public void CreateBamReader()
{
// Should return a fully functional bam reader
var tempPath = $"TemporaryBamFile_{Guid.NewGuid()}.bam";
if (File.Exists(tempPath))
{
File.Delete(tempPath);
}
using (var bamWriter = new BamWriter(tempPath, "header", new List <GenomeMetadata.SequenceMetadata>()))
{
var bamWriterHandle = new BamWriterHandle(bamWriter);
bamWriterHandle.WriteAlignment(TestHelpers.CreateBamAlignment("ATCG", 8, 10, 30, true));
bamWriterHandle.WriteAlignment(null);
}
Assert.True(File.Exists(tempPath));
var stitcherOptions = new StitcherOptions();
var factory = new GeminiDataSourceFactory(stitcherOptions, "fakeGenomePath", false);
using (var bamReader = factory.CreateBamReader(tempPath))
{
BamAlignment alignment = new BamAlignment();
var getNext = bamReader.GetNextAlignment(ref alignment, true);
Assert.True(getNext);
Assert.Equal(7, alignment.Position);
}
File.Delete(tempPath);
}
public void Dispose()
{
if (_writer != null)
{
_writer.Close();
_writer.Dispose();
_writer = null;
}
}
/// <summary>
/// We're finished processing a read. Add it to the read buffer, and flush
/// old nodes to disk.
/// </summary>
public void WriteRead(ref BamAlignment read, bool remapped)
{
if (remapped)
{
var info = new RemapInfo(read.Position,
read.Position + (int)read.CigarData.GetReferenceSpan() - 1);
_remappings[string.Format("{0}-{1}", read.Name, read.IsFirstMate() ? 1 : 2)] = info;
}
LinkedListNode <ReadsForPosition> node = _readBuffer.First;
LinkedListNode <ReadsForPosition> nextNode;
byte[] buffer = BamWriter.SerializeAlignment(ref read);
while (node != null)
{
// Flush reads that are earlier than the earliest possible read we could see in the future (i.e. 2x max shift less than current read).
// Reasoning: The realign shift could go in two directions. So, for example with max shift of 10, you could receive
// and buffer reads in an order like this: 100, 100, 110, 115 (originally 110, shifted right to 115),
// 100 (originally 110, shifted left to 100), 120 (originally 110, shifted right to 120),
// 101 (originally 110, shifted left to 100), 121, 130, etc. If you had been flushing reads
// using the max shift only as the threshold, upon hitting the 115 (fourth read) you would have
// flushed the 100s (because 100 < 115 - 10), even though there still may be 100s coming through.
// By the time we hit 121, we know that all of the reads we encounter in the future are going to be > 100 (at minimum, the 121 could represent a max-right-shift from 111 and other 111 reads could be max-left-shifted to 101).
if (node.Value.Position < read.Position - (_maxRealignShift * 2))
{
MinimumRealignmentStartPosition = Math.Max(MinimumRealignmentStartPosition, node.Value.Position);
nextNode = node.Next;
FlushBufferedRecords(node.Value);
_readBuffer.Remove(node);
node = nextNode;
continue;
}
if (node.Value.Position == read.Position)
{
node.Value.Reads.Add(buffer);
return;
}
if (node.Value.Position > read.Position)
{
ReadsForPosition reads = new ReadsForPosition();
reads.Position = read.Position;
reads.Reads.Add(buffer);
_readBuffer.AddBefore(node, reads);
return;
}
node = node.Next;
}
ReadsForPosition readList = new ReadsForPosition();
readList.Position = read.Position;
readList.Reads.Add(buffer);
_readBuffer.AddLast(readList);
}
private void AdjustMates(string tmpFile, BamWriter writer)
{
// Second pass: Adjust flags on mates
Logger.WriteToLog("Writing reads with corrected mate flags, {0} total remapped reads", _remappings.Count);
var read = new BamAlignment();
using (var reader = new BamReader(tmpFile))
{
while (true)
{
var result = reader.GetNextAlignment(ref read, false);
if (!result)
{
break;
}
// Adjust flags as needed:
var mateKey = string.Format("{0}-{1}", read.Name, read.IsFirstMate() ? 2 : 1);
RemapInfo info;
if (!_remappings.TryGetValue(mateKey, out info))
{
writer.WriteAlignment(read);
continue;
}
if (info.Start == -1)
{
read.SetIsMateUnmapped(true);
read.SetIsProperPair(false);
read.FragmentLength = 0;
}
else
{
read.MatePosition = info.Start;
}
if (read.IsMateMapped() && read.IsProperPair())
{
int readEnd = read.Position + (int)read.CigarData.GetReferenceSpan() - 1;
