private static void PostProcessAndWriteResults(SegmentationInput segmentationInput, string outPartitionedFile,
PloidyInfo referencePloidy, GenomeSegmentationResults segmentationResults)
{
var segments = segmentationInput.PostProcessSegments(segmentationResults, referencePloidy);
segmentationInput.WriteCanvasPartitionResults(outPartitionedFile, segments);
}
public void NoSegments_ReturnsNoBedGraphEntries()
{
var calculator = new CopyNumberBedGraphCalculator();
var segments = Enumerable.Empty <CanvasSegment>().ToList();
var ploidyInfo = new PloidyInfo();
var results = calculator.Calculate(segments, ploidyInfo);
Assert.Empty(results);
}
private SampleMetrics(double meanCoverage, double meanMafCoverage, double variance,
double mafVariance, int maxCoverage, PloidyInfo ploidy)
{
MeanCoverage = meanCoverage;
MeanMafCoverage = meanMafCoverage;
Variance = variance;
MafVariance = mafVariance;
MaxCoverage = maxCoverage;
Ploidy = ploidy;
}
public static int GetPloidy(PloidyInfo referencePloidy, string chrom, int start, int end, int defaultPloidy = 2)
{
if (referencePloidy == null)
{
return(defaultPloidy);
}
CanvasSegment segment = new CanvasSegment(chrom, start, end, new List <float>());
return(referencePloidy.GetReferenceCopyNumber(segment));
}
public static void RatiosToCounts(IEnumerable <SampleGenomicBin> ratios, IFileLocation referencePloidyBedFile,
IFileLocation outputPath)
{
PloidyInfo referencePloidy = null;
if (referencePloidyBedFile != null && referencePloidyBedFile.Exists)
{
referencePloidy = PloidyInfo.LoadPloidyFromBedFile(referencePloidyBedFile.FullName);
}
CanvasIO.WriteToTextFile(outputPath.FullName, RatiosToCounts(ratios, referencePloidy));
}
public Dictionary <string, List <SegmentWithBins> > PostProcessSegments(
GenomeSegmentationResults segmentationResults,
PloidyInfo referencePloidy)
{
var excludedIntervals = new Dictionary <string, List <SampleGenomicBin> >();
if (!string.IsNullOrEmpty(ForbiddenIntervalBedPath))
{
excludedIntervals = CanvasCommon.Utilities.LoadBedFile(ForbiddenIntervalBedPath);
}
return(_processor.PostProcessSegments(segmentationResults, referencePloidy, excludedIntervals,
CoverageInfo));
}
public static SampleMetrics GetSampleInfo(IReadOnlyList <CanvasSegment> segments, string ploidyBedPath, int numberOfTrimmedBins, SampleId id)
{
double meanMafCoverage = new SortedList <int>(segments.SelectMany(x => x.Balleles.TotalCoverage)).Median();
double variance = Utilities.Variance(segments.Select(x => x.TruncatedMedianCount(numberOfTrimmedBins)).ToList());
double mafVariance = Utilities.Variance(segments.Where(x => x.Balleles.TotalCoverage.Count > 0)
.Select(x => x.Balleles.TotalCoverage.Average()).ToList());
double meanCoverage = new SortedList <float>(segments.SelectMany(x => x.Counts).Select(x => x)).Median();
int maxCoverage = Convert.ToInt16(segments.Select(x => x.TruncatedMedianCount(numberOfTrimmedBins)).Max()) + 10;
var ploidy = new PloidyInfo();
if (!ploidyBedPath.IsNullOrEmpty() && File.Exists(ploidyBedPath))
{
ploidy = PloidyInfo.LoadPloidyFromVcfFile(ploidyBedPath, id.ToString());
}
return(new SampleMetrics(meanCoverage, meanMafCoverage, variance, mafVariance, maxCoverage, ploidy));
}
private bool IsNewSegment(Dictionary <string, bool> starts, string chr, List <SampleGenomicBin> excludeIntervals, uint previousBinEnd,
uint end, uint start, ref int excludeIndex, PloidyInfo referencePloidy)
{
string key = chr + ":" + start;
bool newSegment = starts.ContainsKey(key);
if (excludeIntervals != null)
{
while (excludeIndex < excludeIntervals.Count && excludeIntervals[excludeIndex].Stop < previousBinEnd)
{
excludeIndex++;
}
if (excludeIndex < excludeIntervals.Count)
{
// Note: forbiddenZoneMid should never fall inside a bin, becuase these intervals were already excluded
// from consideration during the call to CanvasBin.
