/// <summary>
/// Checks if any of the conditions is true:
/// 1. The read is a duplicate,
/// 2. The read failed QC,
/// 3. The read is of low mapping quality.
/// </summary>
/// <param name="alignment"></param>
/// <returns></returns>
public static bool IsDuplicateFailedQCLowQuality(BamAlignment alignment, uint qualityThreshold)
{
if (alignment.IsDuplicate())
{
return(true);
}
if (alignment.IsFailedQC())
{
return(true);
}
if (alignment.MapQuality == FragmentBinnerConstants.MappingQualityNotAvailable ||
alignment.MapQuality < qualityThreshold)
{
return(true);
}
return(false);
}
public void FromBam()
{
var alignment = new BamAlignment
{
Bases = "ATCTTA",
Position = 100,
MatePosition = 500,
Name = "test",
CigarData = new CigarAlignment("5M1S"),
MapQuality = 10,
Qualities = new[] { (byte)10, (byte)20, (byte)30 }
};
alignment.SetIsDuplicate(true);
alignment.SetIsProperPair(true);
alignment.SetIsSecondaryAlignment(true);
alignment.SetIsUnmapped(true);
var read = new Read("chr1", alignment);
Assert.Equal(read.Chromosome, "chr1");
Assert.Equal(read.Sequence, alignment.Bases);
Assert.Equal(read.Position, alignment.Position + 1);
Assert.Equal(read.MatePosition, alignment.MatePosition + 1);
Assert.Equal(read.Name, alignment.Name);
Assert.Equal(read.CigarData, alignment.CigarData);
Assert.Equal(read.IsMapped, alignment.IsMapped());
Assert.Equal(read.IsProperPair, alignment.IsProperPair());
Assert.Equal(read.IsPrimaryAlignment, alignment.IsPrimaryAlignment());
Assert.Equal(read.IsPcrDuplicate, alignment.IsDuplicate());
foreach (var direction in read.SequencedBaseDirectionMap)
{
Assert.Equal(direction, DirectionType.Forward);
}
for (var i = 0; i < read.Qualities.Length; i++)
{
Assert.Equal(read.Qualities[i], alignment.Qualities[i]);
}
}
private PairStatus SingleReadStatus(BamAlignment alignment)
{
if ((alignment.RefID != alignment.MateRefID && alignment.IsPaired()))
{
return(PairStatus.SplitChromosomes); // Stitched reads will have split ref ids too but not the same thing
}
if (((!alignment.IsMateMapped() && alignment.RefID == -1) || (!alignment.IsMapped() && alignment.MateRefID == -1)))
{
return(PairStatus.MateUnmapped);
}
if (alignment.IsDuplicate())
{
return(PairStatus.Duplicate);
}
if (_considerInsertSize)
{
if (alignment.IsPaired() && !OverlapsMate(alignment))
{
return(PairStatus.LongFragment);
}
}
return(PairStatus.Unknown);
}
/// <summary>
/// Step 2: Get the ref and variant allele frequencies for the variants of interest, in the tumor bam file.
/// </summary>
protected void ProcessBamFile(string bamPath)
{
Console.WriteLine("{0} Looping over bam records from {1}", DateTime.Now, bamPath);
int overallCount = 0;
int nextVariantIndex = 0;
using (BamReader reader = new BamReader(bamPath))
{
BamAlignment read = new BamAlignment();
int refID = reader.GetReferenceIndex(this.Chromosome);
if (refID < 0)
{
throw new ArgumentException(string.Format("Error: Chromosome name '{0}' does not match bam file at '{1}'", this.Chromosome, bamPath));
}
Console.WriteLine("Jump to refid {0} {1}", refID, this.Chromosome);
reader.Jump(refID, 0);
while (true)
{
bool result = reader.GetNextAlignment(ref read, false);
if (!result) break;
if (!read.HasPosition() || read.RefID > refID) break; // We're past our chromosome of interest.
