private static List <BamAlignment> ExtractReadsFromRealignerAndCombiner(PairResult pair, string refSeq,
int refSeqOffset, List <PreIndel> preIndels, bool hasExistingIndels = false)
{
var stitchedPairHandler =
new PairHandler(new Dictionary <int, string>()
{
{ 1, "chr1" }
}, new BasicStitcher(0), tryStitch: true);
var snippetSource = new Mock <IGenomeSnippetSource>();
var genomeSnippet = new GenomeSnippet()
{
Chromosome = "chr1",
Sequence = new string('A', refSeqOffset) + refSeq + new string('T', 1000),
StartPosition = 0
};
snippetSource.Setup(x => x.GetGenomeSnippet(It.IsAny <int>())).Returns(genomeSnippet);
var mockStatusHandler = new Mock <IStatusHandler>();
var comparer = new GemBasicAlignmentComparer(false, false);
var readRealigner = new GeminiReadRealigner(comparer, remaskSoftclips: false,
keepProbeSoftclips: false, keepBothSideSoftclips: false,
trackActualMismatches: false, checkSoftclipsForMismatches: true,
debug: false, maskNsOnly: false, maskPartialInsertion: false,
minimumUnanchoredInsertionLength: 1,
minInsertionSizeToAllowMismatchingBases: 4,
maxProportionInsertSequenceMismatch: 0.2); // TODO fix // TODO figure out what I was saying to fix here...
var filterer = new Mock <IRegionFilterer>();
filterer.Setup(x => x.AnyIndelsNearby(It.IsAny <int>())).Returns(true);
var indels = preIndels.Select(x => HashableIndelSource.GetHashableIndel(genomeSnippet, x, 0, false)).ToList();
var indelSource = new ChromosomeIndelSource(indels, snippetSource.Object);
var realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
var combiner = new ReadPairRealignerAndCombiner(new NonSnowballEvidenceCollector(),
new PostRealignmentStitcher(stitchedPairHandler, new DebugStatusHandler(new ReadStatusCounter())),
realignmentEvaluator, new PairSpecificIndelFinder(), "chr1", false, hasExistingIndels: hasExistingIndels);
var nmCalc = new NmCalculator(snippetSource.Object);
var result = combiner.ExtractReads(pair, nmCalc);
return(result);
}
public void GetFinalAlignment_NonMock()
{
var snippetSource = new Mock <IGenomeSnippetSource>();
var genomeSnippet = new GenomeSnippet()
{
Chromosome = "chr1",
Sequence = new string('A', 1000) + "ATCGATTGA" + new string('T', 1000),
StartPosition = 1000
};
snippetSource.Setup(x => x.GetGenomeSnippet(It.IsAny <int>())).Returns(genomeSnippet);
var mockStatusHandler = new Mock <IStatusHandler>();
var comparer = new GemBasicAlignmentComparer(false, false);
var readRealigner = new GeminiReadRealigner(comparer, remaskSoftclips: false,
keepProbeSoftclips: false, keepBothSideSoftclips: false,
trackActualMismatches: false, checkSoftclipsForMismatches: true,
debug: false, maskNsOnly: false, maskPartialInsertion: false,
minimumUnanchoredInsertionLength: 1,
minInsertionSizeToAllowMismatchingBases: 4, maxProportionInsertSequenceMismatch: 0.2); // TODO fix // TODO figure out what I was saying to fix here...