var excludeStart = excludeIntervals[excludeIndex].Start;
var excludeStop = excludeIntervals[excludeIndex].Stop;
int forbiddenZoneMid = (excludeStart + excludeStop) / 2;
if (previousBinEnd < forbiddenZoneMid && end >= forbiddenZoneMid)
{
//Debug.Assert(previousBinEnd <= excludeStart);
//Debug.Assert(start >= excludeStop);
newSegment = true;
}
}
}
if (previousBinEnd > 0 && _maxInterBinDistInSegment >= 0 && previousBinEnd + _maxInterBinDistInSegment < start &&
!newSegment)
{
newSegment = true;
}
// also start new segment if reference ploidy changes between end of last and end of this;
// note that Interval takes 1-based positions, so using "previousBinEnd" effectively
// includes the last base of the previous bin, allowing for a change at the bin boundary
if (!newSegment && referencePloidy != null)
{
var refIval = new ReferenceInterval(chr, new Interval(previousBinEnd > 0 ? previousBinEnd : 1, end));
if (!referencePloidy.IsUniformReferencePloidy(refIval))
{
newSegment = true;
}
}
return(newSegment);
}
private bool IsPassVariant(CanvasSegment segment, PloidyInfo ploidyInfo)
{
if (!segment.Filter.IsPass)
{
return(false);
}
var referenceCopyNumber = ploidyInfo?.GetReferenceCopyNumber(segment) ?? 2;
if (segment.CopyNumber != referenceCopyNumber)
{
return(true);
}
if (segment.CopyNumber == 2 && segment.MajorChromosomeCount == 2)
{
return(true); //LOH
}
return(false);
}
private static void GetBinRatio(string tumorBinnedPath, string normalBinnedPath, string ratioBinnedPath,
string ploidyBedPath, NexteraManifest manifest = null)
{
PloidyInfo referencePloidy = String.IsNullOrEmpty(ploidyBedPath) ? null : PloidyInfo.LoadPloidyFromBedFile(ploidyBedPath);
double tumorMedian = (new BinCounts(tumorBinnedPath, manifest: manifest)).OnTargetMedianBinCount;
double normalMedian = (new BinCounts(normalBinnedPath, manifest: manifest)).OnTargetMedianBinCount;
double librarySizeFactor = (tumorMedian > 0 && normalMedian > 0) ? normalMedian / tumorMedian : 1;
using (GzipReader tumorReader = new GzipReader(tumorBinnedPath))
using (GzipReader normalReader = new GzipReader(normalBinnedPath))
using (GzipWriter writer = new GzipWriter(ratioBinnedPath))
{
string normalLine;
string tumorLine;
string[] normalToks;
string[] tumorToks;
double normalCount;
double tumorCount;
double ratio;
while ((normalLine = normalReader.ReadLine()) != null)
{
tumorLine = tumorReader.ReadLine();
normalToks = normalLine.Split('\t');
tumorToks = tumorLine.Split('\t');
normalCount = double.Parse(normalToks[3]);
tumorCount = double.Parse(tumorToks[3]);
// The weighted average count of a bin could be less than 1.
// Using these small counts for coverage normalization creates large ratios.
// It would be better to just drop these bins so we don't introduce too much noise into segmentation and CNV calling.
