public static PloidyInfo LoadPloidyFromBedFile(string filePath)
{
PloidyInfo ploidy = new PloidyInfo();
int count = 0;
using (GzipReader reader = new GzipReader(filePath))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null) break;
if (fileLine.StartsWith("##ExpectedSexChromosomeKaryotype"))
{
ploidy.HeaderLine = fileLine.Trim();
continue;
}
if (fileLine.Length == 0 || fileLine[0] == '#') continue;
string[] bits = fileLine.Split('\t');
string chromosome = bits[0];
if (!ploidy.PloidyByChromosome.ContainsKey(chromosome))
{
ploidy.PloidyByChromosome[chromosome] = new List<PloidyInterval>();
}
PloidyInterval interval = new PloidyInterval();
interval.Start = int.Parse(bits[1]);
interval.End = int.Parse(bits[2]);
interval.Ploidy = int.Parse(bits[4]);
ploidy.PloidyByChromosome[chromosome].Add(interval);
count++;
}
}
Console.WriteLine("Reference ploidy: Loaded {0} intervals across {1} chromosomes", count, ploidy.PloidyByChromosome.Keys.Count);
return ploidy;
}
public static PloidyInfo LoadPloidyFromVcfFile(string vcfPath, string sampleName)
{
int sampleIndex = 0;
PloidyInfo ploidy = new PloidyInfo();
using (VcfReader reader = new VcfReader(vcfPath))
{
if (!sampleName.IsNullOrEmpty())
{
if (!sampleName.IsNullOrEmpty() && reader.Samples.Count < 2)
{
throw new ArgumentException(
$"File '{vcfPath}' must be a multi-sample sample VCF containing > 1 samples");
}
if (reader.Samples.Select(x => Convert.ToInt32(x == sampleName)).Sum() != 1)
{
throw new ArgumentException(
$"File '{vcfPath}' should contain one genotypes column corresponding to sample {sampleName}");
}
sampleIndex = reader.Samples.IndexOf(sampleName);
}
ploidy.HeaderLine = string.Join(" ", reader.HeaderLines);
while (true)
{
VcfVariant record;
bool result = reader.GetNextVariant(out record);
if (!result)
{
break;
}
string chromosome = record.ReferenceName;
if (!ploidy.PloidyByChromosome.ContainsKey(chromosome))
{
ploidy.PloidyByChromosome[chromosome] = new List <PloidyInterval>();
}
PloidyInterval interval = new PloidyInterval(chromosome);
interval.Start = record.ReferencePosition;
interval.End = int.Parse(record.InfoFields["END"]);
var genotypeColumn = record.GenotypeColumns[sampleIndex];
if (genotypeColumn.ContainsKey("CN"))
{
var value = genotypeColumn["CN"];
interval.Ploidy = value == "." ? 2 : int.Parse(value);
}
else
{
throw new ArgumentException($"File '{vcfPath}' must contain one genotype CN column!");
}
ploidy.PloidyByChromosome[chromosome].Add(interval);
}
}
return(ploidy);
}
public static void WriteSegments(string outVcfPath, List <CanvasSegment> segments, double?diploidCoverage,
string wholeGenomeFastaDirectory, string sampleName,
List <string> extraHeaders, PloidyInfo ploidy, int qualityThreshold, bool isPedigreeInfoSupplied, int?denovoQualityThreshold = null)
{
using (BgzipOrStreamWriter writer = new BgzipOrStreamWriter(outVcfPath))
{
var genome = WriteVcfHeader(segments, diploidCoverage, wholeGenomeFastaDirectory, new List <string> {
sampleName
},
extraHeaders, qualityThreshold, writer, denovoQualityThreshold);
WriteVariants(new List <List <CanvasSegment> > {
segments.ToList()
}, ploidy, genome, writer, isPedigreeInfoSupplied, denovoQualityThreshold);
}
}
public static PloidyInfo LoadPloidyFromBedFile(string filePath)
{
PloidyInfo ploidy = new PloidyInfo();
if (string.IsNullOrEmpty(filePath))
{
return(ploidy);
}
int count = 0;
using (GzipReader reader = new GzipReader(filePath))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null)
{
break;
}
// save anything that looks like a vcf header line (we will add it to the output vcf)
// TODO: support adding multiple header lines to the output vcf
if (fileLine.StartsWith("##"))
{
ploidy.HeaderLine = fileLine.Trim();
continue;
}
if (fileLine.Length == 0 || fileLine[0] == '#')
{
continue;
}
