// ReSharper restore InconsistentNaming
public IntermediateSampleFields(string[] vcfColumns, FormatIndices formatIndices, string[] sampleCols)
{
VcfRefAllele = vcfColumns[VcfCommon.RefIndex];
AltAlleles = vcfColumns[VcfCommon.AltIndex].Split(',');
FormatIndices = formatIndices;
SampleColumns = sampleCols;
(TAR, TIR) = GetLinkedIntegers(GetFirstValue(GetString(formatIndices.TAR, sampleCols)), GetFirstValue(GetString(formatIndices.TIR, sampleCols)));
(NR, NV) = GetLinkedIntegers(GetString(formatIndices.NR, sampleCols), GetString(formatIndices.NV, sampleCols));
RepeatNumberSpan = GetString(formatIndices.CI, sampleCols);
MajorChromosomeCount = GetInteger(GetString(formatIndices.MCC, sampleCols));
DenovoQuality = GetFloat(GetString(formatIndices.DQ, sampleCols));
MAD = GetIntegers(GetString(formatIndices.MAD, sampleCols));
SCH = GetString(formatIndices.SCH, sampleCols);
PLG = GetIntegers(GetString(formatIndices.PLG, sampleCols));
PCN = GetIntegers(GetString(formatIndices.PCN, sampleCols));
DCS = GetStrings(GetString(formatIndices.DCS, sampleCols));
DID = GetStrings(GetString(formatIndices.DID, sampleCols));
DST = GetStrings(GetString(formatIndices.DST, sampleCols));
PCH = GetIntegers(GetString(formatIndices.PCH, sampleCols));
CHC = GetBool(GetString(formatIndices.CHC, sampleCols), "+");
AQ = GetFloat(GetString(formatIndices.AQ, sampleCols));
LQ = GetFloat(GetString(formatIndices.LQ, sampleCols));
VF = GetDouble(GetString(formatIndices.VF, sampleCols));
(CopyNumber, RepeatNumber) = GetCopyNumber(GetString(formatIndices.CN, sampleCols), vcfColumns[VcfCommon.AltIndex].Contains("STR"));
(ACount, CCount, GCount, TCount, TotalAlleleCount) = GetAlleleCounts(
GetString(formatIndices.AU, sampleCols), GetString(formatIndices.CU, sampleCols),
GetString(formatIndices.GU, sampleCols), GetString(formatIndices.TU, sampleCols));
}
internal static ISample ExtractSample(string sampleColumn, FormatIndices formatIndices, ISimplePosition simplePosition,
IMitoHeteroplasmyProvider mitoHeteroplasmyProvider, LegacySampleFieldExtractor legacyExtractor = null, bool enableDq = false)
{
// sanity check: make sure we have a format column
if (string.IsNullOrEmpty(sampleColumn))
{
return(Sample.EmptySample);
}
string[] sampleColumns = sampleColumn.OptimizedSplit(':', formatIndices.NumColumns);
if (sampleColumns.Length == 1 && sampleColumns[0] == ".")
{
return(Sample.EmptySample);
}
sampleColumns.NormalizeNulls();
if (legacyExtractor != null)
{
return(legacyExtractor.ExtractSample(sampleColumn));
}
int[] alleleDepths = sampleColumns.GetString(formatIndices.AD).GetIntegers();
float?artifactAdjustedQualityScore = sampleColumns.GetString(formatIndices.AQ).GetFloat();
int? copyNumber = sampleColumns.GetString(formatIndices.CN).GetInteger();
string[] diseaseAffectedStatuses = sampleColumns.GetString(formatIndices.DST).GetStrings();
bool failedFilter = sampleColumns.GetString(formatIndices.FT).GetFailedFilter();
string genotype = sampleColumns.GetString(formatIndices.GT);
int? genotypeQuality = sampleColumns.GetString(formatIndices.GQ).GetInteger();
bool isDeNovo = sampleColumns.GetString(formatIndices.DN).IsDeNovo();
double? deNovoQuality = enableDq? sampleColumns.GetString(formatIndices.DQ).GetDouble():null;
float? likelihoodRatioQualityScore = sampleColumns.GetString(formatIndices.LQ).GetFloat();
int[] pairedEndReadCounts = sampleColumns.GetString(formatIndices.PR).GetIntegers();
int[] repeatUnitCounts = sampleColumns.GetString(formatIndices.REPCN).GetIntegers('/');
