public override void Write(IEnumerable <CalledAllele> calledAlleles, IRegionMapper mapper = null)
{
var comparer = new AlleleCompareByLoci();
var sortedVariants = calledAlleles.OrderBy(a => a, comparer).ThenBy(a => a.ReferenceAllele).ThenBy(a => a.AlternateAllele);
base.Write(sortedVariants);
}
/// <summary>
/// Step forward with the reader, assembling a list of variants at your CurrentVariant position.
/// </summary>
/// <param name="Reader"></param>
/// <param name="CurrentVariant"></param>
/// <param name="BackLogExists"></param>
/// <param name="TheBackLog"></param>
/// <returns></returns>
private static List <CalledAllele> AssembleColocatedList(
VcfReader Reader, CalledAllele CurrentVariant, bool mFirst,
ref bool BackLogExists, ref List <CalledAllele> TheBackLog)
{
List <CalledAllele> CoLocatedVariants = new List <CalledAllele>();
bool ContinueReadA = true;
while (ContinueReadA)
{
var NextVariantList = new List <CalledAllele>();
if (BackLogExists)
{
NextVariantList = TheBackLog;
BackLogExists = false;
}
else
{
VcfVariant NextVariant = new VcfVariant();
ContinueReadA = Reader.GetNextVariant(NextVariant);
if (!ContinueReadA)
{
break;
}
NextVariantList = VcfVariantUtilities.Convert(new List <VcfVariant> {
NextVariant
}).ToList();
}
// VarOrde = -1 if Current comes first, 0 if co-located.
int VarOrder = (AlleleCompareByLoci.OrderAlleles(CurrentVariant, NextVariantList.First(), mFirst));
switch (VarOrder)
{
case 0: //the variant we just got is at out current position
CoLocatedVariants.AddRange(NextVariantList);
break;
case -1: //the variant we just got is after our current position, and needs to go to the backlog.
TheBackLog = NextVariantList; //NextVariant;
ContinueReadA = false;
BackLogExists = true;
break;
default: //
{
throw new InvalidDataException("Vcf needs to be ordered.");
}
}
}
if (!BackLogExists)
{
TheBackLog = null;
}
return(CoLocatedVariants);
}
public static List <CalledAllele> GetMergedListOfVariants(IVcfNeighborhood completedNbhd, List <CalledAllele> originalVariantsInsideRange)
{
var mergedVariantList = new List <CalledAllele>();
Dictionary <int, List <CalledAllele> > foundMNVS = completedNbhd.CalledVariants;
//track which variants got used for MNV phasing.
var indexesOfVariantsRecalledByMnvCaller = completedNbhd.GetOriginalVcfVariants();
//decide which of the original variants we keep, and which get replaced.
for (int i = 0; i < originalVariantsInsideRange.Count; i++)
{
var originalCall = originalVariantsInsideRange[i];
var currentPosition = originalCall.ReferencePosition;
bool variantWasAlreadyUsed = CheckIfUsed(indexesOfVariantsRecalledByMnvCaller, originalCall);
if (foundMNVS.ContainsKey(currentPosition))
{
//add all the MNVs
mergedVariantList.AddRange(foundMNVS[currentPosition]);
foundMNVS[currentPosition] = new List <CalledAllele>();//empty out the list, but leave the fact that there were MNVs here.
//add back any original variants,
//so long as they were not used by the caller, and not reference
if (!(variantWasAlreadyUsed) &&
(originalCall.Type != Pisces.Domain.Types.AlleleCategory.Reference))
{
mergedVariantList.Add(originalCall);
}
continue;
}
//Else, we did not find any MNVs here.
//Then this position should be either the original call from the vcf,
//or a new reference call converted from a variant that got used by the MNV caller.
if (variantWasAlreadyUsed)
{
var newRef = completedNbhd.CalledRefs[originalCall.ReferencePosition];
//sometimes several variants were used, all at the same position. we dont need to add new references for all of them.
if ((mergedVariantList.Count == 0) || (mergedVariantList.Last().ReferencePosition != currentPosition))
{
mergedVariantList.Add(newRef);
}
}
else //wasnt used for MNV calling
{
mergedVariantList.Add(originalVariantsInsideRange[i]);
}
}
//incase we called any MNVs past the edge of the input VCF.