string[] bits = fileLine.Split('\t');
string chromosome = bits[0];
if (!ploidy.PloidyByChromosome.ContainsKey(chromosome))
{
ploidy.PloidyByChromosome[chromosome] = new List <PloidyInterval>();
}
PloidyInterval interval = new PloidyInterval(chromosome);
interval.Start = int.Parse(bits[1]);
interval.End = int.Parse(bits[2]);
interval.Ploidy = int.Parse(bits[4]);
ploidy.PloidyByChromosome[chromosome].Add(interval);
count++;
}
}
Console.WriteLine("Reference ploidy: Loaded {0} intervals across {1} chromosomes", count, ploidy.PloidyByChromosome.Keys.Count);
return(ploidy);
}
public static void WriteSegments(string outVcfPath, List <CanvasSegment> segments, double?diploidCoverage,
string wholeGenomeFastaDirectory, string sampleName,
List <string> extraHeaders, PloidyInfo ploidy, int qualityThreshold, bool isPedigreeInfoSupplied, int?denovoQualityThreshold, int?sizeThreshold)
{
using (BgzipOrStreamWriter writer = new BgzipOrStreamWriter(outVcfPath))
{
var genome = WriteVcfHeader(segments, diploidCoverage, wholeGenomeFastaDirectory, new List <string> {
sampleName
},
extraHeaders, writer, qualityThreshold, denovoQualityThreshold, sizeThreshold);
var sampleId = new SampleId(sampleName);
var segmentsOfAllSamples = segments.Select(x => new SampleMap <CanvasSegment> {
{ sampleId, x }
});
WriteVariants(segmentsOfAllSamples, new List <PloidyInfo> {
ploidy
}, genome, writer, denovoQualityThreshold);
}
}
private static PloidyInfo LoadPloidyFromVcfFile(string vcfPath, int sampleIndex)
{
PloidyInfo ploidy = new PloidyInfo();
using (VcfReader reader = new VcfReader(vcfPath))
{
//the ploidy.vcf header lines need to be updated to include reference sex chromosome info for one or multiple samples
//ploidy.HeaderLine = string.Join(" ", reader.HeaderLines);
while (true)
{
bool result = reader.GetNextVariant(out var record);
if (!result)
{
break;
}
string chromosome = record.ReferenceName;
if (!ploidy.PloidyByChromosome.ContainsKey(chromosome))
{
ploidy.PloidyByChromosome[chromosome] = new List <PloidyInterval>();
}
PloidyInterval interval = new PloidyInterval(chromosome)
{
Start = record.ReferencePosition,
End = int.Parse(record.InfoFields["END"])
};
var genotypeColumn = record.GenotypeColumns[sampleIndex];
if (genotypeColumn.ContainsKey("CN"))
{
var value = genotypeColumn["CN"];
interval.Ploidy = value == "." ? 2 : int.Parse(value);
}
else
{
throw new ArgumentException($"File '{vcfPath}' must contain one genotype CN column!");
}
ploidy.PloidyByChromosome[chromosome].Add(interval);
}
}
return(ploidy);
}
public static PloidyInfo LoadPloidyFromBedFile(string filePath)
{
PloidyInfo ploidy = new PloidyInfo();
int count = 0;
using (GzipReader reader = new GzipReader(filePath))
{
while (true)
{
string fileLine = reader.ReadLine();
if (fileLine == null)
{
break;
}
if (fileLine.StartsWith("##ExpectedSexChromosomeKaryotype"))
{
ploidy.HeaderLine = fileLine.Trim();
continue;
}
if (fileLine.Length == 0 || fileLine[0] == '#')
{
continue;
}
string[] bits = fileLine.Split('\t');
string chromosome = bits[0];
if (!ploidy.PloidyByChromosome.ContainsKey(chromosome))
{
ploidy.PloidyByChromosome[chromosome] = new List <PloidyInterval>();
}
PloidyInterval interval = new PloidyInterval();
interval.Start = int.Parse(bits[1]);
interval.End = int.Parse(bits[2]);
interval.Ploidy = int.Parse(bits[4]);
ploidy.PloidyByChromosome[chromosome].Add(interval);
count++;
}
}
Console.WriteLine("Reference ploidy: Loaded {0} intervals across {1} chromosomes", count, ploidy.PloidyByChromosome.Keys.Count);
return(ploidy);
}
private 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);
}
/// <summary>
/// Generate a tabular file with information about coverage and allele frequency for each chunk of the genome.
/// This file can be used to generate a pretty plot of coverage versus MAF.
