public CallableNeighborhood(VcfNeighborhood vcfNeighborhood, VariantCallingParameters variantCallingParams, ChrReference chrReference = null)
{
//housekeeping
_nbhdGTcalculator = GenotypeCreator.CreateGenotypeCalculator(variantCallingParams.PloidyModel, variantCallingParams.MinimumFrequencyFilter,
variantCallingParams.MinimumCoverage,
variantCallingParams.DiploidSNVThresholdingParameters,
variantCallingParams.DiploidINDELThresholdingParameters,
variantCallingParams.AdaptiveGenotypingParameters,
variantCallingParams.MinimumGenotypeQScore, variantCallingParams.MaximumGenotypeQScore, variantCallingParams.TargetLODFrequency);
_vcfNeighborhood = vcfNeighborhood;
_acceptedPhasedVariants = new List <CalledAllele>();
_rejectedPhasedVariants = new List <CalledAllele>();
UsedRefCountsLookup = new Dictionary <int, SuckedUpRefRecord>();
MaxQScore = variantCallingParams.MaximumVariantQScore;
//prep vcf nbhd for use, so we know the final range of loci in play
vcfNeighborhood.OrderVariantSitesByFirstTrueStartPosition();
vcfNeighborhood.SetRangeOfInterest();
//set reference bases here, then let go of the chr
if ((chrReference == null) || (chrReference.Sequence == null)) //be gentle if they did not include a ref genome
{
NbhdReferenceSequenceSubstring = new String('R', vcfNeighborhood.LastPositionOfInterestWithLookAhead - vcfNeighborhood.FirstPositionOfInterest);
}
else
{
NbhdReferenceSequenceSubstring = chrReference.Sequence.Substring(vcfNeighborhood.FirstPositionOfInterest - 1, vcfNeighborhood.LastPositionOfInterestWithLookAhead - vcfNeighborhood.FirstPositionOfInterest);
}
}
public AlleleCaller(VariantCallerConfig config, ChrIntervalSet intervalSet = null,
IVariantCollapser variantCollapser = null, ICoverageCalculator coverageCalculator = null)
{
_config = config;
_intervalSet = intervalSet;
_collapser = variantCollapser;
_coverageCalculator = coverageCalculator ?? new CoverageCalculator();
_genotypeCalculator = config.GenotypeCalculator;
}
public VcfNeighborhood(VariantCallingParameters variantCallingParams, string refName, VariantSite vs1, VariantSite vs2, string interveningRef)
{
_nbhdGTcalculator = GenotypeCreator.CreateGenotypeCalculator(variantCallingParams.PloidyModel, variantCallingParams.MinimumFrequencyFilter,
variantCallingParams.MinimumCoverage,
variantCallingParams.DiploidThresholdingParameters,
variantCallingParams.MinimumGenotpyeQScore, variantCallingParams.MaximumGenotpyeQScore);
VcfVariantSites = new List <VariantSite>();
_referenceName = refName;
_acceptedPhasedVariants = new List <CalledAllele>();
_rejectedPhasedVariants = new List <CalledAllele>();
UsedRefCountsLookup = new Dictionary <int, int>();
AddVariantSite(vs1, vs1.VcfReferenceAllele.Substring(0, 1));
AddVariantSite(vs2, interveningRef);
SetID();
}
private void ExecuteDiploidMultiAllelicSiteGenotypeTest(IGenotypeCalculator GTC)
{
//test cases:
// (1) SNP + indel + indel
// (2) indel + SNP + SNP
// (3) 3 indels (OK, one is low VF)
// (4) 3 indels (OK, one is low VF)
// (4) 3 indels - ploidy violation
Genotype expectedGenotype = Genotype.HeterozygousAlt1Alt2;
int expectedNumAllelesToPrune = 1;
// (1) SNP + indel + indel
// should be 1/2 with the lowest freq thrown out
List <float> refFrequencies = new List <float>()
