public static Dictionary <string, List <CandidateAllele> > GetVariantsByChromosome(this VcfReader reader,
bool variantsOnly = false, bool flagIsKnown = false, List <AlleleCategory> typeFilter = null, Func <CandidateAllele, bool> doSkipCandidate = null)
{
var lookup = new Dictionary <string, List <CandidateAllele> >();
var calledVariants = VcfVariantUtilities.Convert(reader.GetVariants());
foreach (var calledVariant in calledVariants)
{
var candidate = BackToCandiate(calledVariant);
if (candidate.Type != AlleleCategory.Unsupported)
{
if (variantsOnly && candidate.Type == AlleleCategory.Reference)
{
continue;
}
if (typeFilter != null && !typeFilter.Contains(candidate.Type))
{
continue;
}
if (doSkipCandidate != null && doSkipCandidate(candidate))
{
continue;
}
if (flagIsKnown)
{
candidate.IsKnown = true;
}
if (!lookup.ContainsKey(candidate.Chromosome))
{
lookup[candidate.Chromosome] = new List <CandidateAllele>();
}
lookup[candidate.Chromosome].Add(candidate);
}
}
return(lookup);
}
/// <summary>
/// populates a called allele object given an array of vcf columns
/// </summary>
protected static void ConvertColumnsToVariant(bool shouldTrimComplexAlleles, string[] cols, CalledAllele allele, int alleleIndex)
{
bool shouldOutputRcCounts = true;
bool shouldOutputTsCounts = true;
if ((cols == null) || (cols.Length == 0))
{
allele = null;
return;
}
//set defaults.
var genotypeQscore = 0;
var referenceSupport = 0;
var altSupport = 0;
var genotypeString = "";
var totalCoverage = 0;
var variantQuality = 0.0;
var numAlts = 1;
var noiseLevel = 0;
var fractionNocalls = 0f;
var strandBiasInGATKScaleCoords = -100f;
var tsCounts = new List <string>();
//
//read in simple data
allele.Chromosome = cols[VcfCommon.ChromIndex];
allele.ReferencePosition = int.Parse(cols[VcfCommon.PosIndex]);
allele.ReferenceAllele = cols[VcfCommon.RefIndex];
allele.Filters = VcfVariantUtilities.MapFilterString(cols[VcfCommon.FilterIndex]);
bool gotQual = double.TryParse(cols[VcfCommon.QualIndex], out variantQuality); // CFTR uses a ".", which is not actually legal... (actually, vcf 4.1 does allow the missing value "." here. Strelka uses it)
if (gotQual)
{
allele.VariantQscore = (int)variantQuality;
}
// parse the variant alleles
var variantAlleles = cols[VcfCommon.AltIndex].Split(',');
allele.AlternateAllele = variantAlleles[alleleIndex];
var isRef = (allele.AlternateAllele == ".");
if (isRef)
{
numAlts = 0;
}
else
{
numAlts = variantAlleles.Count();
}
// parse the info field data (presume, single sample)
Dictionary <string, string> InfoFields = ParseInfoFields(cols);
// parse the genotype data (presume, single sample)
List <Dictionary <string, string> > Genotypes = ParseGenotypeData(cols);
//get more complex allele data...
if (InfoFields.ContainsKey("DP"))
{
totalCoverage = Int32.Parse(InfoFields["DP"]);
}
if ((Genotypes.Count > 0) && (Genotypes[0] != null))
{
if (Genotypes[0].ContainsKey("GQ"))
{
genotypeQscore = Int32.Parse(Genotypes[0]["GQ"]);
}
else if (Genotypes[0].ContainsKey("GQX"))
{
genotypeQscore = Int32.Parse(Genotypes[0]["GQX"]);
}
if (Genotypes[0].ContainsKey("GT"))
{
genotypeString = Genotypes[0]["GT"];
}
if (Genotypes[0].ContainsKey("NL"))
{
noiseLevel = Int32.Parse(Genotypes[0]["NL"]);
}
if (Genotypes[0].ContainsKey("NC"))
{
fractionNocalls = float.Parse(Genotypes[0]["NC"]);
}
if (Genotypes[0].ContainsKey("SB"))
{
strandBiasInGATKScaleCoords = float.Parse(Genotypes[0]["SB"]);
}
var ADstrings = new string[] { "0", "0" };
if (Genotypes[0].ContainsKey("AD"))
{
ADstrings = Genotypes[0]["AD"].Split(',');
}
referenceSupport = int.Parse(ADstrings[0]);
//by default alt support is 0.
if ((!isRef) && (ADstrings.Length > 1))
{
altSupport = int.Parse(ADstrings[1]);
}
if (shouldOutputRcCounts)
{
if (Genotypes[0].ContainsKey("US"))
{
tsCounts = Genotypes[0]["US"].Split(',').ToList();
}
}
allele.Genotype = VcfVariantUtilities.MapGTString(genotypeString, numAlts);
//note this awkward vcf line (pisces)
//"chr4\t10\t.\tAA\tGA,G\t0\tPASS\tDP=5394\tGT:GQ:AD:DP:VF:NL:SB:NC\t1/2:0:2387,2000:5394:0.8133:23:0.0000:0.0000";
//and this one
//chr2 19946216.ATGTGTG ATG,ATGTG,A 0 PASS metal = platinum; cgi =.; bwa_freebayes = HD:0,LOOHD: 0; bwa_platypus =.; bwa_gatk3 = HD:2,LOOHD: 2; cortex =.; isaac2 = HD:1,LOOHD: 1; dist2closest = 192 GT 1 | 2
if ((numAlts >= 2) && (Genotypes[0].ContainsKey("AD")))
{
if (ADstrings.Count() <= numAlts) //in this case we never expressedly gave the ref support, so we have to derive it.
{
int totalAltCount = 0;
for (int altIndex = 0; altIndex < numAlts; altIndex++)
{
var altSupportAtIndex = int.Parse(ADstrings[altIndex]);
totalAltCount += altSupportAtIndex;
if (altIndex == alleleIndex)
{
altSupport = altSupportAtIndex;
}
}
referenceSupport = Math.Max(0, totalCoverage - totalAltCount);
}
}
}
var strandBiasResults = new BiasResults();
strandBiasResults.GATKBiasScore = strandBiasInGATKScaleCoords;
//set the remaining data
allele.TotalCoverage = totalCoverage;
allele.AlleleSupport = isRef ? referenceSupport : altSupport;
allele.ReferenceSupport = referenceSupport;
allele.GenotypeQscore = genotypeQscore;
allele.NoiseLevelApplied = noiseLevel;
allele.StrandBiasResults = strandBiasResults;
allele.IsForcedToReport = allele.Filters.Contains(FilterType.ForcedReport);
//set the derived values
allele.SetType();
allele.ForceFractionNoCalls(fractionNocalls);
//rescue attempt for complex types, ie ACGT -> ACGTGG.
//Get the simplest form of the allele
if ((allele.Type == AlleleCategory.Unsupported) && shouldTrimComplexAlleles)
{
VcfVariantUtilities.TrimUnsupportedAlleleType(allele);
}
if (tsCounts.Count != 0)
{
VcfVariantUtilities.FillInCollapsedReadsCount(shouldOutputRcCounts, shouldOutputTsCounts, allele, tsCounts);
}
}