private IFileLocation WriteManifest(NexteraManifest manifest, IDirectoryLocation sandbox)
{
var path = sandbox.GetFileLocation(manifest.Name);
NexteraManifestUtils.WriteNexteraManifests(manifest, path.FullName);
return(path);
}
private CanvasCleanOutput InvokeCanvasClean(CanvasCallset callset, IFileLocation binnedPath)
{
StringBuilder commandLine = new StringBuilder();
commandLine.Length = 0;
string executablePath = Path.Combine(_canvasFolder, "CanvasClean.exe");
if (CrossPlatform.IsThisMono())
{
commandLine.AppendFormat("{0} ", executablePath);
executablePath = Utilities.GetMonoPath();
}
commandLine.AppendFormat("-i \"{0}\" ", binnedPath);
string cleanedPath = Path.Combine(callset.TempFolder, string.Format("{0}.cleaned", callset.Id));
commandLine.AppendFormat("-o \"{0}\" ", cleanedPath);
commandLine.AppendFormat("-g");
string ffpePath = null;
// TruSight Cancer has 1,737 targeted regions. The cut-off 2000 is somewhat arbitrary.
// TruSight One has 62,309 targeted regions.
// Nextera Rapid Capture v1.1 has 411,513 targeted regions.
if (!callset.IsEnrichment || callset.Manifest.Regions.Count > 2000)
{
ffpePath = Path.Combine(callset.TempFolder, "FilterRegions.txt");
commandLine.AppendFormat(" -s -r -f \"{0}\"", ffpePath);
}
if (callset.IsEnrichment) // manifest
{
if (!File.Exists(callset.TempManifestPath))
{
NexteraManifestUtils.WriteNexteraManifests(callset.Manifest, callset.TempManifestPath);
}
commandLine.AppendFormat(" -t \"{0}\"", callset.TempManifestPath);
}
UnitOfWork cleanJob = new UnitOfWork()
{
ExecutablePath = executablePath,
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = Path.GetFileName(cleanedPath),
CommandLine = commandLine.ToString()
};
if (_customParameters.ContainsKey("CanvasClean"))
{
cleanJob.CommandLine = Utilities.MergeCommandLineOptions(cleanJob.CommandLine, _customParameters["CanvasClean"], true);
}
_workManager.DoWorkSingleThread(cleanJob);
var canvasCleanOutput = new CanvasCleanOutput(new FileLocation(cleanedPath), new FileLocation(ffpePath));
return(canvasCleanOutput);
}
/// <summary>
/// Invoke CanvasNormalize.
/// </summary>
/// <param name="callset"></param>
/// <returns>path to the bin ratio bed file</returns>
protected string InvokeCanvasNormalize(CanvasCallset callset, string tumorBinnedPath, Dictionary <string, string> bamToBinned,
string ploidyBedPath, string mode = "weightedaverage")
{
string ratioBinnedPath = Path.Combine(callset.TempFolder, string.Format("{0}.ratio.binned", callset.Id));
string canvasNormalizePath = Path.Combine(_canvasFolder, "CanvasNormalize.exe");
string executablePath = canvasNormalizePath;
if (CrossPlatform.IsThisMono())
{
executablePath = Utilities.GetMonoPath();
}
StringBuilder commandLine = new StringBuilder();
if (CrossPlatform.IsThisMono())
{
commandLine.AppendFormat("{0} ", canvasNormalizePath);
}
commandLine.AppendFormat("-t {0} ", tumorBinnedPath.WrapWithShellQuote()); // tumor bed
if (callset.IsEnrichment && callset.Manifest.CanvasControlAvailable)
{
commandLine.AppendFormat("-n {0} ", callset.Manifest.CanvasControlBinnedPath.WrapWithShellQuote()); // normal bed
}
else
{
foreach (string normalBinnedPath in callset.NormalBamPaths.Select(path => bamToBinned[path.BamFile.FullName]))
{
commandLine.AppendFormat("-n {0} ", normalBinnedPath.WrapWithShellQuote()); // normal bed
}
}
commandLine.AppendFormat("-w {0} ", callset.NormalBinnedPath.WrapWithShellQuote()); // weighted average normal bed
commandLine.AppendFormat("-o {0} ", ratioBinnedPath.WrapWithShellQuote()); // ratio bed
if (callset.IsEnrichment) // manifest
{
if (!File.Exists(callset.TempManifestPath))
{
NexteraManifestUtils.WriteNexteraManifests(callset.Manifest, callset.TempManifestPath);
}
commandLine.AppendFormat("-f {0} ", callset.TempManifestPath.WrapWithShellQuote());
}
commandLine.AppendFormat("-m {0} ", mode.WrapWithShellQuote());
if (!string.IsNullOrEmpty(ploidyBedPath))
{
commandLine.AppendFormat("-p {0} ", ploidyBedPath.WrapWithShellQuote());
}