{
0.40F, 0.60F, 0.90F
};
List <float> altFrequencies = new List <float>()
{
0.60F, 0.40F, 0.10F
};
List <string> refAllele = new List <string>()
{
"A", "A", "ACT"
};
List <string> altAllele = new List <string>()
{
"C", "AGGG", "A"
};
double coverage = 1000;
var alleles = new List <CalledAllele>();
for (int i = 0; i < 3; i++)
{
var variant = TestHelper.CreatePassingVariant(false);
variant.AlleleSupport = (int)(altFrequencies[i] * coverage);
variant.TotalCoverage = (int)coverage;
variant.ReferenceSupport = (int)(refFrequencies[i] * coverage);
variant.AlternateAllele = altAllele[i];
variant.ReferenceAllele = refAllele[i];
alleles.Add(variant);
}
alleles[1].Type = AlleleCategory.Insertion;
alleles[2].Type = AlleleCategory.Deletion;
//var GTC = new DiploidThresholdingGenotyper();
GTC.MinDepthToGenotype = 100;
var allelesToPrune = GTC.SetGenotypes(alleles);
Assert.Equal(expectedNumAllelesToPrune, allelesToPrune.Count);
foreach (var allele in alleles)
{
Assert.Equal(expectedGenotype, allele.Genotype);
Assert.Equal(0, allele.Filters.Count());
}
Assert.Equal(allelesToPrune[0].ReferenceAllele, "ACT");
Assert.Equal(allelesToPrune[0].AlternateAllele, "A");
Assert.Equal(allelesToPrune[0].Frequency, 0.10F);
// (2) indel + SNP + SNP
// should be 1/2 with the lowest freq thrown out
refFrequencies = new List <float>()
{
0.40F, 0.20F, 0.20F
};
altFrequencies = new List <float>()
{
0.60F, 0.10F, 0.40F
};
refAllele = new List <string>()
{
"A", "A", "A"
};
altAllele = new List <string>()
{
"ACCAT", "G", "C"
};
alleles = new List <CalledAllele>();
for (int i = 0; i < 3; i++)
{
var variant = TestHelper.CreatePassingVariant(false);
variant.AlleleSupport = (int)(altFrequencies[i] * coverage);
variant.TotalCoverage = (int)coverage;
variant.ReferenceSupport = (int)(refFrequencies[i] * coverage);
variant.AlternateAllele = altAllele[i];
variant.ReferenceAllele = refAllele[i];
alleles.Add(variant);
}
alleles[0].Type = AlleleCategory.Insertion;
GTC.MinDepthToGenotype = 100;
allelesToPrune = GTC.SetGenotypes(alleles);
Assert.Equal(expectedNumAllelesToPrune, allelesToPrune.Count);
foreach (var allele in alleles)
{
Assert.Equal(expectedGenotype, allele.Genotype);
Assert.Equal(0, allele.Filters.Count());
}
Assert.Equal(allelesToPrune[0].ReferenceAllele, "A");
Assert.Equal(allelesToPrune[0].AlternateAllele, "G");
Assert.Equal(allelesToPrune[0].Frequency, 0.10F);
// (3) 3 indels (OK)
// should be 1/2 with the lowest freq thrown out
refFrequencies = new List <float>()
{
0.40F, 0.90F, 0.60F
};
altFrequencies = new List <float>()
{
0.60F, 0.10F, 0.40F
};
refAllele = new List <string>()
{
"A", "ACT", "A"
};
altAllele = new List <string>()
{
"ACCAT", "A", "CC"
};
alleles = new List <CalledAllele>();
for (int i = 0; i < 3; i++)
{
var variant = TestHelper.CreatePassingVariant(false);
variant.AlleleSupport = (int)(altFrequencies[i] * coverage);
variant.TotalCoverage = (int)coverage;
variant.ReferenceSupport = (int)(refFrequencies[i] * coverage);
variant.AlternateAllele = altAllele[i];
variant.ReferenceAllele = refAllele[i];
alleles.Add(variant);
}
alleles[0].Type = AlleleCategory.Insertion;
alleles[1].Type = AlleleCategory.Deletion;
alleles[2].Type = AlleleCategory.Insertion;
GTC.MinDepthToGenotype = 100;