/// </summary>
public static void WriteCoveragePlotData(List <CanvasSegment> segments, double?normalDiploidCoverage, PloidyInfo referencePloidy,
string filePath, string referenceFolder)
{
if (segments.Any() && !normalDiploidCoverage.HasValue)
{
throw new Illumina.Common.IlluminaException("normal diploid coverage must be specified");
}
int pointLength = 100000;
int minimumBinsToPlot = GetMinimumBinsForCoveragePlotPoint(segments, pointLength);
Dictionary <string, List <CanvasSegment> > segmentsByChromosome = GetSegmentsByChromosome(segments);
GenomeMetadata genome = new GenomeMetadata();
genome.Deserialize(Path.Combine(referenceFolder, "GenomeSize.xml"));
List <float> counts = new List <float>();
List <float> MAF = new List <float>();
List <float> VF = new List <float>();
using (FileStream stream = new FileStream(filePath, FileMode.Create, FileAccess.Write))
using (StreamWriter writer = new StreamWriter(stream))
{
writer.NewLine = "\n";
writer.Write("#Chromosome\tStart\tEnd\tCopyNumber\tMajorChromosomeCount\tMedianHits\tNormalizedCoverage\tMedianMinorAlleleFrequency\tReferencePloidy\t");
for (int i = 0; i < NumberVariantFrequencyBins; i++)
{
writer.Write("VariantFrequencyBin{0}\t", i);
}
writer.WriteLine();
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
if (!segmentsByChromosome.ContainsKey(chromosome.Name))
{
continue;
}
int pointStartPos = 0; // 0-based start
while (pointStartPos < chromosome.Length)
{
int pointEndPos = (int)Math.Min(chromosome.Length, pointStartPos + pointLength); // 1-based end
counts.Clear();
MAF.Clear();
VF.Clear();
Dictionary <string, long> CopyNumberAndChromCount = new Dictionary <string, long>();
Dictionary <int, long> basesByCopyNumber = new Dictionary <int, long>();
// Accumulate counts and MAF from the segments:
List <CanvasSegment> chrSegments = new List <CanvasSegment>();
if (segmentsByChromosome.ContainsKey(chromosome.Name))
{
chrSegments = segmentsByChromosome[chromosome.Name];
}
List <CanvasSegment> overlapSegments = new List <CanvasSegment>();
foreach (CanvasSegment segment in chrSegments)
{
if (segment.Begin > pointEndPos)
{
continue;
}
if (segment.End < pointStartPos)
{
continue;
}
int weight = Math.Min(segment.End, pointEndPos) - Math.Max(segment.Begin, pointStartPos);
string key = string.Format("{0} {1}", segment.CopyNumber, segment.MajorChromosomeCount);
if (!CopyNumberAndChromCount.ContainsKey(key))
{
CopyNumberAndChromCount[key] = 0;
}
CopyNumberAndChromCount[key] += weight;
if (!basesByCopyNumber.ContainsKey(segment.CopyNumber))
{
basesByCopyNumber[segment.CopyNumber] = 0;
}
basesByCopyNumber[segment.CopyNumber] += weight;
overlapSegments.Add(segment);
}
// Note the most common copy number:
long bestCount = 0;
int majorCopyNumber = 0;
foreach (int key in basesByCopyNumber.Keys)
{
if (basesByCopyNumber[key] > bestCount)
{
bestCount = basesByCopyNumber[key];
majorCopyNumber = key;
}
}
// Find the most common major chromosome count, for the most common copy number:
int?majorChromosomeCount = null;
bestCount = 0;
foreach (string key in CopyNumberAndChromCount.Keys)
{
string[] bits = key.Split();
if (bits[1].Length == 0)
{
continue;
}
if (int.Parse(bits[0]) != majorCopyNumber)
{
continue;
}
long count = CopyNumberAndChromCount[key];
if (count < bestCount)
{
continue;
}
bestCount = count;
majorChromosomeCount = int.Parse(bits[1]);
}
// Note allele frequency and coverage info, for all overlap segments that match (more or less)
// the most common copy number:
foreach (CanvasSegment segment in overlapSegments)
{
if ((majorCopyNumber == 2 && segment.CopyNumber != 2) ||
(majorCopyNumber < 2 && segment.CopyNumber >= 2) ||
(majorCopyNumber > 2 && segment.CopyNumber <= 2))
{
continue;
}
float segLength = segment.End - segment.Begin;
// Add counts to the overall list:
int firstIndex = 0;
if (pointStartPos > segment.Begin)
{
firstIndex = (int)((float)segment.Counts.Count * (pointStartPos - segment.Begin) / segLength);
}
int lastIndex = segment.Counts.Count;
if (pointEndPos < segment.End)
{
lastIndex = (int)((float)segment.Counts.Count * (pointEndPos - segment.Begin) / segLength);
}
for (int index = firstIndex; index < lastIndex; index++)
{
counts.Add(segment.Counts[index]);
}
// Add MAF to the overall list:
firstIndex = 0;
if (pointStartPos > segment.Begin)
{
firstIndex = (int)((float)segment.Alleles.Frequencies.Count * (pointStartPos - segment.Begin) / segLength);
}
lastIndex = segment.Alleles.Frequencies.Count;
if (pointEndPos < segment.End)
{
lastIndex = (int)((float)segment.Alleles.Frequencies.Count * (pointEndPos - segment.Begin) / segLength);
}
for (int index = firstIndex; index < lastIndex; index++)
{
float tempMAF = segment.Alleles.Frequencies[index];
VF.Add(tempMAF);
if (tempMAF > 0.5)
{
tempMAF = 1 - tempMAF;
}
MAF.Add(tempMAF);
}
}
// Write output for this point:
writer.Write("{0}\t{1}\t{2}\t", chromosome.Name, pointStartPos, pointEndPos);
// Write counts if we have reasonable amounts of data; write MAF if we have reasonable amounts of data.