if (normalCount < 1)
{
continue;
} // skip the bin
string chrom = normalToks[0];
int start = int.Parse(normalToks[1]);
int end = int.Parse(normalToks[2]);
// get the normal ploidy from intervalsWithPloidyByChrom
double factor = CanvasDiploidBinRatioFactor * GetPloidy(referencePloidy, chrom, start, end) / 2.0;
ratio = tumorCount / normalCount * factor * librarySizeFactor;
normalToks[3] = String.Format("{0}", ratio);
writer.WriteLine(String.Join("\t", normalToks));
}
}
}
public void ReferenceCopyNumber_IsExcluded()
{
var calculator = new CopyNumberBedGraphCalculator();
var segments = new List <CanvasSegment>
{
new CanvasSegment("chr1", 0, 1, new List <SampleGenomicBin>
{
new SampleGenomicBin("chr1", 0, 1, 3f)
})
{
CopyNumber = 2,
Filter = CanvasFilter.PassFilter
}
};
var ploidyInfo = new PloidyInfo();
var results = calculator.Calculate(segments, ploidyInfo).ToList();
Assert.Empty(results);
}
public void FiltersNonPassSegments()
{
var calculator = new CopyNumberBedGraphCalculator();
var segments = new List <CanvasSegment>
{
new CanvasSegment("chr1", 0, 1, new List <SampleGenomicBin>
{
new SampleGenomicBin("chr1", 0, 1, 3f)
})
{
CopyNumber = 0,
Filter = CanvasFilter.Create("NonPass".Yield())
}
};
var ploidyInfo = new PloidyInfo();
var results = calculator.Calculate(segments, ploidyInfo);
Assert.Empty(results);
}
public void LOH_IsIncluded()
{
var calculator = new CopyNumberBedGraphCalculator();
var segments = new List <CanvasSegment>
{
new CanvasSegment("chr1", 0, 1, new List <SampleGenomicBin>
{
new SampleGenomicBin("chr1", 0, 1, 3f)
})
{
CopyNumber = 2,
MajorChromosomeCount = 2,
Filter = CanvasFilter.PassFilter
}
};
var ploidyInfo = new PloidyInfo();
var results = calculator.Calculate(segments, ploidyInfo).ToList();
Assert.Equal(2m, results.First().Value);
}
public void VariantCopyNumber_ReturnsCopyNumber()
{
var calculator = new CopyNumberBedGraphCalculator();
var segments = new List <CanvasSegment>
{
new CanvasSegment("chr1", 0, 1, new List <SampleGenomicBin>
{
new SampleGenomicBin("chr1", 0, 1, 3f)
})
{
CopyNumber = 1,
Filter = CanvasFilter.PassFilter
}
};
var ploidyInfo = new PloidyInfo();
var results = calculator.Calculate(segments, ploidyInfo).ToList();
Assert.Equal("chr1", results.First().Chromosome);
Assert.Equal(new BedInterval(0, 1), results.First().Interval);
Assert.Equal(1m, results.First().Value);
}
private static IEnumerable <SampleGenomicBin> RatiosToCounts(IEnumerable <SampleGenomicBin> ratios, PloidyInfo referencePloidy)
{
foreach (SampleGenomicBin ratio in ratios)
{
// get the normal ploidy
double factor = CanvasDiploidBinRatioFactor * GetPloidy(referencePloidy, ratio.GenomicBin.Chromosome, ratio.Start, ratio.Stop) / 2.0;
double count = ratio.Count * factor;
yield return(new SampleGenomicBin(ratio.GenomicBin.Chromosome, ratio.Start, ratio.Stop, ratio.GenomicBin.GC, (float)count));
}
}
static int Main(string[] args)
{
CanvasCommon.Utilities.LogCommandLine(args);
List <string> cleanedFiles = new List <string>();
List <string> outPartitionedFiles = new List <string>();
List <string> vafFiles = new List <string>();
bool needHelp = false;
bool isGermline = false;
string filterBedFile = null;
string referenceFolder = null;
string commonCNVsbedPath = null;
string evennessMetricFile = null;
SegmentSplitUndo undoMethod = SegmentSplitUndo.None;
SegmentationInput.SegmentationMethod partitionMethod = SegmentationInput.SegmentationMethod.Wavelets;
string parameterconfigPath = Path.Combine(Isas.Framework.Utilities.Utilities.GetAssemblyFolder(typeof(CanvasPartition)), "CanvasPartitionParameters.json");
string ploidyVcfPath = null;
OptionSet p = new OptionSet()
{
{ "i|infile=", "input file - usually generated by CanvasClean", v => cleanedFiles.Add(v) },
{ "v|vaffile=", "variant frequencyfiles - usually generated by CanvasSNV", v => vafFiles.Add(v) },
{ "o|outfile=", "text file to output", v => outPartitionedFiles.Add(v) },
{ "m|method=", "segmentation method (Wavelets/CBS). Default: " + partitionMethod, v => partitionMethod = (SegmentationInput.SegmentationMethod)Enum.Parse(typeof(SegmentationInput.SegmentationMethod), v) },
