private static CalledAllele ProcessDeletion(CalledAllele deletionVar, CalledAllele variant, AlleleReader reader,
RecalibratedVariantsCollection snvLoci, RecalibratedVariantsCollection indelLoci)
{
CalledAllele lastVar;
for (int i = 1; i < deletionVar.ReferenceAllele.Length; i++)
{
if (ShouldSkipVariant(variant))
{
continue;
}
if (variant.HasAnAltAllele)
{
if (GetVariantType(variant) == VariantType.Snv)
{
snvLoci.AddLocus(variant);
}
else
{
indelLoci.AddLocus(variant);
}
}
lastVar = variant;
variant = new CalledAllele();
GetNextUncrushedAllele(reader, out variant);
// If there is multiallelic variant inside deletion, ignore locus
if (variant.IsCoLocatedAllele(lastVar) && lastVar.HasAnAltAllele)
{
if (GetVariantType(lastVar) == VariantType.Snv)
{
snvLoci.RemoveLastEntry();
}
else
{
indelLoci.RemoveLastEntry();
}
}
while (variant.ReferencePosition == lastVar.ReferencePosition &&
variant.Chromosome == lastVar.Chromosome)
{
lastVar = variant;
variant = new CalledAllele();
GetNextUncrushedAllele(reader, out variant);
}
if (variant.ReferencePosition > deletionVar.ReferencePosition + deletionVar.ReferenceAllele.Length - 1 &&
variant.Chromosome == deletionVar.Chromosome)
{
break;
}
}
return(variant);
}
public static List <RecalibratedVariantsCollection> GetVariantFrequencies(string vcfIn)
{
CalledAllele variant = new CalledAllele();
CalledAllele lastVar = new CalledAllele();
var snvLoci = new RecalibratedVariantsCollection();
var indelLoci = new RecalibratedVariantsCollection();
var alleleReader = new AlleleReader(vcfIn);
// Check headers
CheckHeader(alleleReader);
using (alleleReader)
{
while (GetNextUncrushedAllele(alleleReader, out variant))
{
try
{
// Check if multiallelic
if (variant.IsCoLocatedAllele(lastVar))
{
variant = ProcessMultiAllelicVariant(lastVar, variant, alleleReader, snvLoci, indelLoci);
}
// Check if within deletion
if (lastVar.ReferenceAllele != null && variant != null &&
lastVar.ReferenceAllele.Length > 1 &&
lastVar.Genotype != Genotype.HomozygousRef &&
variant.ReferencePosition == lastVar.ReferencePosition + 1)
{
variant = ProcessDeletion(lastVar, variant, alleleReader, snvLoci, indelLoci);
}
// this happens if last variants in file are multi-allelic or a deletion
if (variant == null)
{
break;
}
if (ShouldSkipVariant(variant) || !variant.Chromosome.Any(char.IsDigit))
{
continue;
}
var variantType = GetVariantType(variant);
if (variantType == VariantType.NoVariant)
{
snvLoci.AddLocus(variant);
indelLoci.AddLocus(variant);
}
else if (variantType == VariantType.Snv)
{
snvLoci.AddLocus(variant);
}
else if (variantType == VariantType.Indel)
{
indelLoci.AddLocus(variant);
}
lastVar = variant;
variant = new CalledAllele();
}
catch (Exception ex)
{
Logger.WriteToLog(string.Format("Fatal error processing vcf; Check {0}, position {1}. Exception: {2}",
variant.Chromosome, variant.ReferencePosition, ex));
throw;
}
}
}
return(new List <RecalibratedVariantsCollection> {
snvLoci, indelLoci
});
}
private static CalledAllele ProcessMultiAllelicVariant(CalledAllele lastVar, CalledAllele variant,
AlleleReader reader, RecalibratedVariantsCollection snvLoci, RecalibratedVariantsCollection indelLoci)
{
// SNPs and insertions are processed the same way--check and see if the two major variants have total VF > 0.8
// 1/2 variants should not be used in the modeling because this model is for alignment bias of the reference
List <CalledAllele> variants = new List <CalledAllele>()
{
lastVar, variant
};
// Use VF to find the two major variants
List <double> vf = new List <double>()
{
GetAlternateAlleleFrequency(lastVar),
GetAlternateAlleleFrequency(variant)
};
int[] topIndices = new int[] { 0, 1 };
Array.Sort(vf.ToArray(), topIndices);
Array.Reverse(topIndices);
// Keep track of ref vf
// NB: refVf is only approximate and could be negative if ref alleles are different lengths
double refVf = 1 - vf[0] - vf[1];
int currIndex = 2;
while (GetNextUncrushedAllele(reader, out variant))
{
if (!variant.IsCoLocatedAllele(lastVar))
{
break;
}
// Handle variant and update top 2 in VF
variants.Add(variant);
double newVf = GetAlternateAlleleFrequency(variant);
vf.Add(newVf);
if (newVf > vf[topIndices[0]])
{
topIndices[1] = topIndices[0];
topIndices[0] = currIndex;
}
else if (newVf > vf[topIndices[1]])
{
topIndices[1] = currIndex;
}
refVf = refVf - vf[currIndex];
currIndex++;
lastVar = variant;
variant = new CalledAllele();
}
// Remove the last entry
if (GetVariantType(variants[0]) == VariantType.Snv)
{
snvLoci.RemoveLastEntry();
}
else if (GetVariantType(variants[0]) == VariantType.Indel)
{
indelLoci.RemoveLastEntry();
}
// Determine type of entry
RecalibratedVariantsCollection currLoci;
if (GetVariantType(variants[topIndices[0]]) == VariantType.Snv &&
GetVariantType(variants[topIndices[1]]) == VariantType.Snv)
{
currLoci = snvLoci;
}
else if (GetVariantType(variants[topIndices[0]]) == VariantType.Indel &&
GetVariantType(variants[topIndices[1]]) == VariantType.Indel)
{
currLoci = indelLoci;
}
else // mixed type
{
return(variant);
}
if ((currLoci == snvLoci && // multiallelic check for SNVs
(GetAlternateAlleleFrequency(variants[topIndices[0]]) + GetAlternateAlleleFrequency(variants[topIndices[1]]) > MultiAllelicThreshold || // top 2 VF > 0.8
GetAlternateAlleleFrequency(variants[topIndices[0]]) + refVf > MultiAllelicThreshold) ||
currLoci == indelLoci) &&
!ShouldSkipVariant(variants[topIndices[0]]) && // should not skip variant
!(vf[topIndices[0]] > HetThreshold && vf[topIndices[0]] < HomAltThreshold && vf[topIndices[1]] > HetThreshold)) // not 1/2
{
currLoci.AddLocus(variants[topIndices[0]]);
}
return(variant);
}