var filterer = GetMockRegionFilterer();
var indels = new List <HashableIndel>();
var indelSource = new ChromosomeIndelSource(indels, snippetSource.Object);
var realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
var origBamAlignment =
TestHelpers.CreateBamAlignment("AAAAAAATTCA", 1500, 1500, 30, true, cigar: new CigarAlignment("11M"));
var realigned = realignmentEvaluator.GetFinalAlignment(origBamAlignment, out bool changed, out bool forcedSoftclip,
out bool confirmed, out bool sketchy);
// No indels
Assert.False(changed);
Assert.False(confirmed);
indels = new List <HashableIndel>()
{
new HashableIndel()
{
Chromosome = "chr1",
ReferencePosition = 1506,
ReferenceAllele = "A",
AlternateAllele = "ATT",
Type = AlleleCategory.Insertion,
Length = 2
}
};
indelSource = new ChromosomeIndelSource(indels, snippetSource.Object);
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
realigned = realignmentEvaluator.GetFinalAlignment(origBamAlignment, out changed, out forcedSoftclip,
out confirmed, out sketchy);
Assert.True(changed);
Assert.False(confirmed);
Assert.Equal("7M2I2M", realigned.CigarData.ToString());
var confirmedAccepteds = new List <HashableIndel>();
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
var reRealigned = realignmentEvaluator.GetFinalAlignment(realigned, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds);
Assert.False(changed);
Assert.True(confirmed);
Assert.Equal("7M2I2M", reRealigned.CigarData.ToString());
// Existing indel is best (and only)
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
reRealigned = realignmentEvaluator.GetFinalAlignment(realigned, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds, existingIndels: new List <PreIndel>()
{
new PreIndel(new CandidateAllele("chr1", 1506, "A", "ATT", AlleleCategory.Insertion))
});
Assert.False(changed);
Assert.True(confirmed);
Assert.Equal("7M2I2M", reRealigned.CigarData.ToString());
// Existing indel is unsanctioned but good fit - keep it
var alignmentWithInsertion =
TestHelpers.CreateBamAlignment("AAAAAAATTCA", 1500, 1500, 30, true, cigar: new CigarAlignment("7M3I1M"));
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, false, filterer.Object, false);
var realignedExistingIns = realignmentEvaluator.GetFinalAlignment(alignmentWithInsertion, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds, existingIndels: new List <PreIndel>()
{
new PreIndel(new CandidateAllele("chr1", 1506, "A", "ATTC", AlleleCategory.Insertion))
});
Assert.False(changed);
Assert.False(confirmed);
Assert.Equal("7M3I1M", realignedExistingIns.CigarData.ToString());
// Existing indel is unsanctioned and we're softclipping unknowns - softclip it
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
realignedExistingIns = realignmentEvaluator.GetFinalAlignment(alignmentWithInsertion, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds, existingIndels: new List <PreIndel>()
{
new PreIndel(new CandidateAllele("chr1", 1506, "A", "ATTC", AlleleCategory.Insertion))
});
Assert.False(changed);
Assert.False(confirmed);
Assert.Equal("7M4S", realignedExistingIns.CigarData.ToString());
indels = new List <HashableIndel>()
{
new HashableIndel()
{
Chromosome = "chr1",
ReferencePosition = 1506,
ReferenceAllele = "A",
AlternateAllele = "ATT",
Type = AlleleCategory.Insertion,
Length = 2,
Score = 1000
},
new HashableIndel()
{
Chromosome = "chr1",
ReferencePosition = 1506,
ReferenceAllele = "A",
AlternateAllele = "ATTC",
Type = AlleleCategory.Insertion,
Length = 3,
Score = 760
},
new HashableIndel()
{
Chromosome = "chr1",
ReferencePosition = 1506,
ReferenceAllele = "A",
AlternateAllele = "ATTG",
Type = AlleleCategory.Insertion,
Length = 3,
Score = 10
}
};
indelSource = new ChromosomeIndelSource(indels, snippetSource.Object);
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
realigned = realignmentEvaluator.GetFinalAlignment(origBamAlignment, out changed, out forcedSoftclip,
out confirmed, out sketchy);
Assert.True(changed);
Assert.False(confirmed);
Assert.Equal("7M3I1M", realigned.CigarData.ToString());
confirmedAccepteds = new List <HashableIndel>();
