispose();
var expectedHeader1 = new List <string>
{
"##fileformat=VCFv4.1",
"##fileDate=20160620",
"##source=Pisces 1.0.0.0",
"##Pisces_cmdline=\"-B KRAS_42_S1.bam -g -MinimumFrequency 0.01 -MinBaseCallQuality 21 -MaxVariantQScore 100 -MinCoverage 300 -MaxAcceptableStrandBiasFilter 0.5 -MinVariantQScore 20 -VariantQualityFilter 20 -gVCF true -CallMNVs True -out \\myout",
"##VariantPhaser=Scylla 1.0.0.0",
"##reference=WholeGenomeFASTA",
"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Total Depth\">",
"##FILTER=<ID=q20,Description=\"Quality score less than 20\">",
"##FILTER=<ID=SB,Description=\"Variant strand bias too high\">",
"##FILTER=<ID=R8,Description=\"Indel repeat greater than or equal to 8\">",
"##FILTER=<ID=R5x9,Description=\"Repeats of part or all of the variant allele (max repeat length 5) in the reference greater than or equal to 9\">",
"##FILTER=<ID=q30,Description=\"Quality score less than 30, by Scylla\">",
"##FILTER=<ID=LowDP,Description=\"Low coverage (DP tag), therefore no genotype called\">",
"##FILTER=<ID=LowVariantFreq,Description=\"Variant frequency less than 0.0070\">",
"##FILTER=<ID=MultiAllelicSite,Description=\"Variant does not conform to diploid model\">",
"##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">",
"##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">",
"#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT HD700n560_miseq1_S7.bam"
};
Assert.Equal(expectedHeader1.Count, writtenHeader.Count);
for (int i = 0; i < expectedHeader1.Count; i++)
{
if (expectedHeader1[i].StartsWith("##VariantPhaser="))
{
Assert.True(writtenHeader[i].StartsWith("##VariantPhaser="));
continue;
}
Assert.Equal(expectedHeader1[i], writtenHeader[i]);
}
config = new VcfWriterConfig
{
DepthFilterThreshold = 500,
VariantQualityFilterThreshold = 22,
FrequencyFilterThreshold = 0.007f,
EstimatedBaseCallQuality = 23,
PloidyModel = PloidyModel.Somatic,
};
originalHeader = new List <string>
{
"##fileformat=VCFv4.1",
"##fileDate=20160620",
"##source=Pisces 1.0.0.0",
"##Pisces_cmdline=\"-B KRAS_42_S1.bam -g -MinimumFrequency 0.01 -MinBaseCallQuality 21 -MaxVariantQScore 100 -MinCoverage 300 -MaxAcceptableStrandBiasFilter 0.5 -MinVariantQScore 20 -VariantQualityFilter 20 -gVCF true -CallMNVs True -out \\myout",
"##reference=WholeGenomeFASTA",
"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Total Depth\">",
"##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">",
"##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">",
"#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT HD700n560_miseq1_S7.bam"
};
writer = new PhasedVcfWriter(outputFilePath, config, new VcfWriterInputContext(), originalHeader, null);
var expectedHeader2 = new List <string>
{
"##fileformat=VCFv4.1",
"##fileDate=20160620",
"##source=Pisces 1.0.0.0",
"##Pisces_cmdline=\"-B KRAS_42_S1.bam -g -MinimumFrequency 0.01 -MinBaseCallQuality 21 -MaxVariantQScore 100 -MinCoverage 300 -MaxAcceptableStrandBiasFilter 0.5 -MinVariantQScore 20 -VariantQualityFilter 20 -gVCF true -CallMNVs True -out \\myout",
"##VariantPhaser=Scylla 1.0.0.0",
"##reference=WholeGenomeFASTA",
"##INFO=<ID=DP,Number=1,Type=Integer,Description=\"Total Depth\">",
"##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">",
"##FORMAT=<ID=GQ,Number=1,Type=Integer,Description=\"Genotype Quality\">",
"##FILTER=<ID=q22,Description=\"Quality score less than 22\">",
"##FILTER=<ID=LowDP,Description=\"Low coverage (DP tag), therefore no genotype called\">",
"##FILTER=<ID=LowVariantFreq,Description=\"Variant frequency less than 0.0070\">",
"#CHROM POS ID REF ALT QUAL FILTER INFO FORMAT HD700n560_miseq1_S7.bam",
};
variants[0].Filters = new List <FilterType>();
variants[1].Filters = new List <FilterType>();
writer.WriteHeader();
writer.Write(variants);
writer.Dispose();
reader = new VcfReader(outputFilePath);
writtenHeader = reader.HeaderLines;
reader.Dispose();
Assert.Equal(expectedHeader2.Count, writtenHeader.Count);
for (int i = 0; i < expectedHeader2.Count; i++)
{
if (expectedHeader2[i].StartsWith("##VariantPhaser="))
{
Assert.True(writtenHeader[i].StartsWith("##VariantPhaser="));
continue;
}
Assert.Equal(expectedHeader2[i], writtenHeader[i]);
}
}
public void Write()
{
//write a normal vcf
var writer = InitializeWriter(false);
//Writer should order the variants by chrom, coord, ref, then alt.