if (read.RefID < refID) continue; // We're not yet on our chromosome of interest.
overallCount++;
if (overallCount % 1000000 == 0)
{
Console.WriteLine("Record {0} at {1}...", overallCount, read.Position);
}
// Skip over unaligned or other non-count-worthy reads:
if (!read.IsPrimaryAlignment()) continue;
if (!read.IsMapped()) continue;
if (read.IsDuplicate()) continue;
if (read.MapQuality <= MinimumMapQ) continue;
// Scan forward through the variants list, to keep up with our reads:
while (nextVariantIndex < this.Variants.Count && this.Variants[nextVariantIndex].ReferencePosition < read.Position)
{
nextVariantIndex++;
}
if (nextVariantIndex >= this.Variants.Count) break;
// If the read doesn't look like it has a reasonable chance of touching the next variant, continue:
if (read.Position + 1000 < this.Variants[nextVariantIndex].ReferencePosition) continue;
// This read potentially overlaps next variant (and further variants). Count bases!
ProcessReadBases(read, nextVariantIndex);
}
}
Console.WriteLine("Looped over {0} bam records in all", overallCount);
}
public ReadPair GetNextEntryUntilNull()
{
var bamAlignment = new BamAlignment();
while (!_hasPassedChrom)
{
bool hasMoreReads;
hasMoreReads = _bamReader.GetNextAlignment(ref bamAlignment, false);
if (!hasMoreReads || _applyChrFilter && bamAlignment.RefID > _refId)
{
FlushUnpaired();
if (_applyChrFilter && bamAlignment.RefID > _refId)
{
_hasPassedChrom = true;
}
break;
}
if (_applyChrFilter && bamAlignment.RefID < _refId)
{
AddStatus(TooLow);
continue;
}
if (bamAlignment.IsDuplicate() && _skipAndRemoveDuplicates)
{
AddStatus(Duplicate);
continue;
}
// TODO did we want to handle secondary/supplementaries differently?
//if (bamAlignment.IsSecondary() || !bamAlignment.IsPrimaryAlignment())
//{
// continue;
//}
if (ShouldSkipRead(bamAlignment))
{
continue;
}
var status = SingleReadStatus(bamAlignment);
if (status != PairStatus.Unknown)
{
AddStatus(WontPair);
return(new ReadPair(bamAlignment)
{
PairStatus = status
});
}
var filteredReadPair = _filter.TryPair(bamAlignment, status);
if (filteredReadPair != null)
{
if (filteredReadPair.PairStatus != PairStatus.OffTarget)
{
filteredReadPair.PairStatus = PairStatus.Paired;
}
return(filteredReadPair);
}
}
while (true)
{
if (_unpaired == null || !_unpaired.Any())
{
break;
}
return(_unpaired.Dequeue());
}
return(null);
}
/// <summary>
/// Step 2: Get the ref and variant allele frequencies for the variants of interest, in the tumor bam file.
/// </summary>
protected void ProcessBamFile(string bamPath)
{
Console.WriteLine("{0} Looping over bam records from {1}", DateTime.Now, bamPath);
int overallCount = 0;
int nextVariantIndex = 0;
using (BamReader reader = new BamReader(bamPath))
{
BamAlignment read = new BamAlignment();
int refID = reader.GetReferenceIndex(this.Chromosome);
if (refID < 0)
{
throw new ArgumentException(string.Format("Error: Chromosome name '{0}' does not match bam file at '{1}'", this.Chromosome, bamPath));
}
Console.WriteLine("Jump to refid {0} {1}", refID, this.Chromosome);
reader.Jump(refID, 0);
while (true)
{
bool result = reader.GetNextAlignment(ref read, false);
if (!result)
{
break;
}
if (!read.HasPosition() || read.RefID > refID)
{
break; // We're past our chromosome of interest.