{ "r|reference=", "folder that contains both genome.fa and GenomeSize.xml", v => referenceFolder = v },
{ "s|split=", "CBS split method (None/Prune/SDUndo). Default: " + undoMethod, v => undoMethod = (SegmentSplitUndo)Enum.Parse(typeof(SegmentSplitUndo), v) },
{ "b|bedfile=", "bed file to exclude (don't span these intervals)", v => filterBedFile = v },
{ "c|commoncnvs=", "bed file with common CNVs (always include these intervals into segmentation results)", v => commonCNVsbedPath = v },
{ "g|germline", "flag indicating that input file represents germline genome", v => isGermline = v != null },
{ $"{CommandLineOptions.EvennessMetricFile}=", "output file for evenness metric (optional)", v => evennessMetricFile = v },
{ "p|ploidyVcfFile=", "vcf file specifying reference ploidy (e.g. for sex chromosomes) (optional)", v => ploidyVcfPath = v },
{ "config=", "parameter configuration path (default {parameterconfigPath})", v => parameterconfigPath = v },
{ "h|help", "show this message and exit", v => needHelp = v != null }
};
List <string> extraArgs = p.Parse(args);
if (extraArgs.Any())
{
throw new IlluminaException($"Unknown arguments: {string.Join(",", extraArgs)}");
}
if (needHelp)
{
ShowHelp(p);
return(0);
}
if (!cleanedFiles.Any() || !outPartitionedFiles.Any() || referenceFolder == null)
{
ShowHelp(p);
return(0);
}
if (cleanedFiles.Any(inFile => !File.Exists(inFile)))
{
Console.WriteLine("CanvasPartition.exe: File {0} does not exist! Exiting.", cleanedFiles);
return(1);
}
if (!string.IsNullOrEmpty(filterBedFile) && !File.Exists(filterBedFile))
{
Console.WriteLine("CanvasPartition.exe: File {0} does not exist! Exiting.", filterBedFile);
return(1);
}
if (!File.Exists(parameterconfigPath))
{
Console.WriteLine($"CanvasPedigreeCaller.exe: File {parameterconfigPath} does not exist! Exiting.");
return(1);
}
if (!string.IsNullOrEmpty(ploidyVcfPath) && !File.Exists(ploidyVcfPath))
{
Console.WriteLine("CanvasPartition.exe: File {0} does not exist! Exiting.", ploidyVcfPath);
return(1);
}
var parameterconfigFile = new FileLocation(parameterconfigPath);
var canvasPartitionParameters = Deserialize <CanvasPartitionParameters>(parameterconfigFile);
ILogger logger = new Logger(Console.Out.ToEnumerable(), Console.Error.ToEnumerable());
var processor = new SegmentationResultsProcessor(canvasPartitionParameters.MaxInterBinDistInSegment);
var segmentationInputs = vafFiles.Count > 0 && vafFiles.Count == cleanedFiles.Count ?
cleanedFiles.Zip(vafFiles, (inFile, vafFile) => new SegmentationInput(inFile, vafFile, filterBedFile,
referenceFolder, evennessMetricFile, logger, processor)).ToList() :
cleanedFiles.Select(inFile => new SegmentationInput(inFile, null, filterBedFile,
referenceFolder, evennessMetricFile, logger, processor)).ToList();
GenomeSegmentationResults segmentationResults;
PloidyInfo referencePloidy = ploidyVcfPath != null?PloidyInfo.LoadPloidyFromVcfFileNoSampleId(ploidyVcfPath) : null;
switch (partitionMethod)
{
default: // use Wavelets if CBS is not selected
Console.WriteLine("{0} Running Wavelet Partitioning", DateTime.Now);
var waveletsRunner = new WaveletsRunner(new WaveletsRunner.WaveletsRunnerParams(isGermline,
commonCNVsbedPath, madFactor:
canvasPartitionParameters.MadFactor,
thresholdLowerMaf: canvasPartitionParameters.ThresholdLowerMaf,
evennessScoreThreshold: canvasPartitionParameters.EvennessScoreThreshold, verbose: 2));
segmentationResults = new GenomeSegmentationResults(waveletsRunner.Run(segmentationInputs.Single(),
canvasPartitionParameters.EvennessScoreWindow));
PostProcessAndWriteResults(segmentationInputs.Single(), outPartitionedFiles.Single(), referencePloidy, segmentationResults);
break;
case SegmentationInput.SegmentationMethod.CBS:
{
Console.WriteLine("{0} Running CBS Partitioning", DateTime.Now);
var cbsRunner = new CBSRunner(canvasPartitionParameters.MaxInterBinDistInSegment, undoMethod,
canvasPartitionParameters.CBSalpha);
var sampleSegmentations = new List <GenomeSegmentationResults>();
foreach (var input in segmentationInputs)
{
var segmentation = new GenomeSegmentationResults(cbsRunner.Run(input, verbose: 2));
sampleSegmentations.Add(segmentation);
}