// (Note: Observed that for germline data on chrY we often had well under 100 counts given the new, smaller bin size)
if (counts.Count >= minimumBinsToPlot)
{
writer.Write("{0}\t", majorCopyNumber);
writer.Write("{0}\t", majorChromosomeCount);
counts.Sort();
double medianHits = counts[counts.Count / 2];
writer.Write("{0:F2}\t", medianHits);
double normalizedCount = 2 * medianHits / normalDiploidCoverage.Value;
writer.Write("{0:F2}\t", normalizedCount);
if (MAF.Count >= 10)
{
MAF.Sort();
writer.Write("{0}\t", MAF[MAF.Count / 2]);
}
else
{
writer.Write("\t");
}
int refPloidy = 2;
if (referencePloidy != null && referencePloidy.PloidyByChromosome.ContainsKey(chromosome.Name))
{
foreach (var interval in referencePloidy.PloidyByChromosome[chromosome.Name])
{
if (interval.Start <= pointEndPos && interval.End >= pointStartPos)
{
refPloidy = interval.Ploidy;
}
}
}
writer.Write("{0}\t", refPloidy);
if (VF.Count >= 10)
{
// bin VF
float[] vfDistribution = new float[NumberVariantFrequencyBins];
foreach (float vf in VF)
{
int binNumber = Math.Min(vfDistribution.Length - 1, (int)Math.Floor(vf / 0.01));
vfDistribution[binNumber]++;
}
for (int i = 0; i < vfDistribution.Length; i++)
{
vfDistribution[i] = vfDistribution[i] / (float)VF.Count * 100.0f;
writer.Write("{0:F2}\t", vfDistribution[i]);
}
}
else
{
for (int i = 0; i < NumberVariantFrequencyBins; i++)
{
writer.Write("\t");
}
}
}
writer.WriteLine();
pointStartPos += pointLength;
}
}
}
}
private const double EMLikelihoodThres = 1; // Controls when to update means
#endregion
/// <summary>
/// Load the expected ploidy for sex chromosomes from a .bed file. This lets us know that, for instance, copy number 2
/// on chrX is a GAIN (not REF) call for a male (XY) sample.
/// </summary>
public void LoadReferencePloidy(string filePath)
{
Console.WriteLine(">>>LoadReferencePloidy({0})", filePath);
ReferencePloidy = PloidyInfo.LoadPloidyFromBedFile(filePath);
}
/// <summary>
/// Generate a tabular file with information about coverage and allele frequency for each chunk of the genome.
/// This file can be used to generate a pretty plot of coverage versus MAF.
/// </summary>
static public void WriteCoveragePlotData(List<CanvasSegment> segments, double normalDiploidCoverage, PloidyInfo referencePloidy,
string filePath, string referenceFolder)
{
Dictionary<string, List<CanvasSegment>> segmentsByChromosome = GetSegmentsByChromosome(segments);
GenomeMetadata genome = new GenomeMetadata();
genome.Deserialize(Path.Combine(referenceFolder, "GenomeSize.xml"));
int pointLength = 100000;
List<float> counts = new List<float>();
List<float> MAF = new List<float>();
List<float> VF = new List<float>();
using (StreamWriter writer = new StreamWriter(filePath))
{
writer.NewLine = "\n";
writer.Write("#Chromosome\tStart\tEnd\tCopyNumber\tMajorChromosomeCount\tMedianHits\tNormalizedCoverage\tMedianMinorAlleleFrequency\tReferencePloidy\t");
for (int i = 0; i < NumberVariantFrequencyBins; i++) { writer.Write("VariantFrequencyBin{0}\t", i); }
writer.WriteLine();
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
if (chromosome.IsMito()) continue;
int pointStartPos = 0; // 0-based start
while (pointStartPos < chromosome.Length)
{
int pointEndPos = (int)Math.Min(chromosome.Length, pointStartPos + pointLength); // 1-based end
counts.Clear();
MAF.Clear();
VF.Clear();
Dictionary<string, long> CopyNumberAndChromCount = new Dictionary<string, long>();
Dictionary<int, long> basesByCopyNumber = new Dictionary<int, long>();
// Accumulate counts and MAF from the segments:
List<CanvasSegment> chrSegments = new List<CanvasSegment>();
if (segmentsByChromosome.ContainsKey(chromosome.Name)) chrSegments = segmentsByChromosome[chromosome.Name];
List<CanvasSegment> overlapSegments = new List<CanvasSegment>();
foreach (CanvasSegment segment in chrSegments)
{
if (segment.Begin > pointEndPos) continue;
if (segment.End < pointStartPos) continue;
int weight = Math.Min(segment.End, pointEndPos) - Math.Max(segment.Begin, pointStartPos);
string key = string.Format("{0} {1}", segment.copyNumber, segment.MajorChromosomeCount);