}
if (read.RefID < refID)
{
continue; // We're not yet on our chromosome of interest.
}
overallCount++;
if (overallCount % 1000000 == 0)
{
Console.WriteLine("Record {0} at {1}...", overallCount, read.Position);
}
// Skip over unaligned or other non-count-worthy reads:
if (!read.IsPrimaryAlignment())
{
continue;
}
if (!read.IsMapped())
{
continue;
}
if (read.IsDuplicate())
{
continue;
}
if (read.MapQuality <= MinimumMapQ)
{
continue;
}
// Scan forward through the variants list, to keep up with our reads:
while (nextVariantIndex < this.Variants.Count && this.Variants[nextVariantIndex].ReferencePosition < read.Position)
{
nextVariantIndex++;
}
if (nextVariantIndex >= this.Variants.Count)
{
break;
}
// If the read doesn't look like it has a reasonable chance of touching the next variant, continue:
if (read.Position + 1000 < this.Variants[nextVariantIndex].ReferencePosition)
{
continue;
}
// This read potentially overlaps next variant (and further variants). Count bases!
ProcessReadBases(read, nextVariantIndex);
}
}
Console.WriteLine("Looped over {0} bam records in all", overallCount);
}
public bool IsDuplicate(BamAlignment alignment)
{
return(alignment.IsDuplicate());
}
/// <summary>
/// Reads in a bam file and marks within the BitArrays which genomic mers are present.
/// </summary>
/// <param name="bamFile">bam file read alignments from.</param>
/// <param name="observedAlignments">Dictioanry of BitArrays, one for each chromosome, to store the alignments in.</param>
static void LoadObservedAlignmentsBAM(string bamFile, bool isPairedEnd, string chromosome, CanvasCoverageMode coverageMode, HitArray observed, Int16[] fragmentLengths)
{
// Sanity check: The .bai file must exist, in order for us to seek to our target chromosome!
string indexPath = bamFile + ".bai";
if (!File.Exists(indexPath))
{
throw new Exception(string.Format("Fatal error: Bam index not found at {0}", indexPath));
}
using (BamReader reader = new BamReader(bamFile))
{
int desiredRefIndex = -1;
desiredRefIndex = reader.GetReferenceIndex(chromosome);
if (desiredRefIndex == -1)
{
throw new ApplicationException(
string.Format("Unable to retrieve the reference sequence index for {0} in {1}.", chromosome,
bamFile));
}
bool result = reader.Jump(desiredRefIndex, 0);
if (!result)
{
// Note: This is not necessarily an error, it just means that there *are* no reads for this chromosome in this
// .bam file. That is not uncommon e.g. for truseq amplicon.
return;
}
int readCount = 0;
int keptReadCount = 0;
string header = reader.GetHeader();
BamAlignment alignment = new BamAlignment();
while (reader.GetNextAlignment(ref alignment, true))
{
readCount++;
// Flag check - Require reads to be aligned, passing filter, non-duplicate:
if (!alignment.IsMapped())
{
continue;
}
if (alignment.IsFailedQC())
{
continue;
}
if (alignment.IsDuplicate())
{
continue;
}
if (alignment.IsReverseStrand())
{
continue;
}
if (!alignment.IsMainAlignment())
{
continue;
}
// Require the alignment to start with 35 bases of non-indel:
if (alignment.CigarData[0].Type != 'M' || alignment.CigarData[0].Length < 35)
{
continue;
}
if (isPairedEnd && !alignment.IsProperPair())
{
continue;
}
int refID = alignment.RefID;
// quit if the current reference index is different from the desired reference index
if (refID != desiredRefIndex)
{
break;
}
if (refID == -1)
{
continue;
}
keptReadCount++;
if (coverageMode == CanvasCoverageMode.Binary)
{
observed.Data[alignment.Position] = 1;
}
else
{
observed.Set(alignment.Position);
}
// store fragment size, make sure it's within Int16 range and is positive (simplification for now)
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
{
fragmentLengths[alignment.Position] = Convert.ToInt16(Math.Max(Math.Min(Int16.MaxValue, alignment.FragmentLength), 0));
}
}
Console.WriteLine("Kept {0} of {1} total reads", keptReadCount, readCount);
}
}
/// <summary>
/// Reads in a bam file and marks within the BitArrays which genomic mers are present.