segmentationResults = GenomeSegmentationResults.SplitOverlappingSegments(sampleSegmentations);
foreach (var(segmentationInput, outPartitionedFile) in segmentationInputs.Zip(outPartitionedFiles))
{
PostProcessAndWriteResults(segmentationInput, outPartitionedFile, referencePloidy, segmentationResults);
}
break;
}
case SegmentationInput.SegmentationMethod.HMM:
{
Console.WriteLine("{0} Running HMM Partitioning", DateTime.Now);
var hiddenMarkovModelsRunner = new HiddenMarkovModelsRunner(cleanedFiles.Count);
bool isPerSample = false;
segmentationResults =
new GenomeSegmentationResults(hiddenMarkovModelsRunner.Run(segmentationInputs, isPerSample));
for (int i = 0; i < segmentationInputs.Count; i++)
{
PostProcessAndWriteResults(segmentationInputs[i], outPartitionedFiles[i], referencePloidy, segmentationResults);
}
break;
}
case SegmentationInput.SegmentationMethod.PerSampleHMM:
{
Console.WriteLine("{0} Running Per-sample HMM Partitioning", DateTime.Now);
var hiddenMarkovModelsRunner = new HiddenMarkovModelsRunner(1);
var sampleSegmentations = new List <GenomeSegmentationResults>();
bool isPerSample = true;
foreach (var input in segmentationInputs)
{
var segmentation =
new GenomeSegmentationResults(
hiddenMarkovModelsRunner.Run(input.Yield().ToList(), isPerSample));
sampleSegmentations.Add(segmentation);
}
segmentationResults = GenomeSegmentationResults.SplitOverlappingSegments(sampleSegmentations);
foreach (var(segmentationInput, outPartitionedFile) in segmentationInputs.Zip(outPartitionedFiles))
{
PostProcessAndWriteResults(segmentationInput, outPartitionedFile, referencePloidy,
segmentationResults);
}
break;
}
}
Console.WriteLine("{0} CanvasPartition results written out", DateTime.Now);
return(0);
}
public IEnumerable <BedGraphEntry> Calculate(IReadOnlyList <CanvasSegment> segments, PloidyInfo ploidyInfo)
{
var variantSegments = segments.Where(segment => IsPassVariant(segment, ploidyInfo));
return(variantSegments.Select(GetCopyNumberEntry));
}
public Dictionary <string, List <SegmentWithBins> > PostProcessSegments(
GenomeSegmentationResults segmentationResults,
PloidyInfo referencePloidy, Dictionary <string, List <SampleGenomicBin> > excludedIntervals, CoverageInfo coverageInfo)
{
var starts = new Dictionary <string, bool>();
var stops = new Dictionary <string, bool>();
foreach (string chr in segmentationResults.SegmentByChr.Keys)
{
for (int segmentIndex = 0; segmentIndex < segmentationResults.SegmentByChr[chr].Length; segmentIndex++)
{
var segment = segmentationResults.SegmentByChr[chr][segmentIndex];
starts[chr + ":" + segment.start] = true;
stops[chr + ":" + segment.end] = true;
}
}
int segmentNum = -1;
var segmentsByChromosome = new Dictionary <string, List <SegmentWithBins> >();
foreach (string chr in coverageInfo.StartByChr.Keys)
{
segmentsByChromosome.Add(chr, new List <SegmentWithBins>());
SegmentWithBins currentSegment = null;
List <SampleGenomicBin> excludeIntervals = null;
if (excludedIntervals.ContainsKey(chr))
{
excludeIntervals = excludedIntervals[chr];
}
var excludeIndex = 0; // Points to the first interval which *doesn't* end before our current position
uint previousBinEnd = 0;
for (int binIndex = 0; binIndex < coverageInfo.StartByChr[chr].Length; binIndex++)
{
uint start = coverageInfo.StartByChr[chr][binIndex];
uint end = coverageInfo.EndByChr[chr][binIndex];
bool newSegment = IsNewSegment(starts, chr, excludeIntervals, previousBinEnd, end, start, ref excludeIndex, referencePloidy);
var bin = new Bin(start, end, coverageInfo.CoverageByChr[chr][binIndex]);
if (newSegment)
{
segmentNum++;
currentSegment = new SegmentWithBins(segmentNum, bin);
segmentsByChromosome[chr].Add(currentSegment);
}
else
{
if (currentSegment == null)
{
currentSegment = new SegmentWithBins(segmentNum, bin);
segmentsByChromosome[chr].Add(currentSegment);
}
else
{
currentSegment.AddBin(bin);
}
}
previousBinEnd = end;
}
}
return(segmentsByChromosome);
}
public void Write(IReadOnlyList <CanvasSegment> segments, PloidyInfo ploidyInfo, BgzfFile location)
{
var entries = _calculator.Calculate(segments, ploidyInfo);
_writer.Write(entries, location);
}
public int CallVariants(string variantFrequencyFile, string inFile, string outFile, string ploidyBedPath, string referenceFolder, string sampleName,