int[] splitReadCounts = sampleColumns.GetString(formatIndices.SR).GetIntegers();
int? totalDepth = sampleColumns.GetString(formatIndices.DP).GetInteger();
double?variantFrequency = sampleColumns.GetString(formatIndices.VF).GetDouble();
int? minorHaplotypeCopyNumber = sampleColumns.GetString(formatIndices.MCN).GetInteger();
double?somaticQuality = sampleColumns.GetString(formatIndices.SQ).GetDouble();
int? binCount = sampleColumns.GetString(formatIndices.BC).GetInteger();
double[] variantFrequencies = VariantFrequency.GetVariantFrequencies(variantFrequency, alleleDepths, simplePosition.AltAlleles.Length);
string[] mitoHeteroplasmyPercentiles = mitoHeteroplasmyProvider?.GetVrfPercentiles(simplePosition.Chromosome, simplePosition.Start,
simplePosition.AltAlleles, variantFrequencies)?.Select(x => x?.ToString("0.##") ?? "null").ToArray();
var isLoh = GetLoh(copyNumber, minorHaplotypeCopyNumber, genotype);
var sample = new Sample(alleleDepths, artifactAdjustedQualityScore, copyNumber, diseaseAffectedStatuses,
failedFilter, genotype, genotypeQuality, isDeNovo, deNovoQuality, likelihoodRatioQualityScore, pairedEndReadCounts,
repeatUnitCounts, splitReadCounts, totalDepth, variantFrequencies, minorHaplotypeCopyNumber, somaticQuality, isLoh, mitoHeteroplasmyPercentiles, binCount);
return(sample);
}
private static bool IsLegacyVariantCaller(FormatIndices formatIndices)
{
return(formatIndices.TAR != null ||
formatIndices.TIR != null ||
formatIndices.AU != null ||
formatIndices.GU != null ||
formatIndices.CU != null ||
formatIndices.TU != null ||
formatIndices.GQX != null ||
formatIndices.DPI != null ||
formatIndices.MCC != null);
}
internal static ISample ExtractSample(string sampleColumn, FormatIndices formatIndices, int numAltAlleles,
LegacySampleFieldExtractor legacyExtractor = null)
{
// sanity check: make sure we have a format column
if (string.IsNullOrEmpty(sampleColumn))
{
return(Sample.EmptySample);
}
string[] sampleColumns = sampleColumn.OptimizedSplit(':', formatIndices.NumColumns);
if (sampleColumns.Length == 1 && sampleColumns[0] == ".")
{
return(Sample.EmptySample);
}
sampleColumns.NormalizeNulls();
if (legacyExtractor != null)
{
return(legacyExtractor.ExtractSample(sampleColumn));
}
int[] alleleDepths = sampleColumns.GetString(formatIndices.AD).GetIntegers();
float?artifactAdjustedQualityScore = sampleColumns.GetString(formatIndices.AQ).GetFloat();
int? copyNumber = sampleColumns.GetString(formatIndices.CN).GetInteger();
string[] diseaseAffectedStatuses = sampleColumns.GetString(formatIndices.DST).GetStrings();
bool failedFilter = sampleColumns.GetString(formatIndices.FT).GetFailedFilter();
string genotype = sampleColumns.GetString(formatIndices.GT);
int? genotypeQuality = sampleColumns.GetString(formatIndices.GQ).GetInteger();
bool isDeNovo = sampleColumns.GetString(formatIndices.DN).IsDeNovo();
float? likelihoodRatioQualityScore = sampleColumns.GetString(formatIndices.LQ).GetFloat();
int[] pairedEndReadCounts = sampleColumns.GetString(formatIndices.PR).GetIntegers();
int[] repeatUnitCounts = sampleColumns.GetString(formatIndices.REPCN).GetIntegers('/');
int[] splitReadCounts = sampleColumns.GetString(formatIndices.SR).GetIntegers();
int? totalDepth = sampleColumns.GetString(formatIndices.DP).GetInteger();
double?variantFrequency = sampleColumns.GetString(formatIndices.VF).GetDouble();
int? minorHaplotypeCopyNumber = sampleColumns.GetString(formatIndices.MCN).GetInteger();
double?somaticQuality = sampleColumns.GetString(formatIndices.SQ).GetDouble();