if (!CopyNumberAndChromCount.ContainsKey(key)) CopyNumberAndChromCount[key] = 0;
CopyNumberAndChromCount[key] += weight;
if (!basesByCopyNumber.ContainsKey(segment.copyNumber)) basesByCopyNumber[segment.copyNumber] = 0;
basesByCopyNumber[segment.copyNumber] += weight;
overlapSegments.Add(segment);
}
// Note the most common copy number:
long bestCount = 0;
int majorCopyNumber = 0;
foreach (int key in basesByCopyNumber.Keys)
{
if (basesByCopyNumber[key] > bestCount)
{
bestCount = basesByCopyNumber[key];
majorCopyNumber = key;
}
}
// Find the most common major chromosome count, for the most common copy number:
int? majorChromosomeCount = null;
bestCount = 0;
foreach (string key in CopyNumberAndChromCount.Keys)
{
string[] bits = key.Split();
if (bits[1].Length == 0) continue;
if (int.Parse(bits[0]) != majorCopyNumber) continue;
long count = CopyNumberAndChromCount[key];
if (count < bestCount) continue;
bestCount = count;
majorChromosomeCount = int.Parse(bits[1]);
}
// Note allele frequency and coverage info, for all overlap segments that match (more or less)
// the most common copy number:
foreach (CanvasSegment segment in overlapSegments)
{
if ((majorCopyNumber == 2 && segment.copyNumber != 2) ||
(majorCopyNumber < 2 && segment.copyNumber >= 2) ||
(majorCopyNumber > 2 && segment.copyNumber <= 2))
continue;
float segLength = segment.End - segment.Begin;
// Add counts to the overall list:
int firstIndex = 0;
if (pointStartPos > segment.Begin)
{
firstIndex = (int)((float)segment.Counts.Count * (pointStartPos - segment.Begin) / segLength);
}
int lastIndex = segment.Counts.Count;
if (pointEndPos < segment.End)
{
lastIndex = (int)((float)segment.Counts.Count * (pointEndPos - segment.Begin) / segLength);
}
for (int index = firstIndex; index < lastIndex; index++) counts.Add(segment.Counts[index]);
// Add MAF to the overall list:
firstIndex = 0;
if (pointStartPos > segment.Begin)
{
firstIndex = (int)((float)segment.VariantFrequencies.Count * (pointStartPos - segment.Begin) / segLength);
}
lastIndex = segment.VariantFrequencies.Count;
if (pointEndPos < segment.End)
{
lastIndex = (int)((float)segment.VariantFrequencies.Count * (pointEndPos - segment.Begin) / segLength);
}
for (int index = firstIndex; index < lastIndex; index++)
{
float tempMAF = segment.VariantFrequencies[index];
VF.Add(tempMAF);
if (tempMAF > 0.5) tempMAF = 1 - tempMAF;
MAF.Add(tempMAF);
}
}
// Write output for this point:
writer.Write("{0}\t{1}\t{2}\t", chromosome.Name, pointStartPos, pointEndPos);
// Write counts if we have reasonable amounts of data; write MAF if we have reasonable amounts of data.
// (Note: Observed that for germline data on chrY we often had well under 100 counts given the new, smaller bin size)
if (counts.Count >= 30)
{
writer.Write("{0}\t", majorCopyNumber);
writer.Write("{0}\t", majorChromosomeCount);
counts.Sort();
double medianHits = counts[counts.Count / 2];
writer.Write("{0:F2}\t", medianHits);
double normalizedCount = 2 * medianHits / normalDiploidCoverage;
writer.Write("{0:F2}\t", normalizedCount);
if (MAF.Count >= 10)
{
MAF.Sort();
writer.Write("{0}\t", MAF[MAF.Count / 2]);
}
else
{
writer.Write("\t");
}
int refPloidy = 2;
if (referencePloidy != null && referencePloidy.PloidyByChromosome.ContainsKey(chromosome.Name))
{
foreach (var interval in referencePloidy.PloidyByChromosome[chromosome.Name])
{
if (interval.Start <= pointEndPos && interval.End >= pointStartPos)
{
refPloidy = interval.Ploidy;
}
}
}
writer.Write("{0}\t", refPloidy);
if (VF.Count >= 10)
{
// bin VF
float[] vfDistribution = new float[NumberVariantFrequencyBins];
foreach (float vf in VF)
{
int binNumber = Math.Min(vfDistribution.Length - 1, (int)Math.Floor(vf / 0.01));
vfDistribution[binNumber]++;
}
for (int i = 0; i < vfDistribution.Length; i++)
{
vfDistribution[i] = vfDistribution[i] / (float)VF.Count * 100.0f;
writer.Write("{0:F2}\t", vfDistribution[i]);
}
}
else
{
for (int i = 0; i < NumberVariantFrequencyBins; i++) writer.Write("\t");
}
}
writer.WriteLine();
pointStartPos += pointLength;
}
}
}
}
/// <summary>
/// Outputs the copy number calls to a text file.