string truthDataPath)
{
if (!string.IsNullOrEmpty(truthDataPath))
{
this.CNOracle = new CopyNumberOracle();
this.CNOracle.LoadKnownCN(truthDataPath);
}
this.Segments = CanvasSegment.ReadSegments(inFile);
this.TempFolder = Path.GetDirectoryName(inFile);
if (this.Segments.Count == 0)
{
Console.WriteLine("CanvasDiploidCaller: No segments loaded; no CNV calls will be made.");
CanvasSegment.WriteSegments(outFile, this.Segments, referenceFolder, sampleName, null, null);
return(0);
}
PloidyInfo ploidy = null;
if (!string.IsNullOrEmpty(ploidyBedPath))
{
ploidy = PloidyInfo.LoadPloidyFromBedFile(ploidyBedPath);
}
// load MAF
this.MeanCoverage = CanvasIO.LoadVariantFrequencies(variantFrequencyFile, this.Segments);
int medianVariantCoverage = AggregateVariantCoverage(ref this.Segments);
// Create new models for different copy number states
this.InitializePloidies();
// Compute statistics on the copy number two regions
float[] diploidCounts = AggregateCounts(ref this.Segments);
DiploidCoverage = CanvasCommon.Utilities.Mean(diploidCounts);
CoverageWeightingFactor = CoverageWeighting / DiploidCoverage;
// new coverage model
this.Model = new CoverageModel();
Model.DiploidCoverage = DiploidCoverage;
List <SegmentInfo> segments = new List <SegmentInfo>();
foreach (CanvasSegment segment in this.Segments)
{
SegmentInfo info = new SegmentInfo();
info.Segment = segment;
List <double> MAF = new List <double>();
foreach (float value in segment.VariantFrequencies)
{
MAF.Add(value > 0.5 ? 1 - value : value);
}
if (MAF.Count > 0)
{
info.MAF = CanvasCommon.Utilities.Median(MAF);
}
else
{
info.MAF = -1;
}
info.Coverage = CanvasCommon.Utilities.Median(segment.Counts);
if (this.Segments.Count > 100)
{
info.Weight = segment.End - segment.Begin;
}
else
{
info.Weight = segment.BinCount;
}
segments.Add(info);
}
// Assign copy number and major chromosome count for each segment
bool useGaussianMixtureModel = false; // For now, this is set false, since we saw weird performance on chrY (CANV-115):
if (useGaussianMixtureModel)
{
// optimize model covariance
double likelihood = FitGaussians(Model, segments);
AssignPloidyCallsGaussianMixture();
}
else
{
AssignPloidyCallsDistance(Model, segments, medianVariantCoverage);
}
// Merge neighboring segments that got the same copy number call.
CanvasSegment.MergeSegments(ref this.Segments);
CanvasSegment.AssignQualityScores(this.Segments, CanvasSegment.QScoreMethod.LogisticGermline);
List <string> extraHeaders = new List <string>();
string coverageOutputPath = CanvasCommon.Utilities.GetCoverageAndVariantFrequencyOutputPath(outFile);
CanvasSegment.WriteCoveragePlotData(this.Segments, Model.DiploidCoverage, ploidy, coverageOutputPath, referenceFolder);
if (this.CNOracle != null)
{
this.GenerateReportVersusKnownCN();
}
if (ploidy != null && !string.IsNullOrEmpty(ploidy.HeaderLine))
{
extraHeaders.Add(ploidy.HeaderLine);
}
CanvasSegment.WriteSegments(outFile, this.Segments, referenceFolder, sampleName, extraHeaders, ploidy);
return(0);
}
public int CallVariants(string variantFrequencyFile, string inFile, string outFile, string ploidyVcfPath, string referenceFolder, string sampleName,
string truthDataPath)
{
if (!string.IsNullOrEmpty(truthDataPath))
{
_cnOracle = new CopyNumberOracle();
_cnOracle.LoadKnownCN(truthDataPath);
}
_segments = Segments.ReadSegments(_logger, new FileLocation(inFile));
_allSegments = _segments.AllSegments.ToList();
TempFolder = Path.GetDirectoryName(inFile);
if (_allSegments.Count == 0)
{
Console.WriteLine("CanvasDiploidCaller: No segments loaded; no CNV calls will be made.");
CanvasSegmentWriter.WriteSegments(outFile, _allSegments, _model?.DiploidCoverage, referenceFolder,
sampleName, null, null, QualityFilterThreshold, false, null, null);
return(0);
}
PloidyInfo ploidy = null;
if (!string.IsNullOrEmpty(ploidyVcfPath))
{
ploidy = PloidyInfo.LoadPloidyFromVcfFileNoSampleId(ploidyVcfPath);
}
// load MAF
var allelesByChromosome = CanvasIO.ReadFrequenciesWrapper(_logger, new FileLocation(variantFrequencyFile), _segments.IntervalsByChromosome);
_segments.AddAlleles(allelesByChromosome);
