private CanvasOutput RunSingleSample(string sampleId, CanvasResequencingInput input, IDirectoryLocation sampleSandbox)
{
if (!_annotationFileProvider.IsSupported(input.GenomeMetadata))
{
_logger.Info($"Skipping Canvas for sample {sampleId}: unsupported reference genome '{input.GenomeMetadata.Name}'");
return(null);
}
if (!_annotationFileProvider.CustomDbSnpVcf(input.GenomeMetadata) && input.Vcf == null)
{
_logger.Info($"Skipping Canvas for sample {sampleId}. A dbSNP VCF file was not provided and no small variant VCF file is available");
return(null);
}
StringBuilder commandLine = new StringBuilder("Germline-WGS");
commandLine.Append(_singleSampleInputCommandLineBuilder.GetSingleSampleCommandLine(sampleId, input.Bam, input.GenomeMetadata, sampleSandbox));
// use sample vcf by default (performance could be similar with dbSNP vcf though)
var bAlleleVcf = input.Vcf.VcfFile;
var bAlleleVcfOptionName = SingleSampleCommonOptionsParser.SampleBAlleleVcfOptionName;
if (_annotationFileProvider.CustomDbSnpVcf(input.GenomeMetadata))
{
bAlleleVcf = _annotationFileProvider.GetDbSnpVcf(input.GenomeMetadata);
bAlleleVcfOptionName = SingleSampleCommonOptionsParser.PopulationBAlleleVcfOptionName;
}
commandLine.Append($" --{bAlleleVcfOptionName} \"{bAlleleVcf}\"");
IFileLocation ploidyBed = _canvasPloidyBedCreator.CreateGermlinePloidyBed(input.Vcf, input.GenomeMetadata, sampleSandbox);
if (ploidyBed != null)
{
commandLine.Append($" --{SingleSampleCommonOptionsParser.PloidyBedOptionName} \"{ploidyBed}\"");
}
var canvasPartitionParam = $@"--commoncnvs {_annotationFileProvider.GetCanvasAnnotationFile(input.GenomeMetadata, "commoncnvs.bed")}";
var moreCustomParameters = new Dictionary <string, string>();
moreCustomParameters["CanvasPartition"] = canvasPartitionParam;
commandLine.Append(_singleSampleInputCommandLineBuilder.GetCustomParameters(moreCustomParameters));
commandLine = _singleSampleInputCommandLineBuilder.MergeCustomCanvasParameters(commandLine);
UnitOfWork singleSampleJob = new UnitOfWork()
{
ExecutablePath = CrossPlatform.IsThisLinux() ? _runtimeExecutable.FullName : _canvasExe.FullName,
CommandLine = CrossPlatform.IsThisLinux() ? _canvasExe + " " + commandLine : commandLine.ToString(),
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = "Canvas_" + sampleId,
};
_workManager.DoWorkSingleThread(singleSampleJob);
return(GetCanvasOutput(sampleId, sampleSandbox));
}
private CanvasOutput RunSingleSample(string sampleId, CanvasTumorNormalEnrichmentInput input, IFileLocation bAlleleVcf, bool isDbSnpVcf, IFileLocation manifest, IDirectoryLocation sampleSandbox)
{
if (!_annotationFileProvider.IsSupported(input.GenomeMetadata))
{
_logger.Info($"Skipping Canvas for sample {sampleId}: unsupported reference genome '{input.GenomeMetadata.Name}'");
return(null);
}
StringBuilder commandLine = new StringBuilder("Tumor-normal-enrichment");
commandLine.Append(_singleSampleInputCommandLineBuilder.GetSingleSampleCommandLine(sampleId, input.TumorBam, input.GenomeMetadata, sampleSandbox));
commandLine.Append($" --normal-bam \"{input.NormalBam.BamFile}\"");
commandLine.Append($" --manifest \"{manifest}\"");
var bAlleleVcfOptionName = isDbSnpVcf ?