foreach (var mnvsLeft in foundMNVS)
{
mergedVariantList.AddRange(mnvsLeft.Value);
}
var comparer = new AlleleCompareByLoci();
mergedVariantList.Sort(comparer);
return(mergedVariantList);
}
public void OrderVariants()
{
var chr10 = TestHelper.CreateDummyAllele("chr10", 123, "A", "C", 1000, 156);
var chrX = TestHelper.CreateDummyAllele("chrX", 123, "A", "C", 1000, 156);
var chrXSecond = TestHelper.CreateDummyAllele("chrX", 124, "A", "C", 1000, 156);
var chrM = TestHelper.CreateDummyAllele("chrM", 123, "A", "C", 1000, 156);
var chrMSecond = TestHelper.CreateDummyAllele("chrM", 124, "A", "C", 1000, 156);
var chr9 = TestHelper.CreateDummyAllele("chr9", 123, "A", "C", 1000, 156);
var chr9Second = TestHelper.CreateDummyAllele("chr9", 124, "A", "C", 1000, 156);
var nonstandardChrZ = TestHelper.CreateDummyAllele("chrZ", 123, "A", "C", 1000, 156);
var nonstandardChrA = TestHelper.CreateDummyAllele("chrA", 123, "A", "C", 1000, 156);
// ---------------------------------------------------------------------------
// When neither or both is on chrM, shouldn't matter if we set option to prioritize chrM
// ---------------------------------------------------------------------------
// Same chrom, different positions - numeric chrom
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chr9Second, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chr9Second, false));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chr9Second, chr9, true));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chr9Second, chr9, false));
// Same chrom, different positions - chrX
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, chrXSecond, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, chrXSecond, false));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrXSecond, chrX, true));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrXSecond, chrX, false));
// Same chrom, different positions - chrM
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrM, chrMSecond, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrM, chrMSecond, false));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrMSecond, chrM, true));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrMSecond, chrM, false));
// Different chroms, one is >=10 (direct string compare would not sort these chroms correctly)
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chr10, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chr10, false));
// One numeric, one chrX
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chrX, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chrX, false));
// Same chrom, same position
Assert.Equal(0, AlleleCompareByLoci.OrderAlleles(chr9, chr9, true));
Assert.Equal(0, AlleleCompareByLoci.OrderAlleles(chrX, chrX, true));
Assert.Equal(0, AlleleCompareByLoci.OrderAlleles(chrM, chrM, true));
// ---------------------------------------------------------------------------
// If one is on chrM, option to prioritize chrM matters
// ---------------------------------------------------------------------------
// One numeric, one chrM
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chr9, chrM, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, chrM, false));
// One chrX, one chrM
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrX, chrM, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, chrM, false));
// ---------------------------------------------------------------------------
// Nonstandard chroms should be below numerics and then ordered alphabetically
// ---------------------------------------------------------------------------
// One numeric, one weird
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, nonstandardChrZ, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, nonstandardChrZ, false));
// One chrX, one weird
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrZ, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrZ, false));
// One chrM, one weird
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrZ, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrZ, false));
// One numeric, one funny
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, nonstandardChrA, true));
Assert.Equal(-1, AlleleCompareByLoci.OrderAlleles(chr9, nonstandardChrA, false));
// One chrX, one funny
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrA, true));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrA, false));
// One chrM, one funny
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrA, true));
Assert.Equal(1, AlleleCompareByLoci.OrderAlleles(chrX, nonstandardChrA, false));
}
/// <summary>
/// perfom a Venn split between two samples
/// </summary>
/// <param name="sampleName"></param>
/// <param name="consensusFilePath"></param>
/// <param name="inputPaths"></param>
/// <param name="outputTwoSampleResults"></param>
public void DoPairwiseVenn(bool mFirst)
{
bool doConsensus = (consensusBuilder != null);
bool requireGenotypes = false;
using (VcfReader ReaderA = new VcfReader(_inputPaths[0], requireGenotypes))
using (VcfReader ReaderB = new VcfReader(_inputPaths[1], requireGenotypes))
{
if (doConsensus)
{
consensusBuilder.OpenConsensusFile(ReaderA.HeaderLines);
}
OpenVennDiagramStreams(ReaderA.HeaderLines);
//read the first variant from each gvcf file...
var currentAllele = new CalledAllele();
var backLogPoolAVcfVariant = new VcfVariant();
var backLogPoolBVcfVariant = new VcfVariant();
var backLogExistPoolA = ReaderA.GetNextVariant(backLogPoolAVcfVariant);
var backLogExistPoolB = ReaderB.GetNextVariant(backLogPoolBVcfVariant);
var backLogPoolAAlleles = backLogExistPoolA ? VcfVariantUtilities.Convert(new List <VcfVariant> {
backLogPoolAVcfVariant
}).ToList() : null;
var backLogPoolBAlleles = backLogExistPoolB ? VcfVariantUtilities.Convert(new List <VcfVariant> {
backLogPoolBVcfVariant
}).ToList() : null;
//keep reading and processing until we are done with both gvcfs
while (true)
{
try
{
//1) Get the next set of variants. Pull from the backlog first,
//choosing all the variants at the first available position.
var coLocatedPoolAAlleles = new List <CalledAllele>();
var coLocatedPoolBAlleles = new List <CalledAllele>();
//We need to set up which location to look at next.