/// </summary>
/// <param name="bamFile">bam file read alignments from.</param>
/// <param name="observedAlignments">Dictioanry of BitArrays, one for each chromosome, to store the alignments in.</param>
static void LoadObservedAlignmentsBAM(string bamFile, bool isPairedEnd, string chromosome, CanvasCoverageMode coverageMode, HitArray observed, Int16[] fragmentLengths)
{
// Sanity check: The .bai file must exist, in order for us to seek to our target chromosome!
string indexPath = bamFile + ".bai";
if (!File.Exists(indexPath))
{
throw new Exception(string.Format("Fatal error: Bam index not found at {0}", indexPath));
}
using (BamReader reader = new BamReader(bamFile))
{
int desiredRefIndex = -1;
desiredRefIndex = reader.GetReferenceIndex(chromosome);
if (desiredRefIndex == -1)
{
throw new ApplicationException(
string.Format("Unable to retrieve the reference sequence index for {0} in {1}.", chromosome,
bamFile));
}
bool result = reader.Jump(desiredRefIndex, 0);
if (!result)
{
// Note: This is not necessarily an error, it just means that there *are* no reads for this chromosome in this
// .bam file. That is not uncommon e.g. for truseq amplicon.
return;
}
int readCount = 0;
int keptReadCount = 0;
string header = reader.GetHeader();
BamAlignment alignment = new BamAlignment();
while (reader.GetNextAlignment(ref alignment, true))
{
readCount++;
// Flag check - Require reads to be aligned, passing filter, non-duplicate:
if (!alignment.IsMapped()) continue;
if (alignment.IsFailedQC()) continue;
if (alignment.IsDuplicate()) continue;
if (alignment.IsReverseStrand()) continue;
if (!alignment.IsMainAlignment()) continue;
// Require the alignment to start with 35 bases of non-indel:
if (alignment.CigarData[0].Type != 'M' || alignment.CigarData[0].Length < 35) continue;
if (isPairedEnd && !alignment.IsProperPair()) continue;
int refID = alignment.RefID;
// quit if the current reference index is different from the desired reference index
if (refID != desiredRefIndex)
break;
if (refID == -1)
continue;
keptReadCount++;
if (coverageMode == CanvasCoverageMode.Binary)
{
observed.Data[alignment.Position] = 1;
}
else
{
observed.Set(alignment.Position);
}
// store fragment size, make sure it's within Int16 range and is positive (simplification for now)
if (coverageMode == CanvasCoverageMode.GCContentWeighted)
fragmentLengths[alignment.Position] = Convert.ToInt16(Math.Max(Math.Min(Int16.MaxValue, alignment.FragmentLength), 0));
}
Console.WriteLine("Kept {0} of {1} total reads", keptReadCount, readCount);
}
}
/// <summary>
/// Checks if any of the conditions is true:
/// 1. The read is a duplicate,
/// 2. The read failed QC,
/// 3. The read is of low mapping quality.