UnitOfWork normalizeJob = new UnitOfWork()
{
ExecutablePath = executablePath,
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = Path.GetFileName(ratioBinnedPath),
CommandLine = commandLine.ToString()
};
if (_customParameters.ContainsKey("CanvasNormalize"))
{
normalizeJob.CommandLine = Utilities.MergeCommandLineOptions(normalizeJob.CommandLine, _customParameters["CanvasNormalize"], true);
}
_workManager.DoWorkSingleThread(normalizeJob);
return(ratioBinnedPath);
}
private int GetBinSize(CanvasCallset callset, string bamPath, List <string> intermediateDataPaths,
string canvasReferencePath, string canvasBedPath)
{
string canvasBinPath = Path.Combine(_canvasFolder, "CanvasBin.exe");
string executablePath = canvasBinPath;
if (CrossPlatform.IsThisMono())
{
executablePath = Utilities.GetMonoPath();
}
StringBuilder commandLine = new StringBuilder();
if (CrossPlatform.IsThisMono())
{
commandLine.AppendFormat("{0} ", canvasBinPath);
}
commandLine.AppendFormat("-b \"{0}\" ", bamPath);
commandLine.AppendFormat("-p "); // Paired-end input mode (Isaac or BWA output)
commandLine.AppendFormat("-r \"{0}\" ", canvasReferencePath);
foreach (string path in intermediateDataPaths)
{
commandLine.AppendFormat("-i \"{0}\" ", path);
}
commandLine.AppendFormat("-y "); // bin size only
if (callset.IsEnrichment) // manifest
{
if (!File.Exists(callset.TempManifestPath))
{
NexteraManifestUtils.WriteNexteraManifests(callset.Manifest, callset.TempManifestPath);
}
commandLine.AppendFormat("-t \"{0}\" ", callset.TempManifestPath);
}
string outputStub = Path.Combine(Path.GetDirectoryName(callset.BinSizePath), Path.GetFileNameWithoutExtension(callset.BinSizePath));
commandLine.AppendFormat("-f \"{0}\" -d {1} -o \"{2}\"", canvasBedPath, _countsPerBin, outputStub);
UnitOfWork binJob = new UnitOfWork()
{
ExecutablePath = executablePath,
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = Path.GetFileNameWithoutExtension(callset.BinSizePath),
CommandLine = commandLine.ToString()
};
if (_customParameters.ContainsKey("CanvasBin"))
{
binJob.CommandLine = Utilities.MergeCommandLineOptions(binJob.CommandLine, _customParameters["CanvasBin"], true);
}
_workManager.DoWorkSingleThread(binJob);
int binSize;
using (StreamReader reader = new StreamReader(callset.BinSizePath))
{
binSize = int.Parse(reader.ReadLine());
}
return(binSize);
}
/// <summary>
/// Invoke CanvasBin. Return null if this fails and we need to abort CNV calling for this sample.
/// </summary>
protected IFileLocation InvokeCanvasBin(CanvasCallset callset, string canvasReferencePath, string canvasBedPath, string ploidyBedPath)
{
StringBuilder commandLine = new StringBuilder();
string canvasBinPath = Path.Combine(_canvasFolder, "CanvasBin.exe");
string executablePath = canvasBinPath;
if (CrossPlatform.IsThisMono())
{
executablePath = Utilities.GetMonoPath();
}
//use bam as input
if (callset.Bam == null)
{
Console.WriteLine("Input bam file not seen for sample {0}_{1} - no CNV calls", callset.SampleName, callset.Id);
return(null);
}
List <string> bamPaths = new List <string>();
bamPaths.Add(callset.Bam.BamFile.FullName);
if (!(callset.IsEnrichment && callset.Manifest.CanvasControlAvailable)) // do not add normal BAMs if Canvas Control is available
{
bamPaths.AddRange(callset.NormalBamPaths.Select(bam => bam.BamFile.FullName));
}
// loop over the reference sequences in that genome
GenomeMetadata genomeMetadata = callset.GenomeMetadata;
List <UnitOfWork> binJobs = new List <UnitOfWork>();
Dictionary <string, List <string> > intermediateDataPathsByBamPath = new Dictionary <string, List <string> >();
foreach (string bamPath in bamPaths)
{
intermediateDataPathsByBamPath[bamPath] = new List <string>();
}
for (int bamIndex = 0; bamIndex < bamPaths.Count; bamIndex++)
{
foreach (GenomeMetadata.SequenceMetadata sequenceMetadata in genomeMetadata.Sequences.OrderByDescending(sequence => sequence.Length))
{
// Only invoke CanvasBin for autosomes + allosomes;
// don't invoke it for mitochondrial chromosome or extra contigs or decoys