allelesToPrune = GTC.SetGenotypes(alleles);
Assert.Equal(expectedNumAllelesToPrune, allelesToPrune.Count);
foreach (var allele in alleles)
{
Assert.Equal(expectedGenotype, allele.Genotype);
Assert.Equal(0, allele.Filters.Count());
}
Assert.Equal(allelesToPrune[0].ReferenceAllele, "ACT");
Assert.Equal(allelesToPrune[0].AlternateAllele, "A");
Assert.Equal(allelesToPrune[0].Frequency, 0.10F);
// (4) 3 indels - excused ploidy violation,
// b /c its possible to have these together in a diploid individual
// should be 1/2 with the lowest freq thrown out
refFrequencies = new List <float>()
{
0.60F, 0.60F, 0.60F
};
altFrequencies = new List <float>()
{
0.31F, 0.30F, 0.31F
};
refAllele = new List <string>()
{
"A", "ACT", "A"
};
altAllele = new List <string>()
{
"ACCAT", "A", "AC"
};
expectedGenotype = Genotype.HeterozygousAlt1Alt2;
alleles = new List <CalledAllele>();
for (int i = 0; i < 3; i++)
{
var variant = TestHelper.CreatePassingVariant(false);
variant.AlleleSupport = (int)(altFrequencies[i] * coverage);
variant.TotalCoverage = (int)coverage;
variant.ReferenceSupport = (int)(refFrequencies[i] * coverage);
variant.AlternateAllele = altAllele[i];
variant.ReferenceAllele = refAllele[i];
alleles.Add(variant);
}
alleles[0].Type = AlleleCategory.Insertion;
alleles[1].Type = AlleleCategory.Deletion;
alleles[2].Type = AlleleCategory.Insertion;
GTC.MinDepthToGenotype = 100;
allelesToPrune = GTC.SetGenotypes(alleles);
Assert.Equal(expectedNumAllelesToPrune, allelesToPrune.Count);
foreach (var allele in alleles)
{
Assert.Equal(expectedGenotype, allele.Genotype);
Assert.Equal(0, allele.Filters.Count);
}
Assert.Equal(allelesToPrune[0].ReferenceAllele, "ACT");
Assert.Equal(allelesToPrune[0].AlternateAllele, "A");
Assert.Equal(allelesToPrune[0].Frequency, 0.30F);
//
// (5) 3 SNPS - NOT excused ploidy violation,
// b /c its NOT possible to thave these together in a diploid individual
// should be ./. with the lowest freq thrown out
refFrequencies = new List <float>()
{
0.01F, 0.01F, 0.01F
};
altFrequencies = new List <float>()
{
0.31F, 0.30F, 0.31F
};
refAllele = new List <string>()
{
"A", "A", "A"
};
altAllele = new List <string>()
{
"C", "T", "G"
};
expectedGenotype = Genotype.Alt12LikeNoCall;
alleles = new List <CalledAllele>();
for (int i = 0; i < 3; i++)
{
var variant = TestHelper.CreatePassingVariant(false);
variant.AlleleSupport = (int)(altFrequencies[i] * coverage);
variant.TotalCoverage = (int)coverage;
variant.ReferenceSupport = (int)(refFrequencies[i] * coverage);
variant.AlternateAllele = altAllele[i];
variant.ReferenceAllele = refAllele[i];
alleles.Add(variant);
}
alleles[0].Type = AlleleCategory.Snv;
alleles[1].Type = AlleleCategory.Snv;
alleles[2].Type = AlleleCategory.Snv;
//GTC = new DiploidThresholdingGenotyper();
GTC.MinDepthToGenotype = 100;
allelesToPrune = GTC.SetGenotypes(alleles);
Assert.Equal(expectedNumAllelesToPrune, allelesToPrune.Count);
foreach (var allele in alleles)
{
Assert.Equal(expectedGenotype, allele.Genotype);
Assert.Equal(FilterType.MultiAllelicSite, allele.Filters[0]);
}
Assert.Equal(allelesToPrune[0].ReferenceAllele, "A");
Assert.Equal(allelesToPrune[0].AlternateAllele, "T");
Assert.Equal(allelesToPrune[0].Frequency, 0.30F);
}