/// </summary>
/// <param name="alignment"></param>
/// <returns></returns>
public static bool IsDuplicateFailedQCLowQuality(BamAlignment alignment, uint qualityThreshold)
{
if (alignment.IsDuplicate()) { return true; }
if (alignment.IsFailedQC()) { return true; }
if (alignment.MapQuality == FragmentBinnerConstants.MappingQualityNotAvailable
|| alignment.MapQuality < qualityThreshold)
{
return true;
}
return false;
}
public BamAlignment GetFinalAlignment(BamAlignment origBamAlignment, out bool changed, out bool forcedSoftclip, out bool confirmed, out bool sketchy,
List <PreIndel> selectedIndels = null, List <PreIndel> existingIndels = null,
bool assumeImperfect = true, List <HashableIndel> confirmedAccepteds = null, List <PreIndel> mateIndels = null)
{
sketchy = false;
forcedSoftclip = false;
bool forcedAlignment = false;
var presumeStartPositionForForcedAlignment = 0;
if (origBamAlignment.CigarData.Count == 0)
{
// This was something weird that came up in the halo dataset... mapq is 0 but is still mapped, no cigar
if (origBamAlignment.Position <= 0 && origBamAlignment.FragmentLength != 0) // No sense trying to fiddle with the position otherwise
{
// TODO does this really even move the needle? Is it helping enough to outweigh its weirdness?
var presumedEndPosition = origBamAlignment.MatePosition < origBamAlignment.Position
? origBamAlignment.MatePosition - origBamAlignment.FragmentLength
: origBamAlignment.MatePosition + origBamAlignment.FragmentLength;
presumeStartPositionForForcedAlignment = presumedEndPosition - origBamAlignment.Bases.Length;
forcedAlignment = true;
}
else
{
presumeStartPositionForForcedAlignment = origBamAlignment.Position;
forcedAlignment = true;
}
}
var anyIndelsAtAll = _regionFilterer.AnyIndelsNearby(origBamAlignment.Position);
bool isRealignable = true;
if (anyIndelsAtAll)
{
var isImperfectRead = false || ((origBamAlignment.ContainsDisallowedCigarOps(_suspectCigarOps) ||
origBamAlignment.GetIntTag("NM") > 0 || forcedAlignment));
var isReadWorthCaringAbout = !origBamAlignment.IsDuplicate() && !origBamAlignment.IsSecondary();
isRealignable = isImperfectRead && isReadWorthCaringAbout && origBamAlignment.Bases.Distinct().Count() > 1;
}
else
{
_statusCounter.AddStatusCount("No indels nearby at all");
isRealignable = false;
}
if (!isRealignable)
{
confirmed = false;
changed = false;
sketchy = false;
return(origBamAlignment);
}
// TODO maybe flag (or return all) if there's a lot or high quality stuff that we're missing! Esp with pair specific
var indels = _indelSource.GetRelevantIndels(forcedAlignment ? presumeStartPositionForForcedAlignment : origBamAlignment.Position,
mateIndels, confirmedAccepteds);
// Don't realign around single indels if we already have them
bool hasExistingUnsanctionedIndels = false;
bool existingSanctionedIndelIsBest = false;
bool hasVeryGoodIndel = false;
bool hasHardToCallIndel = false;
var existingMatches = new List <PreIndel>();
HashableIndel existingConfirmedIndel = new HashableIndel();
var existingMatchHashables = new List <HashableIndel>();
if (indels.Any() && existingIndels != null && existingIndels.Any())
{
var topScore = (float)(indels.Max(x => x.Key.Score));
var matchesFound = 0;
var nonPreExistingIndels = new List <KeyValuePair <HashableIndel, GenomeSnippet> >();
var index = 0;
foreach (var kvp in indels)
{
var indel = kvp.Key;
var matches = existingIndels.Where(e => Helper.IsMatch(e, indel));
var isMatch = matches.Any();
if (isMatch)
{
matchesFound++;
if (!indel.InMulti && index == 0)
{
existingSanctionedIndelIsBest = true;
existingConfirmedIndel = indel;
}
var proportionOfTopScore = indel.Score / (float)topScore;
if (proportionOfTopScore >= 0.75)
{
hasVeryGoodIndel = true;
}
if (indel.HardToCall)
{
hasHardToCallIndel = true;
}
existingMatches.AddRange(matches);
// TODO do we need special handling of multis?
existingMatchHashables.Add(indel);
}
if (!isMatch || indel.InMulti)
{
nonPreExistingIndels.Add(kvp);
}
index++;
}
// TODO do we actually want to replace indels with non-pre-existing only?
indels = nonPreExistingIndels;
if (matchesFound == 0)
{
hasExistingUnsanctionedIndels = true;
}
}
// TODO this precludes us from having good multis
if (existingSanctionedIndelIsBest)
{
// If it already had the top ranked indel, there's not really any point in trying to realign around others (here we assume that it's also the best fitting indel for the read, hence why it was originally called by the regular aligner).