/// </summary>
/// <param name="outVcfPath">File to write to.</param>
/// <param name="segments">List of segments to write out.</param>
public static void WriteSegments(string outVcfPath, List<CanvasSegment> segments, string reference, string sampleName,
List<string> extraHeaders, bool reportPloidy, PloidyInfo ploidy, bool reportAllSites = false, bool reportGermlineGenotype = false)
{
string cnvtype = null;
string filter = null;
// report GT for resequencing workflow and MCC for tumour-normal workflow
if (reportGermlineGenotype && reportPloidy)
{
throw new Exception("WriteSegments VCF file output error: reportGermlineGenotype and reportPloidy can not be both true");
}
using (BgzipOrStreamWriter writer = new BgzipOrStreamWriter(outVcfPath))
{
// Write the VCF header:
writer.WriteLine("##fileformat=VCFv4.1");
writer.WriteLine("##source=Isas," + CanvasCommon.CanvasVersionInfo.NameString + " " + CanvasCommon.CanvasVersionInfo.VersionString);
writer.WriteLine("##reference={0}", Path.Combine(reference, "genome.fa"));
if (extraHeaders != null)
{
foreach (string header in extraHeaders)
{
writer.WriteLine(header);
}
}
GenomeMetadata genome = new GenomeMetadata();
genome.Deserialize(Path.Combine(reference, "GenomeSize.xml"));
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
writer.WriteLine("##contig=<ID={0},length={1}>", chromosome.Name, chromosome.Length);
}
writer.WriteLine("##ALT=<ID=CNV,Description=\"Copy number variable region\">");
writer.WriteLine("##FILTER=<ID=q10,Description=\"Quality below 10\">");
writer.WriteLine("##FILTER=<ID=L10kb,Description=\"Length shorter than 10kb\">");
writer.WriteLine("##INFO=<ID=SVTYPE,Number=1,Type=String,Description=\"Type of structural variant\">");
writer.WriteLine("##INFO=<ID=END,Number=1,Type=Integer,Description=\"End position of the variant described in this record\">");
writer.WriteLine("##INFO=<ID=CNVLEN,Number=1,Type=Integer,Description=\"Number of reference positions spanned by this CNV\">");
if (reportGermlineGenotype)
writer.WriteLine("##FORMAT=<ID=GT,Number=1,Type=String,Description=\"Genotype\">");
writer.WriteLine("##FORMAT=<ID=RC,Number=1,Type=Float,Description=\"Mean counts per bin in the region\">");
writer.WriteLine("##FORMAT=<ID=BC,Number=1,Type=Float,Description=\"Number of bins in the region\">");
writer.WriteLine("##FORMAT=<ID=CN,Number=1,Type=Integer,Description=\"Copy number genotype for imprecise events\">");
if (reportPloidy)
writer.WriteLine("##FORMAT=<ID=MCC,Number=1,Type=Integer,Description=\"Major chromosome count (equal to copy number for LOH regions)\">");
writer.WriteLine("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\t" + sampleName);
SanityCheckChromosomeNames(genome, segments);
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
foreach (CanvasSegment segment in segments)
{
if (segment.Chr.ToLowerInvariant() != chromosome.Name.ToLowerInvariant()) continue;
int referenceCN = 2;
if (ploidy != null) referenceCN = ploidy.GetReferenceCopyNumber(segment);
filter = null;
bool isReferenceCall = false;
if (segment.CopyNumber == referenceCN) isReferenceCall = true;
if (reportPloidy && segment.CopyNumber == 2 && segment.MajorChromosomeCount.HasValue && segment.MajorChromosomeCount != 1) isReferenceCall = false; // If we're reporting ploidy and there's LOH, this isn't a reference call.