MeanCoverage = allelesByChromosome.SelectMany(x => x.Value).SelectMany(y => y.TotalCoverage).Average();
AggregateVariantCoverage(ref _allSegments);
// Create new models for different copy number states
InitializePloidies();
// Compute statistics on the copy number two regions
float[] diploidCounts = AggregateCounts(ref _allSegments);
_diploidCoverage = Utilities.Mean(diploidCounts);
_coverageWeightingFactor = CoverageWeighting / _diploidCoverage;
// new coverage model
_model = new CoverageModel {
DiploidCoverage = _diploidCoverage
};
List <SegmentInfo> segments = new List <SegmentInfo>();
foreach (CanvasSegment segment in _allSegments)
{
SegmentInfo info = new SegmentInfo {
Segment = segment
};
List <double> mafs = new List <double>();
foreach (float value in segment.Balleles.Frequencies)
{
mafs.Add(value > 0.5 ? 1 - value : value);
}
if (mafs.Count > 0)
{
info.Maf = Utilities.Median(mafs);
}
else
{
info.Maf = -1;
}
info.Coverage = Utilities.Median(segment.Counts);
info.Weight = _allSegments.Count > 100 ? segment.Length : segment.BinCount;
segments.Add(info);
}
AssignPloidyCallsDistance(_model);
CanvasSegment.AssignQualityScores(_allSegments, CanvasSegment.QScoreMethod.LogisticGermline, _germlineScoreParameters);
// Merge neighboring segments that got the same copy number call.
// merging segments requires quality scores so we do it after quality scores have been assigned
var mergedSegments = CanvasSegment.MergeSegments(_allSegments);
// recalculating qscores after merging segments improves performance!
CanvasSegment.AssignQualityScores(mergedSegments, CanvasSegment.QScoreMethod.LogisticGermline, _germlineScoreParameters);
CanvasSegment.SetFilterForSegments(QualityFilterThreshold, mergedSegments, CanvasFilter.SegmentSizeCutoff);
List <string> extraHeaders = new List <string>();
var coverageOutputPath = SingleSampleCallset.GetCoverageAndVariantFrequencyOutputPath(outFile);
CanvasSegment.WriteCoveragePlotData(mergedSegments, _model.DiploidCoverage, ploidy, coverageOutputPath, referenceFolder);
if (_cnOracle != null)
{
GenerateReportVersusKnownCopyNumber();
}
if (!string.IsNullOrEmpty(ploidy?.HeaderLine))
{
extraHeaders.Add(ploidy.HeaderLine);
}
CanvasSegmentWriter.WriteSegments(outFile, mergedSegments, _model.DiploidCoverage, referenceFolder, sampleName,
extraHeaders, ploidy, QualityFilterThreshold, false, null, null);
return(0);
}
public void PostProcessSegmentsTests()
{
var processor = new SegmentationResultsProcessor(100);
var chr1Segments = new List <SegmentationInput.Segment>();
chr1Segments.Add(new SegmentationInput.Segment()
{
start = 1, end = 1000
});
chr1Segments.Add(new SegmentationInput.Segment()
{
start = 1100, end = 4500
});
chr1Segments.Add(new SegmentationInput.Segment()
{
start = 4600, end = 5000
});
var segmentsByChrom = new Dictionary <string, SegmentationInput.Segment[]>();
segmentsByChrom.Add("chr1", chr1Segments.ToArray());
var segmentationResults = new GenomeSegmentationResults(segmentsByChrom);
var ploidyInfo = new PloidyInfo();
var excludedIntervals = new Dictionary <string, List <SampleGenomicBin> >();
var coverageInfo = new CoverageInfo();
coverageInfo.CoverageByChr = new Dictionary <string, double[]>();
coverageInfo.EndByChr = new Dictionary <string, uint[]>();
coverageInfo.StartByChr = new Dictionary <string, uint[]>();
coverageInfo.CoverageByChr.Add("chr1", new double[] { 10, 10, 50, 100, 25, 10 });
coverageInfo.StartByChr.Add("chr1", new uint[] { 100, 600, 1200, 1300, 4001, 5000 });
coverageInfo.EndByChr.Add("chr1", new uint[] { 500, 890, 1299, 4000, 4500, 5050 });
var results = processor.PostProcessSegments(segmentationResults, ploidyInfo, excludedIntervals, coverageInfo);
var chr1Results = results["chr1"];
Assert.Equal(3, chr1Results.Count);
// Final segments should reflect boundaries of actual bins within them
// (in practice, these probably shouldn't disagree? but let's go theoretical here)
SegmentTestHelpers.CheckSegment(chr1Results[0], 100, 890, 10, 2);
SegmentTestHelpers.CheckSegment(chr1Results[1], 1200, 4500, 50, 3);
SegmentTestHelpers.CheckSegment(chr1Results[2], 5000, 5050, 10, 1); // Bin extends past segment - still keep it (?)
