private void MoveIntermediateOutput(SampleInfo info, IntermediateOutput output, IFileMover fileMover)
{
var stub = GetSingleSampleOutputStub(info);
// Output:
fileMover.Move(output.CnvVcf.VcfFile, SingleSampleCallset.GetVcfOutput(stub));
// Files for visualization:
fileMover.Move(output.CoverageBigwig, SingleSampleCallset.GetCoverageBigWig(stub));
var targetBAlleleBedgraph = SingleSampleCallset.GetBAlleleBedGraph(stub);
fileMover.Move(output.BAlleleBedgraph.FileLocation, targetBAlleleBedgraph.FileLocation);
fileMover.Move(output.BAlleleBedgraph.TabixIndex, targetBAlleleBedgraph.TabixIndex);
var targetCopyNumbedBedgraph = SingleSampleCallset.GetCopyNumberBedGraph(stub);
fileMover.Move(output.CopyNumberBedgraph.FileLocation, targetCopyNumbedBedgraph.FileLocation);
fileMover.Move(output.CopyNumberBedgraph.TabixIndex, targetCopyNumbedBedgraph.TabixIndex);
// Deprecated files:
#pragma warning disable CS0618 // Type or member is obsolete
fileMover.Move(output.CoverageAndVariantFrequencies, SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(stub)); // Used for (non-dynamic) plotting
fileMover.Move(output.Partitioned, SingleSampleCallset.GetPartitionedPath(stub)); // used by BSVI
fileMover.Move(output.VariantFrequencies, SingleSampleCallset.GetVfSummaryPath(stub)); // used by BSVI
fileMover.Move(output.VariantFrequenciesBaf, SingleSampleCallset.GetVfSummaryBafPath(stub)); // used by BSVI
#pragma warning restore CS0618 // Type or member is obsolete
}
private void MoveIntermediateOutput(SampleInfo info, IntermediateOutput output, IFileMover fileMover)
{
var stub = GetStub(info.Id);
fileMover.Move(output.CoverageAndVariantFrequencies, SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(stub));
if (_canvasWorkerFactory.IncludeIntermediateResults())
{
fileMover.Move(output.Partitioned, SingleSampleCallset.GetPartitionedPath(stub));
fileMover.Move(output.VariantFrequencies, SingleSampleCallset.GetVfSummaryPath(stub));
fileMover.Move(output.VariantFrequenciesBaf, SingleSampleCallset.GetVfSummaryPath(stub));
}
}
private CanvasSmallPedigreeOutput GetCanvasOutput(SampleSet <Bam> pedigreeBams, IDirectoryLocation sampleSandbox)
{
var readGroupSamples = pedigreeBams.SelectData(GetReadGroupSample);
var intermediateResults = readGroupSamples.SelectData(readGroupSample =>
{
var variantFrequencies = SingleSampleCallset.GetVfSummaryPath(sampleSandbox, readGroupSample);
var variantFrequenciesBaf = SingleSampleCallset.GetVfSummaryBafPath(sampleSandbox, readGroupSample);
var partitioned = SingleSampleCallset.GetPartitionedPath(sampleSandbox, readGroupSample);
var coverageAndVariantFrequencies = SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(sampleSandbox, readGroupSample);
return(new IntermediateOutput(coverageAndVariantFrequencies, variantFrequencies, variantFrequenciesBaf, partitioned));
});
var cnvVcf = new Vcf(sampleSandbox.GetFileLocation("CNV.vcf.gz"));
return(new CanvasSmallPedigreeOutput(cnvVcf, intermediateResults));
}
private CanvasSmallPedigreeOutput Load(CanvasSmallPedigreeInput input)
{
var intermediateOutputs = input.Samples.SelectData((info, sample) =>
{
var stub = GetStub(info.Id);
var coverageAndVariantFrequency = SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(stub);
if (!_canvasWorkerFactory.IncludeIntermediateResults())
{
return(new IntermediateOutput(coverageAndVariantFrequency, null, null, null));
}
var partitioned = SingleSampleCallset.GetPartitionedPath(stub);
var variantFrequencies = SingleSampleCallset.GetVfSummaryPath(stub);
var variantFrequenciesBaf = SingleSampleCallset.GetVfSummaryPath(stub);
return(new IntermediateOutput(coverageAndVariantFrequency, variantFrequencies, variantFrequenciesBaf, partitioned));
});