var variants = new List <CalledAllele>
{
PhasedVariantTestUtilities.CreateDummyAllele("chrX", 123, "A", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr10", 124, "A", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr9", 123, "T", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr9", 123, "T", "A", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr9", 123, "A", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr8", 123, "A", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chr9", 124, "A", "C", 1000, 156),
PhasedVariantTestUtilities.CreateDummyAllele("chrM", 123, "A", "C", 1000, 156),
};
// Order should be:
var expected = new List <string> {
"chrM\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr8\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr9\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr9\t123\t.\tT\tA\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr9\t123\t.\tT\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr9\t124\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr10\t124\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chrX\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000"
};
writer.Write(variants);
writer.Dispose();
Assert.Throws <Exception>(() => writer.WriteHeader());
Assert.Throws <Exception>(() => writer.Write(new List <CalledAllele> {
PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "G", 1000, 156)
}));
writer.Dispose();
var fileLines = File.ReadAllLines(_outputFile);
Assert.Equal(variants.Count, fileLines.Length);
for (int i = 0; i < expected.Count; i++)
{
Assert.Equal(expected[i], fileLines[i]);
}
//write a crushed vcf
writer = InitializeWriter(true);
writer.Write(variants);
writer.Dispose();
fileLines = File.ReadAllLines(_outputFile);
expected = new List <string> {
"chrM\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr8\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr9\t123\t.\tA\tC,A,C\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr9\t124\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chr10\t124\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000",
"chrX\t123\t.\tA\tC\t100\tPASS\tDP=1000\tGT:GQ:AD:DP:VF:NL:SB:NC\t0/1:0:844,156:1000:0.156:0:0.0000:0.0000"
};
Assert.Equal(6, fileLines.Length); //only variants at diff positions
for (int i = 0; i < expected.Count; i++)
{
Assert.Equal(expected[i], fileLines[i]);
}
}
public void GetAcceptedVariants_MergeVariants()
{
var originalVcfVariant = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "T", 1000, 156);
var originalVcfVariant2 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 124, "A", "T", 1000, 156);
var originalVcfVariant3 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 234, "A", "T", 1000, 156);
var originalVcfVariant4 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 234, "A", "C", 1000, 156);
var vcfVariant0asRef = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 123,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }
}
},
};
var vcfVariant3asRef = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 234,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }
}
},
};
var vcfVariant2asNull = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "./." }
}
},
};
var newMNV = new CalledAllele()
{
Chromosome = "chr1",
Coordinate = 229,
Reference = "AA",
Alternate = "T",
Genotype = Genotype.HeterozygousAltRef
};
var stagedVcfVariants = new List <CalledAllele> {
originalVcfVariant, originalVcfVariant2, originalVcfVariant3, originalVcfVariant4
};
var variantsUsedByCaller2 = new List <CalledAllele>()
{
originalVcfVariant, originalVcfVariant2, originalVcfVariant3
};
var nbhd = new Mock <IVcfNeighborhood>();
nbhd.Setup(n => n.GetOriginalVcfVariants()).Returns(variantsUsedByCaller2.ToList());
var stagedCalledMNVs2 = new Dictionary <int, List <CalledAllele> >()
{
{ newMNV.Coordinate, new List <CalledAllele>()
{
newMNV
} }
};
nbhd.Setup(n => n.CalledVariants).Returns(stagedCalledMNVs2);
// If one has been sucked up all the way, we should output it as a nocall
// (but we have to statge it already as a no call allready, becasue the merger can't do the conversion.
var stagedCalledRefs2 = new Dictionary <int, CalledAllele>()
{
{ 123, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 123, Chromosome = "chr1", Reference = "A", Alternate = "."