// We can skip reporting of reference sites:
if (!reportAllSites && isReferenceCall)
continue;
if (segment.QScore < 10)
filter = "q10";
if (segment.End - segment.Begin < 10000)
{
if (filter != null)
filter = filter + ";L10kb";
else
filter = "L10kb";
}
if (filter == null)
filter = "PASS";
if (segment.CopyNumber < referenceCN)
cnvtype = "LOSS";
else if (segment.CopyNumber > referenceCN)
cnvtype = "GAIN";
else
cnvtype = "REF";
// The Dude abides... from vcf 4.1 spec:
// If any of the ALT alleles is a symbolic allele (an angle-bracketed ID String “<ID>”) then the padding base is required and POS denotes the
// coordinate of the base preceding the polymorphism.
writer.Write("{0}\t{1}\tCanvas:{2}:{0}:{3}-{4}\t", segment.Chr, isReferenceCall ? segment.Begin + 1 : segment.Begin, cnvtype, segment.Begin + 1, segment.End);
writer.Write("N\t{0}\t{1}\t{2}\t", isReferenceCall ? "." : "<CNV>", segment.QScore, filter);
if (segment.copyNumber != referenceCN)
writer.Write("SVTYPE=CNV;");
else if (!isReferenceCall)
writer.Write("SVTYPE=LOH;");
if (segment.copyNumber != referenceCN || !isReferenceCall)
writer.Write("END={0};CNVLEN={1}", segment.End, segment.End - segment.Begin);
else
writer.Write("END={0}", segment.End);
// FORMAT field
if (reportGermlineGenotype)
writer.Write("\tGT:RC:BC:CN", segment.End);
else
writer.Write("\tRC:BC:CN", segment.End);
if (reportPloidy && segment.MajorChromosomeCount.HasValue) writer.Write(":MCC");
// writing GT for resequencing workflow
if (reportGermlineGenotype)
{
writer.Write("\t{0}/{1}:", segment.MajorChromosomeCount, segment.CopyNumber);
}
else
writer.Write("\t");
writer.Write("{1}:{2}:{3}", segment.End, Math.Round(segment.MeanCount, 0, MidpointRounding.AwayFromZero), segment.BinCount, segment.CopyNumber);
// writing MCC for tumour-normal workflow
if (reportPloidy && segment.MajorChromosomeCount.HasValue)
{
writer.Write(":{0}", segment.MajorChromosomeCount);
}
writer.WriteLine();
}
}
}
}
public static void WriteMultiSampleSegments(string outVcfPath, List <List <CanvasSegment> > segments, List <double?> diploidCoverage,
string wholeGenomeFastaDirectory, List <string> sampleNames, List <string> extraHeaders, PloidyInfo ploidy, int qualityThreshold,
bool isPedigreeInfoSupplied = true, int?denovoQualityThreshold = null)
{
using (BgzipOrStreamWriter writer = new BgzipOrStreamWriter(outVcfPath))
{
var genome = WriteVcfHeader(segments.First(), GetMean(diploidCoverage), wholeGenomeFastaDirectory, sampleNames,
extraHeaders, qualityThreshold, writer, denovoQualityThreshold);
WriteVariants(segments, ploidy, genome, writer, isPedigreeInfoSupplied, denovoQualityThreshold);
}
}
/// <summary>
/// Outputs the copy number calls to a text file.
/// </summary>
private static void WriteVariants(IReadOnlyCollection <List <CanvasSegment> > segments, PloidyInfo ploidy, GenomeMetadata genome,
BgzipOrStreamWriter writer, bool isPedigreeInfoSupplied = true, int?denovoQualityThreshold = null)
{
var nSamples = segments.Count;
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
for (int segmentIndex = 0; segmentIndex < segments.First().Count; segmentIndex++)
{
var firstSampleSegment = segments.First()[segmentIndex];
if (!isPedigreeInfoSupplied && segments.Select(sample => sample[segmentIndex].Filter == "PASS").Any() && segments.Count > 1)
{
firstSampleSegment.Filter = "PASS";
}
if (!firstSampleSegment.Chr.Equals(chromosome.Name, StringComparison.OrdinalIgnoreCase))
{
continue;
}
var referenceCopyNumbers = segments.Select(segment => ploidy?.GetReferenceCopyNumber(segment[segmentIndex]) ?? 2).ToList();
var currentSegments = segments.Select(x => x[segmentIndex]).ToList();
var cnvTypes = new List <CnvType>();
for (int sampleIndex = 0; sampleIndex < nSamples; sampleIndex++)
{
cnvTypes.Add(currentSegments[sampleIndex].GetCnvType(referenceCopyNumbers[sampleIndex]));
}
CnvType cnvType;
if (cnvTypes.TrueForAll(x => x == CnvType.Reference))
{
cnvType = CnvType.Reference;
}
else if (cnvTypes.TrueForAll(x => x == CnvType.Reference | x == CnvType.Loss))
{
cnvType = CnvType.Loss;
}
else if (cnvTypes.TrueForAll(x => x == CnvType.Reference | x == CnvType.Gain))
{
cnvType = CnvType.Gain;
}
else if (cnvTypes.TrueForAll(x => x == CnvType.Reference | x == CnvType.LossOfHeterozygosity))
{
cnvType = CnvType.LossOfHeterozygosity;
}
else
{
cnvType = CnvType.ComplexCnv;
}
WriteInfoField(writer, firstSampleSegment, cnvType, denovoQualityThreshold, isMultisample: segments.Count > 1);
// FORMAT field
if (segments.Count == 1)
{
WriteSingleSampleInfo(writer, firstSampleSegment);
}
else
{
WriteFormatField(writer, currentSegments);
}
}
}
}
/// <summary>
/// Outputs the copy number calls to a text file.