// Add forbidden zone between two bins of the same original segment, this should split up the affected segment
excludedIntervals.Add("chr1", new List <SampleGenomicBin>()
{
GetForbiddenZone("chr1", 525, 575)
}); // Mid = 550, in between the bins of the first segment
results = processor.PostProcessSegments(segmentationResults, ploidyInfo, excludedIntervals, coverageInfo);
chr1Results = results["chr1"];
Assert.Equal(4, chr1Results.Count);
SegmentTestHelpers.CheckSegment(chr1Results[0], 100, 500, 10, 1);
SegmentTestHelpers.CheckSegment(chr1Results[1], 600, 890, 10, 1);
SegmentTestHelpers.CheckSegment(chr1Results[2], 1200, 4500, 50, 3);
SegmentTestHelpers.CheckSegment(chr1Results[3], 5000, 5050, 10, 1); // Bin extends past segment - still keep it (?)
// Forbidden zone midpoint is in the second bin -- apparently this is presumed to never happen because it would have already been taken care of
// This fails the test with the Debug Asserts in there. Otherwise it would be counted as a new bin
excludedIntervals.Clear();
excludedIntervals.Add("chr1", new List <SampleGenomicBin>()
{
GetForbiddenZone("chr1", 585, 635)
}); // Mid = 610, in second bin
results = processor.PostProcessSegments(segmentationResults, ploidyInfo, excludedIntervals, coverageInfo);
chr1Results = results["chr1"];
Assert.Equal(4, chr1Results.Count);
SegmentTestHelpers.CheckSegment(chr1Results[0], 100, 500, 10, 1);
SegmentTestHelpers.CheckSegment(chr1Results[1], 600, 890, 10, 1);
SegmentTestHelpers.CheckSegment(chr1Results[2], 1200, 4500, 50, 3);
SegmentTestHelpers.CheckSegment(chr1Results[3], 5000, 5050, 10, 1); // Bin extends past segment - still keep it (?)
// Forbidden zone midpoint is in the first bin although it ends between bins -- apparently this is presumed to never happen because it would have already been taken care of
// Note the asymmetry compared to the above
excludedIntervals.Clear();
excludedIntervals.Add("chr1", new List <SampleGenomicBin>()
{
GetForbiddenZone("chr1", 465, 515)
}); // Mid = 490, in first bin of first segment
results = processor.PostProcessSegments(segmentationResults, ploidyInfo, excludedIntervals, coverageInfo);
chr1Results = results["chr1"];
// Would fail - asymmetry. What do we want?
// Leave as-is for now so as not to change the behavior in this (unrelated) feature addition
//Assert.Equal(4, chr1Results.Count);
//SegmentTestHelpers.CheckSegment(chr1Results[0], 100, 500, 10, 1);
//SegmentTestHelpers.CheckSegment(chr1Results[1], 600, 890, 10, 1);
//SegmentTestHelpers.CheckSegment(chr1Results[2], 1200, 4500, 50, 3);
//SegmentTestHelpers.CheckSegment(chr1Results[3], 5000, 5050, 10, 1); // Bin extends past segment - still keep it (?)
// TODO test where no segment covers bins?
// TODO overlapping segments or bins?
// TODO bin starts before segment
// TODO test interbin dist
}