return(new CanvasSmallPedigreeOutput(new Vcf(GetCnvVcf()), intermediateOutputs));
}
private CanvasSmallPedigreeOutput GetCanvasOutput(SampleSet <CanvasPedigreeSample> pedigreeSamples, IDirectoryLocation sampleSandbox)
{
var intermediateResults = pedigreeSamples.SelectSamples(sampleInfo =>
{
var sampleId = sampleInfo.Id;
var variantFrequencies = SingleSampleCallset.GetVfSummaryPath(sampleSandbox, sampleId);
var variantFrequenciesBaf = SingleSampleCallset.GetVfSummaryBafPath(sampleSandbox, sampleId);
var partitioned = SingleSampleCallset.GetPartitionedPath(sampleSandbox, sampleId);
var coverageAndVariantFrequencies = SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(sampleSandbox, sampleId);
var singleSampleVcf = SingleSampleCallset.GetVcfOutput(sampleSandbox, sampleId);
var coverageBigwig = SingleSampleCallset.GetCoverageBigWig(sampleSandbox, sampleId);
var bAlleleBedgraph = SingleSampleCallset.GetBAlleleBedGraph(sampleSandbox, sampleId);
var copyNumberBedgraph = SingleSampleCallset.GetCopyNumberBedGraph(sampleSandbox, sampleId);
return(new IntermediateOutput(new Vcf(singleSampleVcf), coverageAndVariantFrequencies, variantFrequencies, variantFrequenciesBaf,
partitioned, coverageBigwig, bAlleleBedgraph, copyNumberBedgraph));
});
var cnvVcf = new Vcf(sampleSandbox.GetFileLocation("CNV.vcf.gz"));
return(new CanvasSmallPedigreeOutput(cnvVcf, intermediateResults));
}
private CanvasSmallPedigreeOutput Load(CanvasSmallPedigreeInput input)
{
var intermediateOutputs = input.Samples.SelectData((info, sample) =>
{
var stub = GetSingleSampleOutputStub(info);
var coverageAndVariantFrequency = SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(stub);
var singleSampleVcf = new Vcf(SingleSampleCallset.GetVcfOutput(stub));
var partitioned = SingleSampleCallset.GetPartitionedPath(stub);
var variantFrequencies = SingleSampleCallset.GetVfSummaryPath(stub);
var variantFrequenciesBaf = SingleSampleCallset.GetVfSummaryBafPath(stub);
var coverageBigwig = SingleSampleCallset.GetCoverageBigWig(stub);
var bAlleleBedgraph = SingleSampleCallset.GetBAlleleBedGraph(stub);
var copyNumberBedgraph = SingleSampleCallset.GetCopyNumberBedGraph(stub);
return(new IntermediateOutput(singleSampleVcf, coverageAndVariantFrequency, variantFrequencies, variantFrequenciesBaf, partitioned,
coverageBigwig, bAlleleBedgraph, copyNumberBedgraph));
});
return(new CanvasSmallPedigreeOutput(new Vcf(GetPedigreeVcf()), intermediateOutputs));
}
internal int CallVariants(List <string> variantFrequencyFiles, List <string> segmentFiles,
IFileLocation outVcfFile, string ploidyBedPath, string referenceFolder, List <string> sampleNames, string commonCnvsBedPath, List <SampleType> sampleTypes)
{
// load files
// initialize data structures and classes
var fileCounter = 0;
var samplesInfo = new SampleMap <SampleMetrics>();
var sampleSegments = new SampleMap <Segments>();
var copyNumberModels = new SampleMap <ICopyNumberModel>();
var variantFrequencyFilesSampleList = new SampleMap <string>();
var kinships = new SampleMap <SampleType>();
foreach (string sampleName in sampleNames)
{
var sampleId = new SampleId(sampleName);
var segment = Segments.ReadSegments(_logger, new FileLocation(segmentFiles[fileCounter]));
segment.AddAlleles(CanvasIO.ReadFrequenciesWrapper(_logger, new FileLocation(variantFrequencyFiles[fileCounter]), segment.IntervalsByChromosome));
sampleSegments.Add(sampleId, segment);
var sampleInfo = SampleMetrics.GetSampleInfo(segment.AllSegments, ploidyBedPath, _callerParameters.NumberOfTrimmedBins, sampleId);
var copyNumberModel = _copyNumberModelFactory.CreateModel(_callerParameters.MaximumCopyNumber, sampleInfo.MaxCoverage, sampleInfo.MeanCoverage, sampleInfo.MeanMafCoverage);
samplesInfo.Add(sampleId, sampleInfo);