//Choose the first one from the backlog.
if (backLogExistPoolA || backLogExistPoolB)
{
if (backLogExistPoolA && backLogExistPoolB)
{
int OrderResult = AlleleCompareByLoci.OrderAlleles(
backLogPoolAAlleles.First(), backLogPoolBAlleles.First(), mFirst);
if (OrderResult < 0)
{
currentAllele.Chromosome = backLogPoolAAlleles.First().Chromosome;
currentAllele.ReferencePosition = backLogPoolAAlleles.First().ReferencePosition;
}
else
{
currentAllele.Chromosome = backLogPoolBAlleles.First().Chromosome;
currentAllele.ReferencePosition = backLogPoolBAlleles.First().ReferencePosition;
}
}
else if (backLogExistPoolB)
{
currentAllele.Chromosome = backLogPoolBAlleles.First().Chromosome;
currentAllele.ReferencePosition = backLogPoolBAlleles.First().ReferencePosition;
}
else //if (backLogExistPoolA)
{
currentAllele.Chromosome = backLogPoolAAlleles.First().Chromosome;
currentAllele.ReferencePosition = backLogPoolAAlleles.First().ReferencePosition;
}
//assemble lists of co-located variants at the position of the current variant
coLocatedPoolAAlleles = AssembleColocatedList(ReaderA, currentAllele, mFirst,
ref backLogExistPoolA, ref backLogPoolAAlleles);
coLocatedPoolBAlleles = AssembleColocatedList(ReaderB, currentAllele, mFirst,
ref backLogExistPoolB, ref backLogPoolBAlleles);
} //else, if there is nothing in either backlog, the colocated-variant list should stay empty.
//2) Now we have finshed reading out all the co-located variants...
//We need organize them into pairs, to know which allele to compare with which.
var Pairs = SelectPairs(coLocatedPoolAAlleles, coLocatedPoolBAlleles);
var ConsensusVariants = new List <CalledAllele>();
AggregateAllele lastConsensusReferenceCall = null;
//3) For each pair, combine them and mark if biased or not.
for (int PairIndex = 0; PairIndex < Pairs.Count; PairIndex++)
{
var VariantA = Pairs[PairIndex][0];
var VariantB = Pairs[PairIndex][1];
var ComparisonCase = GetComparisonCase(VariantA, VariantB);
//add VarA and VarB to appropriate venn diagram files.
WriteVarsToVennFiles(ComparisonCase, VariantA, VariantB);
AggregateAllele Consensus = null;
if (doConsensus)
{
Consensus = consensusBuilder.CombineVariants(
VariantA, VariantB, ComparisonCase);
//Its possible for multiallelic sites, a pair of variants could
//end up as a concensus reference. And we already may have
//called a reference for this loci already.
//we might have some cleaning up to do...
if (Consensus.Genotype == Pisces.Domain.Types.Genotype.HomozygousRef)
{
//this is the first time we see a reference at this loci
if (lastConsensusReferenceCall == null)
{
lastConsensusReferenceCall = Consensus;
//its OK to fall through and add our Consensus variant to the list.
}
//Else, if we have already called a reference variant
// for this loci already
// we want to merge the results from this reference with the old one.
// *before* we write it to file.
else
{
//the chr, pos, ref, alt,and depth should be correct.
//We'll merge the filters,
//and take the max SB and PB. (where a higher value indicates worse value, so we stay conservative)
lastConsensusReferenceCall.Filters = ConsensusBuilder.CombineFilters(lastConsensusReferenceCall, Consensus);
lastConsensusReferenceCall.StrandBiasResults = new Pisces.Domain.Models.BiasResults()
{
GATKBiasScore = Math.Max(lastConsensusReferenceCall.StrandBiasResults.GATKBiasScore, Consensus.StrandBiasResults.GATKBiasScore)
};
lastConsensusReferenceCall.PoolBiasResults = new Pisces.Domain.Models.BiasResults()
{
GATKBiasScore = Math.Max(lastConsensusReferenceCall.PoolBiasResults.GATKBiasScore, Consensus.PoolBiasResults.GATKBiasScore)
};
//we are going to take the min Q and NL score, to be conservative
lastConsensusReferenceCall.NoiseLevelApplied = Math.Min(lastConsensusReferenceCall.NoiseLevelApplied, Consensus.NoiseLevelApplied);
lastConsensusReferenceCall.GenotypeQscore = Math.Min(lastConsensusReferenceCall.GenotypeQscore, Consensus.GenotypeQscore);
lastConsensusReferenceCall.VariantQscore = Math.Min(lastConsensusReferenceCall.VariantQscore, Consensus.GenotypeQscore);
continue;
}
}
ConsensusVariants.Add(Consensus);
}
}
//4) Write out the results to file. (this will be a list of co-located variants)
if (doConsensus)
{
consensusBuilder.WriteConsensusVariantsToFile(ConsensusVariants);
}
//we assembled everyone and no one is left.
if ((backLogPoolAAlleles == null) &&
(backLogPoolBAlleles == null))
{
break;
}
}
catch (Exception ex)
{
OnError(string.Format("Fatal error encountered comparing paired sample vcfs; Check {0}, position {1}. Exception: {2}",
currentAllele.Chromosome, currentAllele.ReferencePosition, ex));
throw;
}
} //close assemble list
}//close usings
if (doConsensus)
{
consensusBuilder.CloseConsensusFile();
}
CloseVennDiagramStreams();
}