SingleSampleCommonOptionsParser.PopulationBAlleleVcfOptionName :
SingleSampleCommonOptionsParser.SampleBAlleleVcfOptionName;
commandLine.Append($" --{bAlleleVcfOptionName} {bAlleleVcf.WrapWithShellQuote()}");
commandLine.Append($" --somatic-vcf \"{input.SomaticVcf.VcfFile}\"");
IFileLocation ploidyBed = _canvasPloidyBedCreator.CreatePloidyBed(input.NormalVcf, input.GenomeMetadata, sampleSandbox);
if (ploidyBed != null)
{
commandLine.Append($" --{SingleSampleCommonOptionsParser.PloidyBedOptionName} \"{ploidyBed}\"");
}
commandLine.Append(_singleSampleInputCommandLineBuilder.GetCustomParameters());
commandLine = _singleSampleInputCommandLineBuilder.MergeCustomCanvasParameters(commandLine);
UnitOfWork singleSampleJob = new UnitOfWork
{
ExecutablePath = CrossPlatform.IsThisLinux() ? _runtimeExecutable.FullName : _canvasExe.FullName,
CommandLine = CrossPlatform.IsThisLinux() ? _canvasExe + " " + commandLine : commandLine.ToString(),
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = "Canvas_" + sampleId,
};
_workManager.DoWorkSingleThread(singleSampleJob);
return(GetCanvasOutput(sampleId, sampleSandbox));
}
public CanvasSmallPedigreeOutput Run(CanvasSmallPedigreeInput input, IDirectoryLocation sampleSandbox)
{
if (!_annotationFileProvider.IsSupported(input.GenomeMetadata))
{
_logger.Info($"Skipping Canvas: unsupported reference genome '{input.GenomeMetadata.Name}'");
return(null);
}
if (!_annotationFileProvider.CustomDbSnpVcf(input.GenomeMetadata) && input.Vcf == null)
{
_logger.Info($"Skipping Canvas. A dbSNP VCF file was not provided and no small variant VCF file is available");
return(null);
}
var commandLine = new StringBuilder("SmallPedigree-WGS");
commandLine.Append(GetMultiSampleCommandLine(input.Samples, input.GenomeMetadata, input.Vcf, sampleSandbox));
// use sample vcf by default (performance could be similar with dbSNP vcf though)
var bAlleleVcf = input.Vcf.VcfFile;
var bAlleleVcfOptionName = SingleSampleCommonOptionsParser.SampleBAlleleVcfOptionName;
if (_annotationFileProvider.CustomDbSnpVcf(input.GenomeMetadata))
{
bAlleleVcf = _annotationFileProvider.GetDbSnpVcf(input.GenomeMetadata);
bAlleleVcfOptionName = SingleSampleCommonOptionsParser.PopulationBAlleleVcfOptionName;
}
commandLine.Append($" --{bAlleleVcfOptionName} \"{bAlleleVcf}\"");
var ploidyInfos = input.Samples.SelectData(sample => sample.PloidyInfo);
var ploidyVcf = _canvasPloidyVcfCreator.CreatePloidyVcf(ploidyInfos, input.GenomeMetadata, sampleSandbox);
if (ploidyVcf != null)
{
commandLine.Append($" --{SmallPedigreeOptionsParser.PloidyVcfOptionName} \"{ploidyVcf.VcfFile}\"");
}
var canvasPartitionParam = $@"--commoncnvs {_annotationFileProvider.GetCanvasAnnotationFile(input.GenomeMetadata, "commoncnvs.bed")}";
var moreCustomParameters = new Dictionary <string, string> {
["CanvasPartition"] = canvasPartitionParam
};
commandLine.Append(_singleSampleInputCommandLineBuilder.GetCustomParameters(moreCustomParameters));
commandLine = _singleSampleInputCommandLineBuilder.MergeCustomCanvasParameters(commandLine);
// use Proband or, when proband is not available, first sample as pedigree id
var pedigreeId = input.Samples.Where(x => x.Value.SampleType == SampleType.Proband).Select(x => x.Key.Id).FirstOrDefault();
if (pedigreeId.IsNullOrEmpty())
{
pedigreeId = input.Samples.First().Key.Id;
}
var singleSampleJob = new UnitOfWork()
{
ExecutablePath = CrossPlatform.IsThisLinux() ? _runtimeExecutable.FullName : _canvasExe.FullName,
CommandLine = CrossPlatform.IsThisLinux() ? _canvasExe + " " + commandLine : commandLine.ToString(),
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = "Canvas_" + pedigreeId,
};
_workManager.DoWorkSingleThread(singleSampleJob);
var sampleBams = input.Samples.SelectData(sample => sample.Bam);
return(GetCanvasOutput(sampleBams, sampleSandbox));
}
private CanvasEnrichmentOutput RunSingleSample(string sampleId, CanvasEnrichmentInput input, IFileLocation dbSnpVcf, IFileLocation manifest, IDirectoryLocation sampleSandbox)
