private void WriteChromosomeReads(string inputBamFile, string chrName, IRealignmentWriter writer)
{
using (var extractor = _factory.CreateAlignmentExtractor(inputBamFile, chrName))
{
var read = new Read();
while (extractor.GetNextAlignment(read))
{
var bamAlignment = read.BamAlignment;
writer.WriteRead(ref bamAlignment, false);
}
}
writer.FlushAllBufferedRecords();
}
// todo add documentation on what null
private void Realign(int?upToPosition)
{
var batch = _stateManager.GetCandidatesToProcess(upToPosition);
var candidateIndelGroups = _stateManager.GetCandidateGroups(upToPosition);
if (batch == null ||
(upToPosition.HasValue && batch.ClearedRegions == null)) // nothing to do
{
return;
}
// grab candidates and rank them
var allTargets = batch.GetCandidates().Where(c => c.Length <= _maxIndelSize).Select(c => new CandidateIndel(c)).ToList();
var goodTargets = _caller.Call(allTargets, _stateManager).OrderBy(t => t.ReferencePosition).ToList();
var maxTargetSize = 0;
if (goodTargets.Any())
{
maxTargetSize = goodTargets.Max(t => t.Length) + 1;
}
// realign reads
do
{
if (_lastExtractedRead == null) // first time around, get aligment
{
_lastExtractedRead = new Read();
if (!_extractorForRealign.GetNextAlignment(_lastExtractedRead))
{
break;
}
}
if (batch.MaxClearedPosition.HasValue && _lastExtractedRead.Position >= batch.MaxClearedPosition)
{
break;
}
var bamAlignment = _lastExtractedRead.BamAlignment;
// get original alignment summary
var originalAlignmentSummary = _lastExtractedRead.GetAlignmentSummary(_chrReference.Sequence);
if (_skipAndRemoveDuplicates && bamAlignment.IsDuplicate())
{
continue;
}
if (!goodTargets.Any() ||
!bamAlignment.IsPrimaryAlignment() ||
(_skipDuplicates && bamAlignment.IsDuplicate()) ||
bamAlignment.IsSupplementaryAlignment() ||
bamAlignment.HasSupplementaryAlignment() ||
PassesSuspicion(originalAlignmentSummary)) // skip reads that are or have supplementary alignments
{
_writer.WriteRead(ref bamAlignment, false);
continue;
}
#if false
Console.WriteLine("Original read has {0} mismatches {1} indels", originalAlignmentSummary.NumMismatches, originalAlignmentSummary.NumIndels);
#endif
// try realigning
// take realignment even if it's equal to original, this gives opportunity to known variants
var realignResult = _readRealigner.Realign(_lastExtractedRead, goodTargets, _chrReference.Sequence,
_indelRanker, candidateIndelGroups, maxTargetSize);
if (realignResult != null && RealignmentIsWithinRange(realignResult, bamAlignment) &&
!RealignmentIsUnchanged(realignResult, bamAlignment) &&
RealignmentBetterOrEqual(realignResult, originalAlignmentSummary)
)
{
bamAlignment.Position = realignResult.Position - 1; // 0 base
bamAlignment.CigarData = realignResult.Cigar;
bamAlignment.UpdateIntTagData("NM", realignResult.NumMismatches + realignResult.NumIndelBases); // update NM tag (edit distance)
if (bamAlignment.MapQuality <= 20 && realignResult.NumMismatches == 0 && (_allowRescoringOrig0 || bamAlignment.MapQuality > 0))
{
bamAlignment.MapQuality = 40; // todo what to set this to?
}
_writer.WriteRead(ref bamAlignment, true);
TotalRealignedReads++;
}
else
{
if (realignResult != null && !RealignmentIsWithinRange(realignResult, bamAlignment))
{
Logger.WriteToLog(
string.Format(
"Realignment attempt resulted in an attempted shift of read '{6}' from {0}:{1}:{2} to {0}:{3}:{4}, which is larger than the max realign shift of {5}. Original read will be outputted.",
bamAlignment.RefID, bamAlignment.Position, bamAlignment.CigarData,
realignResult.Position - 1, realignResult.Cigar, _maxRealignShift, bamAlignment.Name));
}
// doesn't look any better
_writer.WriteRead(ref bamAlignment, false);
}
} while (_extractorForRealign.GetNextAlignment(_lastExtractedRead));
_stateManager.DoneProcessing(batch);
}