_statusCounter.AddStatusCount("Existing indel is already the best available");
changed = false;
confirmed = true;
UpdateOutcomeForConfirmed(existingConfirmedIndel);
if (confirmedAccepteds == null)
{
confirmedAccepteds = new List <HashableIndel>();
}
confirmedAccepteds.Add(existingConfirmedIndel);
return(origBamAlignment);
}
if (!indels.Any() || origBamAlignment.EndPosition - origBamAlignment.Position > 500)
{
if (!indels.Any())
{
// TODO maybe do the forced softclip here if the read did have indels?
_statusCounter.AddStatusCount("No indels to realign to");
_statusCounter.AppendStatusStringTag("RX", $"{origBamAlignment.GetStringTag("RX")},No indels to realign to", origBamAlignment);
}
else
{
_statusCounter.AddStatusCount("Alignment reference span longer than we can realign to");
}
changed = false;
confirmed = false;
return(origBamAlignment);
}
// TODO this should relate to cap on indel size... introducing too large of an indel will make us go beyond this context.
var context = indels.First().Value;
var orderedIndels = indels.Select(x => x.Key).ToList();
var numIndels = orderedIndels.Count;
_statusCounter.AddStatusCount("Realigning to " + numIndels);
var bamAlignment = new BamAlignment(origBamAlignment);
if (forcedAlignment)
{
bamAlignment.CigarData = new CigarAlignment(origBamAlignment.Bases.Length + "M");
bamAlignment.Position = presumeStartPositionForForcedAlignment;
}
var realignResult = _readRealigner.Realign(new Read(_chromosome, bamAlignment),
orderedIndels, indels.ToDictionary(x => x.Key, x => x.Value), confirmedAccepteds != null && confirmedAccepteds.Any());
var acceptedIndels = realignResult?.AcceptedIndels;
var hasAnyIndels = acceptedIndels != null && acceptedIndels.Any();
if (realignResult != null)
{
_statusCounter.AddStatusCount("Able to realign at all (may still be worse than original)");
_statusCounter.AppendStatusStringTag("RX", "Able to realign at all(may still be worse than original)", bamAlignment);
}
else
{
_statusCounter.AddStatusCount("Not able to realign at all");
_statusCounter.AppendStatusStringTag("RX", "Not able to realign at all", origBamAlignment);
}
AlignmentSummary originalAlignmentSummary = null;
var realignmentUnchanged = true;
if (realignResult != null)
{
originalAlignmentSummary =
Extensions.GetAlignmentSummary((new Read(_chromosome, origBamAlignment)), context.Sequence,
_trackActualMismatches, _checkSoftclipsForMismatches, context.StartPosition);
realignmentUnchanged = _judger.RealignmentIsUnchanged(realignResult, origBamAlignment);
if (originalAlignmentSummary.NumMismatches > 0)
{
// TODO PERF do we still want to use this ever?
var sumMismatch = Helper.GetSumOfMismatchQualities(origBamAlignment.Qualities,
origBamAlignment.Bases, new Read(_chromosome, origBamAlignment).PositionMap, context.Sequence,
context.StartPosition);
originalAlignmentSummary.SumOfMismatchingQualities = sumMismatch;
}
// Within this logic also checking the same as "!realignmentUnchanged" above.. consolidate this.