{
if (!_annotationFileProvider.IsSupported(input.GenomeMetadata))
{
_logger.Info($"Skipping Canvas for sample {sampleId}: unsupported reference genome '{input.GenomeMetadata.Name}'");
return(null);
}
var moreCustomParameters = new Dictionary <string, string>();
StringBuilder commandLine = new StringBuilder("Somatic-Enrichment");
commandLine.Append(_singleSampleInputCommandLineBuilder.GetSingleSampleCommandLine(sampleId, input.Bam, input.GenomeMetadata, sampleSandbox));
string sexChromosomeKaryotype = null;
if (input.PloidyInfo.IsPloidyAvailable)
{
sexChromosomeKaryotype = input.PloidyInfo.IsXYMale.Value ? PloidyCorrector.PrettyPrintPloidy(1, 1)
: (input.PloidyInfo.IsXXFemale.Value ? PloidyCorrector.PrettyPrintPloidy(2, 0) : null);
}
string controlSexChromosomeKaryotype = null;
if (input.IsCanvasNormalizePcaMode && input.PloidyInfo.IsPloidyAvailable)
{
IFileLocation pcaModelFile = input.PloidyInfo.IsXYMale.Value ? input.PcaModels.MaleModelFile
: (input.PloidyInfo.IsXXFemale.Value ? input.PcaModels.FemaleModelFile : null);
if (pcaModelFile == null)
{
string sampleSex = input.PloidyInfo.IsXYMale.Value ? "male"
: (input.PloidyInfo.IsXXFemale.Value ? "female" : "sex unknown");
_logger.Info($"Skipping Canvas for sample {sampleId}: PCA model file not available for {sampleSex} samples.");
return(null);
}
moreCustomParameters["CanvasNormalize"] = "-m PCA";
commandLine.Append($" --control-binned \"{pcaModelFile}\"");
commandLine.Append($" --control-bin-size 100"); // use a dummy bin size for now
controlSexChromosomeKaryotype = sexChromosomeKaryotype;
}
else if (input.PrecomputedControl?.BinnedPath != null)
{
commandLine.Append($" --control-binned \"{input.PrecomputedControl.BinnedPath}\"");
commandLine.Append($" --control-bin-size {input.PrecomputedControl.BinSize}");
controlSexChromosomeKaryotype = input.PrecomputedControl.SexChromosomeKaryotype;
}
else
{
foreach (var normalBam in input.NormalBamPaths)
{
commandLine.Append($" --control-bam \"{normalBam.BamFile}\"");
}
}
if (controlSexChromosomeKaryotype != null)
{
var controlPloidyBed = sampleSandbox.GetFileLocation("control-ploidy.bed.gz");
if (_canvasPloidyBedCreator.GeneratePloidyBedFileFromSexChromosomeKaryotype(input.GenomeMetadata, input.GenomeMetadata.Sequences.First().FastaPath,
controlSexChromosomeKaryotype, controlPloidyBed.FullName, sampleSandbox.FullName))
{
commandLine.Append($" --{SomaticEnrichmentOptionsParser.ControlPloidyBedOptionName} \"{controlPloidyBed}\"");
}
}
commandLine.Append($" --{SingleSampleCommonOptionsParser.PopulationBAlleleVcfOptionName} \"{dbSnpVcf}\"");
commandLine.Append($" --manifest \"{manifest}\"");
if (sexChromosomeKaryotype != null)
{
var ploidyBed = sampleSandbox.GetFileLocation("ploidy.bed.gz");
if (_canvasPloidyBedCreator.GeneratePloidyBedFileFromSexChromosomeKaryotype(input.GenomeMetadata, input.GenomeMetadata.Sequences.First().FastaPath,
sexChromosomeKaryotype, ploidyBed.FullName, sampleSandbox.FullName))
{
commandLine.Append($" --{SingleSampleCommonOptionsParser.PloidyBedOptionName} \"{ploidyBed}\"");
}
}
if (input.PredefinedBinsFile != null)
{
moreCustomParameters["CanvasBin"] = $"-n={input.PredefinedBinsFile.WrapWithEscapedShellQuote()}";
}
commandLine.Append(_singleSampleInputCommandLineBuilder.GetCustomParameters(moreCustomParameters));
commandLine = _singleSampleInputCommandLineBuilder.MergeCustomCanvasParameters(commandLine);
UnitOfWork singleSampleJob = new UnitOfWork()
{
ExecutablePath = CrossPlatform.IsThisLinux() ? _runtimeExecutable.FullName : _canvasExe.FullName,
CommandLine = CrossPlatform.IsThisLinux() ? _canvasExe + " " + commandLine : commandLine.ToString(),
LoggingFolder = _workManager.LoggingFolder.FullName,
LoggingStub = "Canvas_" + sampleId,
};
_workManager.DoWorkSingleThread(singleSampleJob);
return(GetCanvasOutput(sampleId, sampleSandbox));
}
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);
}