copyNumberModels.Add(sampleId, copyNumberModel);
variantFrequencyFilesSampleList.Add(sampleId, variantFrequencyFiles[fileCounter]);
kinships.Add(sampleId, sampleTypes[fileCounter]);
fileCounter++;
}
var segmentSetsFromCommonCnvs = CreateSegmentSetsFromCommonCnvs(variantFrequencyFilesSampleList,
_callerParameters.MinAlleleCountsThreshold, commonCnvsBedPath, sampleSegments);
var segmentsForVariantCalling = GetHighestLikelihoodSegments(segmentSetsFromCommonCnvs, samplesInfo, copyNumberModels).ToList();
PedigreeInfo pedigreeInfo = PedigreeInfo.GetPedigreeInfo(kinships, _callerParameters);
Parallel.ForEach(
segmentsForVariantCalling,
new ParallelOptions
{
MaxDegreeOfParallelism = Math.Min(Environment.ProcessorCount, _callerParameters.MaxCoreNumber)
},
segments => _variantCaller.CallVariant(segments, samplesInfo, copyNumberModels, pedigreeInfo)
);
var variantCalledSegments = new SampleMap <List <CanvasSegment> >();
foreach (var key in samplesInfo.SampleIds)
{
variantCalledSegments.Add(key, segmentsForVariantCalling.Select(segment => segment[key]).ToList());
}
var mergedVariantCalledSegments = MergeSegments(variantCalledSegments, _callerParameters.MinimumCallSize, _qualityFilterThreshold);
FilterExcessivelyShortSegments(mergedVariantCalledSegments);
var outputFolder = outVcfFile.Directory;
foreach (var sampleId in samplesInfo.SampleIds)
{
var coverageOutputPath = SingleSampleCallset.GetCoverageAndVariantFrequencyOutput(outputFolder,
sampleId.ToString());
CanvasSegment.WriteCoveragePlotData(mergedVariantCalledSegments[sampleId], samplesInfo[sampleId].MeanCoverage,
samplesInfo[sampleId].Ploidy, coverageOutputPath, referenceFolder);
}
bool isPedigreeInfoSupplied = pedigreeInfo != null && pedigreeInfo.HasFullPedigree();
var denovoQualityThreshold = isPedigreeInfoSupplied ? (int?)_deNovoQualityFilterThreshold : null;
var ploidies = samplesInfo.Select(info => info.Value.Ploidy).ToList();
var diploidCoverage = samplesInfo.Select(info => info.Value.MeanCoverage).ToList();
var names = samplesInfo.SampleIds.Select(id => id.ToString()).ToList();
CanvasSegmentWriter.WriteMultiSampleSegments(outVcfFile.FullName, mergedVariantCalledSegments, diploidCoverage, referenceFolder, names,
null, ploidies, _qualityFilterThreshold, denovoQualityThreshold, CanvasFilter.SegmentSizeCutoff, isPedigreeInfoSupplied);
foreach (var sampleId in samplesInfo.SampleIds)
{
var outputVcfPath = SingleSampleCallset.GetVcfOutput(outputFolder, sampleId.ToString());
var sampleMetrics = samplesInfo[sampleId];
var segments = mergedVariantCalledSegments[sampleId];
CanvasSegmentWriter.WriteSegments(outputVcfPath.FullName, segments,
sampleMetrics.MeanCoverage, referenceFolder, sampleId.ToString(), null,
sampleMetrics.Ploidy, _qualityFilterThreshold, isPedigreeInfoSupplied, denovoQualityThreshold, null);
var visualizationTemp = outputFolder.CreateSubdirectory($"VisualizationTemp{sampleId}");
var normalizationFactor = NormalizationCalculator.ComputeNormalizationFactor(segments);
var bigWig = _coverageBigWigWriter.Write(segments, visualizationTemp, normalizationFactor);
bigWig?.MoveTo(SingleSampleCallset.GetCoverageBigWig(outputFolder, sampleId.ToString()));
var copyNumberBedGraph = SingleSampleCallset.GetCopyNumberBedGraph(outputFolder, sampleId.ToString());
_copyNumberBedGraphWriter.Write(segments, sampleMetrics.Ploidy, copyNumberBedGraph);
var partitionBedgraphHeader = "track type=bedGraph visibility=full autoScale=on graphType=points";
var originalSegments = sampleSegments[sampleId];
_partitionCoverageBedGraphWriter.Write(originalSegments.AllSegments, SingleSampleCallset.GetPartitionBedGraph(outputFolder, sampleId.ToString()), normalizationFactor, partitionBedgraphHeader);
}
return(0);
}