} },
{ 124, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 124, Chromosome = "chr1", Reference = "A", Alternate = ".", Genotype = Genotype.RefLikeNoCall
} },
{ 234, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 234, Chromosome = "chr1", Reference = "A", Alternate = ".", Genotype = Genotype.HomozygousRef
} }
};
nbhd.Setup(n => n.CalledRefs).Returns(stagedCalledRefs2);
var accepted = VcfMerger.GetMergedListOfVariants(nbhd.Object, stagedVcfVariants.ToList());
Assert.Equal(5, accepted.Count);
CheckVariantsMatch(vcfVariant0asRef, accepted[0]);
CheckVariantsMatch(vcfVariant2asNull, accepted[1]);
CheckVariantsMatch(newMNV, accepted[2]);
CheckVariantsMatch(vcfVariant3asRef, accepted[3]);
CheckVariantsMatch(originalVcfVariant4, accepted[4]);
}
public void WriteANbhd()
{
var outputFilePath = Path.Combine(UnitTestPaths.TestDataDirectory, "PhasedVcfFileNbhdWriterTest.vcf");
var inputFilePath = Path.Combine(UnitTestPaths.TestDataDirectory, "MergerInput.vcf");
var expectedFilePath = Path.Combine(UnitTestPaths.TestDataDirectory, "MergerOutput.vcf");
File.Delete(outputFilePath);
var context = new VcfWriterInputContext
{
CommandLine = new[] { "myCommandLine" },
SampleName = "mySample",
ReferenceName = "myReference",
ContigsByChr = new List <Tuple <string, long> >
{
new Tuple <string, long>("chr1", 10001),
new Tuple <string, long>("chrX", 500)
}
};
var config = new VcfWriterConfig
{
DepthFilterThreshold = 500,
VariantQualityFilterThreshold = 30,
FrequencyFilterThreshold = 0.007f,
ShouldOutputNoCallFraction = true,
ShouldOutputStrandBiasAndNoiseLevel = true,
EstimatedBaseCallQuality = 23,
PloidyModel = PloidyModel.Somatic,
AllowMultipleVcfLinesPerLoci = true
};
var writer = new PhasedVcfWriter(outputFilePath, config, new VcfWriterInputContext(), new List <string>()
{
}, null);
var reader = new VcfReader(inputFilePath, true);
//set up the original variants
var originalVcfVariant1 = PhasedVariantTestUtilities.CreateDummyAllele("chr2", 116380048, "A", "New", 1000, 156);
var originalVcfVariant2 = PhasedVariantTestUtilities.CreateDummyAllele("chr2", 116380048, "AAA", "New", 1000, 156);
var originalVcfVariant4 = PhasedVariantTestUtilities.CreateDummyAllele("chr7", 116380051, "A", "New", 1000, 156);
var originalVcfVariant5 = PhasedVariantTestUtilities.CreateDummyAllele("chr7", 116380052, "AC", "New", 1000, 156);
var vs1 = new VariantSite((originalVcfVariant1));
var vs2 = new VariantSite((originalVcfVariant2));
var vs4 = new VariantSite((originalVcfVariant4));
var vs5 = new VariantSite((originalVcfVariant5));
//have to replace variants at positon 116380048 (we call two new MNVS here)
var nbhd1 = new VcfNeighborhood(new VariantCallingParameters(), "chr2", vs1, vs2, "");
nbhd1.SetRangeOfInterest();
//have to replace variants at positon 116380051 and 52 (we call one new MNV at 51)
var nbhd2 = new VcfNeighborhood(new VariantCallingParameters(), "chr7", vs4, vs5, "");
nbhd2.SetRangeOfInterest();
VcfMerger merger = new VcfMerger(reader);
List <CalledAllele> allelesPastNbh = new List <CalledAllele>();
nbhd1.CalledVariants = new Dictionary <int, List <CalledAllele> > {
{ originalVcfVariant1.Coordinate, new List <CalledAllele> {
originalVcfVariant1, originalVcfVariant2
} }
};
nbhd2.CalledVariants = new Dictionary <int, List <CalledAllele> > {
{ originalVcfVariant4.Coordinate, new List <CalledAllele> {
originalVcfVariant4
} }
};
allelesPastNbh = merger.WriteVariantsUptoChr(writer, allelesPastNbh, nbhd1.ReferenceName);
allelesPastNbh = merger.WriteVariantsUptoIncludingNbhd(nbhd1, writer, allelesPastNbh);
allelesPastNbh = merger.WriteVariantsUptoChr(writer, allelesPastNbh, nbhd2.ReferenceName);
allelesPastNbh = merger.WriteVariantsUptoIncludingNbhd(nbhd2, writer, allelesPastNbh);
merger.WriteRemainingVariants(writer, allelesPastNbh);
writer.Dispose();