// todo jg - should FragmentLength be 0 if the reads are mapped to diff chrs
read.FragmentLength = (read.Position < info.Start
? info.End - read.Position + 1
: info.Start - readEnd - 1);
}
writer.WriteAlignment(read);
}
}
}
public void Process(string inputBam, string outFolder, StitcherOptions stitcherOptions)
{
var jobManager = new JobManager(10);
var jobs = new List <IJob>();
var perChromBams = new List <string>();
// Process each of the chromosomes separately
foreach (var chrom in _chroms)
{
var intermediateOutput = Path.Combine(outFolder, Path.GetFileNameWithoutExtension(inputBam) + "." + chrom + ".stitched.bam");
perChromBams.Add(intermediateOutput);
var stitcher = new BamStitcher(inputBam, intermediateOutput, stitcherOptions, chrFilter: chrom);
jobs.Add(new GenericJob(() => stitcher.Execute(), "Stitcher_" + chrom));
}
jobManager.Process(jobs);
// Combine the per-chromosome bams
Logger.WriteToLog("Writing final bam.");
var outputBam = Path.Combine(outFolder, Path.GetFileNameWithoutExtension(inputBam) + ".final.stitched.bam");
using (var finalOutput = new BamWriter(outputBam, _header, _references))
{
foreach (var bam in perChromBams)
{
Logger.WriteToLog("Adding " + bam + " to final bam.");
var bamAlignment = new BamAlignment();
using (var bamReader = new BamReader(bam))
{
while (true)
{
var hasMoreReads = bamReader.GetNextAlignment(ref bamAlignment, false);
if (!hasMoreReads)
{
break;
}
finalOutput.WriteAlignment(bamAlignment);
}
}
File.Delete(bam);
}
}
Logger.WriteToLog("Finished combining per-chromosome bams into final bam at " + outputBam);
}
public void Initialize()
{
var outputDirectory = Path.GetDirectoryName(_outputFile);
if (!Directory.Exists(outputDirectory))
{
Directory.CreateDirectory(outputDirectory);
}
using (var reader = new BamReader(_inputFile))
{
var genome = reader.GetReferences();
_bamWriter = new BamWriter(_temp1File, reader.GetHeader(), genome);
}
}
public void HappyPath()
{
//var bamFilePath = Path.Combine(TestPaths.SharedBamDirectory, "Chr17Chr19.bam");
//Assert.True(File.Exists(bamFilePath));
// TODO figure out how to access the shared bams
var tempPath = $"TemporaryBamFile_{Guid.NewGuid()}.bam";
if (File.Exists(tempPath))
{
File.Delete(tempPath);
}
using (var bamWriter = new BamWriter(tempPath, "header", new List <GenomeMetadata.SequenceMetadata>()))
{
bamWriter.WriteAlignment(TestHelpers.CreateBamAlignment("ATCG", 1, 10, 30, true));
}
var bamWriterFactory = new BamWriterFactory(1, tempPath);
var tempPath2 = $"TemporaryBamFile_{Guid.NewGuid()}.bam";
if (File.Exists(tempPath2))
{
File.Delete(tempPath2);
}
var bamWriterHandle = bamWriterFactory.CreateSingleBamWriter(tempPath2);
bamWriterHandle.WriteAlignment(TestHelpers.CreateBamAlignment("ATCAG", 1, 10, 30, true));
bamWriterHandle.WriteAlignment(null);
using (var reader = new BamReader(tempPath2))
{
// TODO more specific?
var header = reader.GetHeader();
Assert.Contains("ID:Gemini", header);
Assert.Contains("PN:Gemini", header);
}
File.Delete(tempPath);
File.Delete(tempPath2);
}
public void HappyPath()
{
var tempPath = $"TemporaryBamFile_{Guid.NewGuid()}.bam";
if (File.Exists(tempPath))
{
File.Delete(tempPath);
}
using (var bamWriter = new BamWriter(tempPath, "header", new List <GenomeMetadata.SequenceMetadata>()))
{
var bamWriterHandle = new BamWriterHandle(bamWriter);
bamWriterHandle.WriteAlignment(TestHelpers.CreateBamAlignment("ATCG", 1, 10, 30, true));
bamWriterHandle.WriteAlignment(null);
}
Assert.True(File.Exists(tempPath));
File.Delete(tempPath);
}
/// <summary>
/// Seek to the unaligned (and mate-unaligned) reads at the tail of the input file, and write them all out to the output file.
/// </summary>
private void WriteUnalignedReads(BamWriter writer)
{
Logger.WriteToLog("Writing unaligned reads");
using (var reader = new BamReader(_inputFile))
{
reader.JumpToUnaligned();
var read = new BamAlignment();
while (true)
{
var result = reader.GetNextAlignment(ref read, false);
if (!result)
{
break;
}
if (read.RefID != -1)
{
continue; // skip over last reads
}
writer.WriteAlignment(read);
}
}
}
public BamWriterHandle(BamWriter writer)
{
_writer = writer;
}
public BamWriterWrapper(BamWriter writer)
{
_writer = writer;
}
