public IEnumerable <VeadGroup> GetVeadGroups(VcfNeighborhood neighborhood)
{
var veadGroups = new Dictionary <string, VeadGroup>();
var neighbors = neighborhood.VcfVariantSites;
var refName = neighbors.First().ReferenceName;
_alignmentExtractor.Jump(refName);
// keep reading the alignments while we're on the same reference sequence
var veadMaker = new VeadFinder(_options);
var debugLog = Path.Combine(_debugLogRoot, refName + "_" + neighborhood.Id + "_ReadsInNbhd.txt");
WriteToReadLog(debugLog, string.Join("\t", "ReadName", "used?", "IsFirstMate", "CigarData", "Read.Position"));
Read read = new Read();
while (true)
{
if (!_alignmentExtractor.GetNextAlignment(read))
{
break; // no more reads
}
if (ShouldSkipRead(read, neighborhood))
{
WriteToReadLog(debugLog, (string.Join("\t", read.Name, "skipped", read.IsFirstMate, read.CigarData.ToString(), read.Position)));
continue;
}
if (PastNeighborhood(read, neighborhood))
{
WriteToReadLog(debugLog, (string.Join("\t", read.Name, "past nbhd", read.IsFirstMate, read.CigarData.ToString(), read.Position)));
break;
}
//Make a vead and add it to our list
var readName = read.Name + "_";
if (read.IsFirstMate)
{
readName += "fwd_" + read.Position;
}
else
{
readName += "rev_" + read.Position;
}
WriteToReadLog(debugLog, (string.Join("\t", read.Name, "will use", read.IsFirstMate, read.CigarData.ToString(), read.Position)));
//map from bases to ref position
var vead = new Vead(readName, veadMaker.FindVariantResults(neighbors, read));
if (vead.SiteResults == null || !vead.SiteResults.Any())
{
continue;
}
// Add vead to a veadgroup.
var hash = vead.ToVariantSequence();
if (!veadGroups.ContainsKey(hash))
{
veadGroups.Add(hash, new VeadGroup(vead));
}
else
{
veadGroups[hash].AddSupport(vead);
}
}
return(veadGroups.Values);
}
//this unit test was made after we found bug ScyllaLoosingRefCalls_PICS-723.
//We had a 1/. GT reported when it should be 1/0.
//The reason for this is that all the refs (the "0"s) got incorrectly sucked up.
//Ie, MNV ACG-> AG claimed 50 refs, so we (incorrectly) subtracted 50 refs from it.
//The bug is that the ref counts got subtractedfrom the exact same mnv that claimed them.
// This should never happen, and was not the intent of the alg.
//
//The affected mehtod is: CreateMnvsFromClusters in VcfNbhd
public void CreateMnvsFromClusters_TakeUpRefCount()
{
var originalVcfVariant1 = TestHelper.CreateDummyAllele("chr1", 123, "ACG", "AT", 1000, 156);
var originalVcfVariant2 = TestHelper.CreateDummyAllele("chr1", 123, "A", "TTTTTT", 1000, 200);
// var originalVcfVariant3 = TestHelper.CreateDummyAllele("chr1", 123, "AC", "TT", 1000, 100);
var vs1 = new VariantSite(originalVcfVariant1);
var vs2 = new VariantSite(originalVcfVariant2);
var caller = new VariantCaller(new VariantCallingParameters(), new BamFilterParameters());
var nbhd = new VcfNeighborhood(0, "chr1", vs1, vs2);
var callableNeighborhood = new CallableNeighborhood(nbhd, new VariantCallingParameters());
callableNeighborhood.AddAcceptedPhasedVariant(
new CalledAllele(AlleleCategory.Snv)
{
Chromosome = "chr1",
ReferencePosition = 123,
ReferenceAllele = "A",
AlternateAllele = "A",
VariantQscore = 100,
TotalCoverage = 1000,
AlleleSupport = 200,
ReferenceSupport = 350
});
callableNeighborhood.AddAcceptedPhasedVariant(
new CalledAllele(AlleleCategory.Mnv)
{
Chromosome = "chr1",
ReferencePosition = 123,
ReferenceAllele = "ACG",
AlternateAllele = "AT",
VariantQscore = 100,
TotalCoverage = 1000,
AlleleSupport = 300,
ReferenceSupport = 350
});
callableNeighborhood.AddAcceptedPhasedVariant(
new CalledAllele(AlleleCategory.Insertion)
{
Chromosome = "chr1",
ReferencePosition = 123,
ReferenceAllele = "A",
AlternateAllele = "AAAAA",
VariantQscore = 100,
TotalCoverage = 1000,
AlleleSupport = 250,
ReferenceSupport = 350
});
//default behavior, nothing gets sucked up
callableNeighborhood.UsedRefCountsLookup = new Dictionary <int, SuckedUpRefRecord>()
{
};
vs1.VcfReferencePosition = 123;
var vead = new Vead("dummy", new VariantSite[] { vs1 });
var vg = new VeadGroup(vead);
var fakeCluster = new Cluster("test", new List <VeadGroup>()
{
vg
});
fakeCluster.ResetConsensus();
callableNeighborhood.CreateMnvsFromClusters(new List <Cluster> {
fakeCluster
},
20);
caller.CallMNVs(callableNeighborhood);
caller.CallRefs(callableNeighborhood);
var acceptedMNVs = callableNeighborhood.CalledVariants;
var acceptedRefs = callableNeighborhood.CalledRefs;
Assert.Equal(2, acceptedMNVs.Count);
Assert.Equal(3, acceptedMNVs[123].Count);
Assert.Equal(1, acceptedRefs.Count);
//check the ref counts on all the MNVs. Nothing should be sucked up.
Assert.Equal(350, acceptedMNVs[123][0].ReferenceSupport); // Previously: total depth - allele suport. overly simple for now)
Assert.Equal(350, acceptedMNVs[123][1].ReferenceSupport); // Now: explicitly set ref support
Assert.Equal(350, acceptedMNVs[123][2].ReferenceSupport); //
// now variant 0 will suck up 100 ref calls:
var suckedUpRefRecord100 = new SuckedUpRefRecord()
{
Counts = 100, AlleleThatClaimedIt = callableNeighborhood.CandidateVariants[0]
};
callableNeighborhood.UsedRefCountsLookup = new Dictionary <int, SuckedUpRefRecord>()
{
{ 123, suckedUpRefRecord100 }
};
callableNeighborhood.CreateMnvsFromClusters(new List <Cluster> {
fakeCluster
},
20);
caller.CallMNVs(callableNeighborhood);
caller.CallRefs(callableNeighborhood);
acceptedMNVs = callableNeighborhood.CalledVariants;
acceptedRefs = callableNeighborhood.CalledRefs;
//check the ref counts on all the MNVs. refs should only be taken up by the first one
Assert.Equal(350, acceptedMNVs[123][0].ReferenceSupport); //Previously: total depth - allele suport. overly simple for now)
//old result - has bug
//Assert.Equal(1000 - 300, acceptedMNVs[123][1].ReferenceSupport); // Previously: total depth - allele suport - sucked up ref)
//Assert.Equal(1000 - 250, acceptedMNVs[123][2].ReferenceSupport); // Now: explicitly set ref support
//new result, fixed
Assert.Equal(350 - 100, acceptedMNVs[123][1].ReferenceSupport); // refSupport - sucked up ref)
Assert.Equal(350 - 100, acceptedMNVs[123][2].ReferenceSupport); // refSupport - sucked up ref)
}
