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 ReadPairRealignerAndCombiner GetRealignPairHandler(bool tryRestitch, bool alreadyStitched,
bool pairAwareRealign,
Dictionary <int, string> refIdMapping, ReadStatusCounter statusCounter, bool isSnowball,
IChromosomeIndelSource indelSource, string chromosome, Dictionary <string, IndelEvidence> masterLookup,
bool hasIndels, Dictionary <HashableIndel, int[]> outcomesLookup, bool skipRestitchIfNothingChanged)
{
var stitcher = GetStitcher();
var stitchedPairHandler = new PairHandler(refIdMapping, stitcher, tryStitch: tryRestitch);
var judger = new RealignmentJudger(GetAlignmentComparer());
var readRealigner = new GeminiReadRealigner(GetAlignmentComparer(), remaskSoftclips: _geminiOptions.RemaskMessySoftclips,
keepProbeSoftclips: _geminiOptions.KeepProbeSoftclip, keepBothSideSoftclips: _geminiOptions.KeepBothSideSoftclips || (_geminiOptions.KeepProbeSoftclip && alreadyStitched),
trackActualMismatches: _realignmentAssessmentOptions.TrackActualMismatches, checkSoftclipsForMismatches: _realignmentAssessmentOptions.CheckSoftclipsForMismatches,
debug: _geminiOptions.Debug, maskNsOnly: !(_geminiOptions.RemaskMessySoftclips || _geminiOptions.KeepProbeSoftclip || _geminiOptions.KeepBothSideSoftclips), maskPartialInsertion: _realignmentOptions.MaskPartialInsertion,
minimumUnanchoredInsertionLength: _realignmentOptions.MinimumUnanchoredInsertionLength,
minInsertionSizeToAllowMismatchingBases: 4, maxProportionInsertSequenceMismatch: 0.2); // TODO fix // TODO figure out what I was saying to fix here...
IStatusHandler statusHandler = new DebugSummaryStatusHandler(statusCounter);
if (_geminiOptions.Debug)
{
statusHandler = new DebugStatusHandler(statusCounter);
}
// Only softclip unknowns if it is not stitched to begin with (we believe in these more, plus it makes our lives simpler for dealing with stitched directions)
var softclipUnknownIndels = _geminiOptions.SoftclipUnknownIndels && !alreadyStitched;
//var regionFilterer = new RegionFilterer(chromosome, indelSource.Indels);
var regionFilterer = new DummyRegionFilterer();
var collector = GetCollector(isSnowball);
var realignmentEvaluator = new RealignmentEvaluator(indelSource.DeepCopy(), statusHandler, readRealigner, judger, chromosome,
_realignmentAssessmentOptions.TrackActualMismatches, _realignmentAssessmentOptions.CheckSoftclipsForMismatches, _geminiOptions.AllowRescoringOrigZero, softclipUnknownIndels,
regionFilterer, _geminiOptions.LightDebug);
return(new ReadPairRealignerAndCombiner(
collector,
GetRestitcher(stitchedPairHandler, statusHandler),
realignmentEvaluator,
GetIndelFinder(pairAwareRealign, chromosome, indelSource), chromosome, alreadyStitched, pairAwareRealign,
masterLookup: masterLookup, hasExistingIndels: hasIndels,
masterOutcomesLookup: outcomesLookup, skipRestitchIfNothingChanged: skipRestitchIfNothingChanged, allowedToStitch: !_geminiOptions.SkipStitching));
}
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());
}