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
reRealigned = realignmentEvaluator.GetFinalAlignment(realigned, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds);
Assert.False(changed);
Assert.True(confirmed);
Assert.Equal("7M3I1M", reRealigned.CigarData.ToString());
// Existing indel is not the top one but is the best fit, keep it
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
reRealigned = realignmentEvaluator.GetFinalAlignment(realigned, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds, existingIndels: new List <PreIndel>()
{
new PreIndel(new CandidateAllele("chr1", 1506, "A", "ATTC", AlleleCategory.Insertion))
});
Assert.False(changed);
Assert.True(confirmed);
Assert.Equal("7M3I1M", reRealigned.CigarData.ToString());
// Has existing unsanctioned indel and there are better ones to realign around - ignore the bad one, take the good
realignmentEvaluator = new RealignmentEvaluator(indelSource, mockStatusHandler.Object, readRealigner,
new RealignmentJudger(comparer), "chr1", false, true, true, true, filterer.Object, false);
reRealigned = realignmentEvaluator.GetFinalAlignment(realigned, out changed, out forcedSoftclip,
out confirmed, out sketchy, confirmedAccepteds: confirmedAccepteds, existingIndels: new List <PreIndel>()
{
new PreIndel(new CandidateAllele("chr1", 1507, "A", "ATC", AlleleCategory.Insertion))
});
Assert.False(changed);
Assert.True(confirmed);
Assert.Equal("7M3I1M", reRealigned.CigarData.ToString());
}
public void GetRelevantIndels()
{
var indel = new HashableIndel()
{
AlternateAllele = "AG",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 1,
ReferencePosition = 10002,
Score = 1,
Type = AlleleCategory.Insertion
};
var indel2 = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 10002,
Score = 10,
Type = AlleleCategory.Insertion
};
var positionWayLower = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 8002,
Score = 10,
Type = AlleleCategory.Insertion
};
var positionLikelyDiffBlockButWithinRange = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 9800,
Score = 10,
Type = AlleleCategory.Insertion
};
var positionWayHigher = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 21000,
Score = 10,
Type = AlleleCategory.Insertion
};
var borderCaseHigh = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 10251,
Score = 10,
Type = AlleleCategory.Insertion
};
var borderCaseLow = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 9752,
Score = 10,
Type = AlleleCategory.Insertion
};
var indelAt0 = new HashableIndel()
{
AlternateAllele = "AGT",
ReferenceAllele = "A",
Chromosome = "chr1",
Length = 2,
ReferencePosition = 0,
Score = 10,
Type = AlleleCategory.Insertion
};
var indels = new List <HashableIndel>()
{
indel,
indel2,
positionWayLower,
positionLikelyDiffBlockButWithinRange,
positionWayHigher,
borderCaseHigh,
borderCaseLow,
indelAt0
};
var snippetSource = new Mock <IGenomeSnippetSource>();
snippetSource.Setup(s => s.GetGenomeSnippet(It.IsAny <int>())).Returns(new GenomeSnippet()
{
Chromosome = "chr1", Sequence = new string('A', 2000), StartPosition = 1
});
var indelSource = new ChromosomeIndelSource(indels, snippetSource.Object);
//var relevant = indelSource.GetRelevantIndels(100);
//Assert.Equal(4, relevant.Count);
// Should get indel1 and 2, border high, border low, withinrange
var relevant = indelSource.GetRelevantIndels(10002);
Assert.Equal(5, relevant.Count());
// Should get indel1 and 2, border low, within range, but not border high (now > 250 away)
relevant = indelSource.GetRelevantIndels(10000);
Assert.Equal(4, relevant.Count());
// Should get all 5 as 10002 did, showing that it is 250 inclusive
relevant = indelSource.GetRelevantIndels(10001);
Assert.Equal(5, relevant.Count());
// Should get the 9752 and the 9800
relevant = indelSource.GetRelevantIndels(9700);
Assert.Equal(2, relevant.Count());
// Not close enough to anything
relevant = indelSource.GetRelevantIndels(9500);
Assert.Equal(0.0, relevant.Count());
relevant = indelSource.GetRelevantIndels(0);
Assert.Equal(1.0, relevant.Count());
relevant = indelSource.GetRelevantIndels(100000);
Assert.Equal(0.0, relevant.Count());
}