if (sequenceMetadata.Type != GenomeMetadata.SequenceType.Allosome && !sequenceMetadata.IsAutosome())
{
continue;
}
string bamPath = bamPaths[bamIndex];
commandLine.Clear();
if (CrossPlatform.IsThisMono())
{
commandLine.AppendFormat("{0} ", canvasBinPath);
}
commandLine.AppendFormat("-b \"{0}\" ", bamPath);
if (callset.Bam.IsPairedEnd)
{
commandLine.AppendFormat("-p ");
}
commandLine.AppendFormat("-r \"{0}\" ", canvasReferencePath);
commandLine.AppendFormat("-c {0} ", sequenceMetadata.Name);
commandLine.AppendFormat("-m {0} ", _coverageMode);
if (callset.IsEnrichment) // manifest
{
if (!File.Exists(callset.TempManifestPath))
{
NexteraManifestUtils.WriteNexteraManifests(callset.Manifest, callset.TempManifestPath);
}
commandLine.AppendFormat("-t \"{0}\" ", callset.TempManifestPath);
}
string intermediateDataPath = Path.Combine(callset.TempFolder, string.Format("{0}_{1}_{2}.dat",
callset.Id, bamIndex, sequenceMetadata.Name));
intermediateDataPathsByBamPath[bamPath].Add(intermediateDataPath);
commandLine.AppendFormat("-f \"{0}\" -d {1} -o \"{2}\" ", canvasBedPath, _countsPerBin, intermediateDataPath);
UnitOfWork binJob = new UnitOfWork()
{
ExecutablePath = executablePath,
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = Path.GetFileName(intermediateDataPath),
CommandLine = commandLine.ToString()
};
if (_customParameters.ContainsKey("CanvasBin"))
{
binJob.CommandLine = Utilities.MergeCommandLineOptions(binJob.CommandLine, _customParameters["CanvasBin"], true);
}
binJobs.Add(binJob);
}
}
_workManager.DoWorkParallelThreads(binJobs);
// get bin size (of the smallest BAM) if normal BAMs are given
int binSize = -1;
if (bamPaths.Count > 1)
{
string smallestBamPath = SmallestFile(bamPaths);
binSize = GetBinSize(callset, smallestBamPath, intermediateDataPathsByBamPath[smallestBamPath],
canvasReferencePath, canvasBedPath);
}
else if (callset.IsEnrichment && callset.Manifest.CanvasControlAvailable)
{
binSize = callset.Manifest.CanvasBinSize.Value;
}
Dictionary <string, string> bamToBinned = new Dictionary <string, string>();
List <UnitOfWork> finalBinJobs = new List <UnitOfWork>();
for (int bamIdx = 0; bamIdx < bamPaths.Count; bamIdx++)
{
string bamPath = bamPaths[bamIdx];
// finish up CanvasBin step by merging intermediate data and finally binning
string binnedPath = Path.Combine(callset.TempFolder, string.Format("{0}_{1}.binned", callset.Id, bamIdx));
bamToBinned[bamPath] = binnedPath;
commandLine.Clear();
if (CrossPlatform.IsThisMono())
{
commandLine.AppendFormat("{0} ", canvasBinPath);
}
commandLine.AppendFormat("-b \"{0}\" ", bamPath);
if (callset.Bam.IsPairedEnd)
{
commandLine.AppendFormat("-p ");
}
commandLine.AppendFormat("-r \"{0}\" ", canvasReferencePath);
commandLine.AppendFormat("-f \"{0}\" -d {1} -o \"{2}\" ", canvasBedPath, _countsPerBin, binnedPath);
if (binSize != -1)
{
commandLine.AppendFormat("-z \"{0}\" ", binSize);
}
foreach (string path in intermediateDataPathsByBamPath[bamPath])
{
commandLine.AppendFormat("-i \"{0}\" ", path);
}
commandLine.AppendFormat("-m {0} ", _coverageMode);
UnitOfWork finalBinJob = new UnitOfWork()
{
ExecutablePath = executablePath,
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = Path.GetFileName(binnedPath),
CommandLine = commandLine.ToString()
};
if (_customParameters.ContainsKey("CanvasBin"))
{
finalBinJob.CommandLine = Utilities.MergeCommandLineOptions(finalBinJob.CommandLine, _customParameters["CanvasBin"], true);
}
finalBinJobs.Add(finalBinJob);
}
_workManager.DoWorkParallel(finalBinJobs, new TaskResourceRequirements(8, 25)); // CanvasBin itself is multi-threaded
string tumorBinnedPath = bamToBinned[callset.Bam.BamFile.FullName]; // binned tumor sample
string outputPath = tumorBinnedPath;
if (callset.NormalBamPaths.Any() || (callset.IsEnrichment && callset.Manifest.CanvasControlAvailable))
{
outputPath = InvokeCanvasNormalize(callset, tumorBinnedPath, bamToBinned, ploidyBedPath);
}
return(new FileLocation(outputPath));
}