var expectedLines = File.ReadLines(expectedFilePath).ToList();
var outputLines = File.ReadLines(outputFilePath).ToList();
Assert.Equal(expectedLines.Count(), outputLines.Count());
for (int i = 0; i < expectedLines.Count; i++)
{
Assert.Equal(expectedLines[i], outputLines[i]);
}
}
public void GetAcceptedVariants_MergeNull()
{
var originalVcfVariant = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 123, "A", "T", 1000, 156);
var originalVcfVariant2 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 124, "A", "T", 1000, 156);
var originalVcfVariant3 = PhasedVariantTestUtilities.CreateDummyAllele("chr1", 234, "A", "T", 1000, 156);
var stagedVcfVariants = new List <CalledAllele> {
originalVcfVariant, originalVcfVariant2, originalVcfVariant3
};
var variantsUsedByCaller = new List <CalledAllele>()
{
originalVcfVariant, originalVcfVariant2
};
var stagedCalledMNV = new CalledAllele(AlleleCategory.Snv)
{
Chromosome = "chr1", Coordinate = 123, Reference = "A", Alternate = "T"
};
var stagedCalledMNVs = new Dictionary <int, List <CalledAllele> >()
{
{ stagedCalledMNV.Coordinate, new List <CalledAllele>()
{
stagedCalledMNV
} }
};
var stagedCalledRefs = new Dictionary <int, CalledAllele>()
{
{ 123, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 123, Chromosome = "chr1", Reference = "A", Alternate = "."
} },
{ 124, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 124, Chromosome = "chr1", Reference = "A", Alternate = "."
} }
};
//since there is an alt at position 124 ( a call of 156 alt / 1000 total, that means 844 original ref calls.
//Of which we said, 100 will get sucked up. So that leaves 744 / 1000 calls for a reference.
//So, we can still make a confident ref call.
var mockNeighborhood = new Mock <IVcfNeighborhood>();
mockNeighborhood.Setup(n => n.GetOriginalVcfVariants()).Returns(variantsUsedByCaller.ToList());
mockNeighborhood.Setup(n => n.CalledVariants).Returns(stagedCalledMNVs);
mockNeighborhood.Setup(n => n.CalledRefs).Returns(stagedCalledRefs);
var accepted = VcfMerger.GetMergedListOfVariants(mockNeighborhood.Object, stagedVcfVariants.ToList());
Assert.Equal(3, accepted.Count);
var vcfVariant2asNull = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "0/0" }, { "DP", "1000" }, { "AD", "744" }
}
},
};
CheckVariantsMatch(originalVcfVariant, accepted[0]);
CheckVariantsMatch(vcfVariant2asNull, accepted[1]);
CheckVariantsMatch(originalVcfVariant3, accepted[2]);
//re-stage the MNVs
var stagedCalledMNVs2 = new Dictionary <int, List <CalledAllele> >()
{
{ stagedCalledMNV.Coordinate, new List <CalledAllele>()
{
stagedCalledMNV
} }
};
mockNeighborhood.Setup(n => n.CalledVariants).Returns(stagedCalledMNVs2);
// If one has been sucked up all the way, we should output it as a nocall
// (but we have to statge it already as a no call allready, becasue the merger can't do the conversion.
var stagedCalledRefs2 = new Dictionary <int, CalledAllele>()
{
{ 123, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 123, Chromosome = "chr1", Reference = "A", Alternate = "."
} },
{ 124, new CalledAllele(AlleleCategory.Reference)
{
Coordinate = 124, Chromosome = "chr1", Reference = "A", Alternate = ".", Genotype = Genotype.RefLikeNoCall
} }
};
mockNeighborhood.Setup(n => n.CalledRefs).Returns(stagedCalledRefs2);
accepted = VcfMerger.GetMergedListOfVariants(mockNeighborhood.Object, stagedVcfVariants);
Assert.Equal(3, accepted.Count);
vcfVariant2asNull = new VcfVariant()
{
ReferenceName = "chr1",
ReferencePosition = 124,
ReferenceAllele = "A",
VariantAlleles = new[] { "." },
Genotypes = new List <Dictionary <string, string> >()
{
new Dictionary <string, string>()
{
{ "GT", "./." }
}
},
};
CheckVariantsMatch(originalVcfVariant, accepted[0]);
CheckVariantsMatch(vcfVariant2asNull, accepted[1]);
CheckVariantsMatch(originalVcfVariant3, accepted[2]);
}