double[] variantFrequencies = VariantFrequency.GetVariantFrequencies(variantFrequency, alleleDepths, numAltAlleles);
var isLoh = GetLoh(copyNumber, minorHaplotypeCopyNumber, genotype);
var sample = new Sample(alleleDepths, artifactAdjustedQualityScore, copyNumber, diseaseAffectedStatuses,
failedFilter, genotype, genotypeQuality, isDeNovo, likelihoodRatioQualityScore, pairedEndReadCounts,
repeatUnitCounts, splitReadCounts, totalDepth, variantFrequencies, minorHaplotypeCopyNumber, somaticQuality, isLoh);
return(sample);
}
internal static ISample[] ToSamples(this string[] vcfColumns, FormatIndices formatIndices, int numAltAlleles, bool isRepeatExpansion)
{
if (vcfColumns.Length < VcfCommon.MinNumColumnsSampleGenotypes)
{
return(null);
}
int numSamples = vcfColumns.Length - VcfCommon.MinNumColumnsSampleGenotypes + 1;
var samples = new ISample[numSamples];
formatIndices.Set(vcfColumns[VcfCommon.FormatIndex]);
for (int index = VcfCommon.GenotypeIndex; index < vcfColumns.Length; index++)
{
samples[index - VcfCommon.GenotypeIndex] = ExtractSample(vcfColumns[index], formatIndices, numAltAlleles, isRepeatExpansion);
}
return(samples);
}
internal static ISample[] ToSamples(this string[] vcfColumns, FormatIndices formatIndices, ISimplePosition simplePosition, IMitoHeteroplasmyProvider mitoHeteroplasmyProvider, bool enableDq = false)
{
if (vcfColumns.Length < VcfCommon.MinNumColumnsSampleGenotypes)
{
return(null);
}
int numSamples = vcfColumns.Length - VcfCommon.MinNumColumnsSampleGenotypes + 1;
var samples = new ISample[numSamples];
formatIndices.Set(vcfColumns[VcfCommon.FormatIndex]);
var legacySampleExtractor = IsLegacyVariantCaller(formatIndices) ? new LegacySampleFieldExtractor(vcfColumns, formatIndices) : null;
for (int index = VcfCommon.GenotypeIndex; index < vcfColumns.Length; index++)
{
samples[index - VcfCommon.GenotypeIndex] = ExtractSample(vcfColumns[index], formatIndices, simplePosition, mitoHeteroplasmyProvider, legacySampleExtractor, enableDq);
}
return(samples);
}
internal static ISample[] ToSamples(this string[] vcfColumns, FormatIndices formatIndices, int numAltAlleles)
{
if (vcfColumns.Length < VcfCommon.MinNumColumnsSampleGenotypes)
{
return(null);
}
int numSamples = vcfColumns.Length - VcfCommon.MinNumColumnsSampleGenotypes + 1;
var samples = new ISample[numSamples];
formatIndices.Set(vcfColumns[VcfCommon.FormatIndex]);
var legacySampleExtractor = IsLegacyVariantCaller(formatIndices) ? new LegacySampleFieldExtractor(vcfColumns, formatIndices) : null;
for (int index = VcfCommon.GenotypeIndex; index < vcfColumns.Length; index++)
{
samples[index - VcfCommon.GenotypeIndex] = ExtractSample(vcfColumns[index], formatIndices, numAltAlleles, legacySampleExtractor);
}
return(samples);
}
/// <summary>
/// extracts the genotype fields from the VCF file and returns a list of JSON samples
/// </summary>
internal ISample[] ExtractSamples()
{
// sanity check: make sure we have enough columns
if (_vcfColumns.Length < VcfCommon.MinNumColumnsSampleGenotypes)
{
return(null);
}
var nSamples = _vcfColumns.Length - VcfCommon.MinNumColumnsSampleGenotypes + 1;
var samples = new ISample[nSamples];
// extract the indices for each genotype field
_formatIndices = FormatIndices.Extract(_vcfColumns[VcfCommon.FormatIndex]);
// add each sample
for (var index = VcfCommon.GenotypeIndex; index < _vcfColumns.Length; index++)
{
samples[index - VcfCommon.GenotypeIndex] = ExtractSample(_vcfColumns[index]);
}
return(samples);
}