if (selectedIndels != null &&
(_judger.RealignmentBetterOrEqual(realignResult, originalAlignmentSummary, confirmedAccepteds != null && confirmedAccepteds.Any())) ||
ResultIsGoodEnough(realignResult, origBamAlignment, originalAlignmentSummary,
realignmentUnchanged, confirmedAccepteds != null && confirmedAccepteds.Any()))
{
UpdateIndelOutcomes(numIndels, orderedIndels, hasAnyIndels, acceptedIndels, confirmedAccepteds, true, realignResult);
if (realignResult.IsSketchy)
{
sketchy = true;
}
return(AcceptRealignment(origBamAlignment, out changed, selectedIndels, existingIndels, realignResult, originalAlignmentSummary, bamAlignment, hasExistingUnsanctionedIndels, out confirmed));
}
}
// At this point, any good realignment would have been returned. If it's realigned and changed now, it's an unaccepted (not good enough) realignment.
// If it had an indel to begin with, it's basically a vote that we don't trust that indel. Optionally softclip it out.
if (!realignmentUnchanged)
{
changed = false;
confirmed = false;
HandleFailedRealignment(origBamAlignment, ref forcedSoftclip, existingIndels, realignResult, hasExistingUnsanctionedIndels, existingMatches);
if ((hasVeryGoodIndel || (hasHardToCallIndel && _judger.IsVeryConfident(originalAlignmentSummary))) && !hasExistingUnsanctionedIndels && existingMatchHashables.Any())
{
// It didn't have the tip-top indel, but it had one that was very close, and we tried realigning around the top guys and failed - this one looks better. Give it credit.
confirmed = true;
foreach (var indel in existingMatchHashables)
{
UpdateOutcomeForConfirmed(indel);
if (confirmedAccepteds != null)
{
confirmedAccepteds.Add(indel);
}
}
}
UpdateIndelOutcomes(numIndels, orderedIndels, hasAnyIndels, acceptedIndels, confirmedAccepteds, false, realignResult);
}
else
{
if (acceptedIndels != null)
{
foreach (var indelNum in acceptedIndels)
{
var indel = orderedIndels[indelNum];
UpdateOutcomeForConfirmed(indel);
}
}
_statusCounter.AddStatusCount("INDEL STATUS\tUnchanged\t" + realignResult?.Indels);
_statusCounter.AppendStatusStringTag("RX", "Unchanged: " + realignResult?.Indels, origBamAlignment);
confirmed = true;
changed = false;
return(origBamAlignment);
}
if (realignResult == null)
{
if (_softclipUnknownIndels && hasExistingUnsanctionedIndels)
{
var unsanctioned = existingIndels.Where(x => !existingMatches.Contains(x));
foreach (var preIndel in unsanctioned.OrderBy(x => x.ReferencePosition))
{
var reverseClip = false;
var clipLength = preIndel.RightAnchor;
if (preIndel.LeftAnchor < preIndel.RightAnchor)
{
reverseClip = true;
clipLength = preIndel.LeftAnchor;
}
// TODO arbitrary number here...
// If it's pretty well-anchored, don't remove the indel
if (clipLength > 20)
{
continue;
}
forcedSoftclip = true;
_statusCounter.AddStatusCount("Softclipped out bad indel");
_statusCounter.AppendStatusStringTag("RX",
$"Softclipped out bad indel({origBamAlignment.CigarData},{string.Join(",", existingIndels)}... No realignment",
origBamAlignment);
_statusCounter.AddStatusCount("INDEL STATUS\tRemoved\t" + string.Join("|", existingIndels));
OverlappingIndelHelpers.SoftclipAfterIndel(origBamAlignment,
reverseClip, preIndel.ReferencePosition);
}
}
}
_statusCounter.AppendStatusStringTag("RX", "Realignment failed", origBamAlignment);
_statusCounter.AddStatusCount("Realignment failed");
return(origBamAlignment);
}