/// </summary>
public static void WriteSegments(string outVcfPath, List<CanvasSegment> segments, string wholeGenomeFastaDirectory, string sampleName,
List<string> extraHeaders, PloidyInfo ploidy, int qualityThreshold = 10)
{
using (BgzipOrStreamWriter writer = new BgzipOrStreamWriter(outVcfPath))
{
// Write the VCF header:
writer.WriteLine("##fileformat=VCFv4.1");
writer.WriteLine($"##source={CanvasVersionInfo.NameString} {CanvasVersionInfo.VersionString}");
writer.WriteLine($"##reference={Path.Combine(wholeGenomeFastaDirectory, "genome.fa")}");
foreach (string header in extraHeaders ?? new List<string>())
{
writer.WriteLine(header);
}
GenomeMetadata genome = new GenomeMetadata();
genome.Deserialize(Path.Combine(wholeGenomeFastaDirectory, "GenomeSize.xml"));
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
writer.WriteLine($"##contig=<ID={chromosome.Name},length={chromosome.Length}>");
}
string qualityFilter = $"q{qualityThreshold}";
writer.WriteLine("##ALT=<ID=CNV,Description=\"Copy number variable region\">");
writer.WriteLine($"##FILTER=<ID={qualityFilter},Description=\"Quality below {qualityThreshold}\">");
writer.WriteLine("##FILTER=<ID=L10kb,Description=\"Length shorter than 10kb\">");
writer.WriteLine("##INFO=<ID=SVTYPE,Number=1,Type=String,Description=\"Type of structural variant\">");
writer.WriteLine("##INFO=<ID=END,Number=1,Type=Integer,Description=\"End position of the variant described in this record\">");
writer.WriteLine("##INFO=<ID=CNVLEN,Number=1,Type=Integer,Description=\"Number of reference positions spanned by this CNV\">");
writer.WriteLine("##FORMAT=<ID=RC,Number=1,Type=Float,Description=\"Mean counts per bin in the region\">");
writer.WriteLine("##FORMAT=<ID=BC,Number=1,Type=Float,Description=\"Number of bins in the region\">");
writer.WriteLine("##FORMAT=<ID=CN,Number=1,Type=Integer,Description=\"Copy number genotype for imprecise events\">");
writer.WriteLine("##FORMAT=<ID=MCC,Number=1,Type=Integer,Description=\"Major chromosome count (equal to copy number for LOH regions)\">");
writer.WriteLine("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\t" + sampleName);
SanityCheckChromosomeNames(genome, segments);
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
foreach (CanvasSegment segment in segments)
{
if (!segment.Chr.Equals(chromosome.Name, StringComparison.OrdinalIgnoreCase)) continue;
int referenceCopyNumber = ploidy?.GetReferenceCopyNumber(segment) ?? 2;
CnvType cnvType = segment.GetCnvType(referenceCopyNumber);
// From vcf 4.1 spec:
// If any of the ALT alleles is a symbolic allele (an angle-bracketed ID String “<ID>”) then the padding base is required and POS denotes the
// coordinate of the base preceding the polymorphism.
string alternateAllele = cnvType.ToAltId();
int position = (alternateAllele.StartsWith("<") && alternateAllele.EndsWith(">")) ? segment.Begin : segment.Begin + 1;
writer.Write($"{segment.Chr}\t{position}\tCanvas:{cnvType.ToVcfId()}:{segment.Chr}:{segment.Begin + 1}-{segment.End}\t");
writer.Write($"N\t{alternateAllele}\t{segment.QScore}\t{segment.Filter}\t", alternateAllele, segment.QScore, segment.Filter);
if (cnvType != CnvType.Reference)
writer.Write($"SVTYPE={cnvType.ToSvType()};");
writer.Write($"END={segment.End}");
if (cnvType != CnvType.Reference)
writer.Write($";CNVLEN={segment.End - segment.Begin}");
// FORMAT field
writer.Write("\tRC:BC:CN", segment.End);
if (segment.MajorChromosomeCount.HasValue)
{
writer.Write(":MCC");
}
writer.Write("\t{1}:{2}:{3}", segment.End, Math.Round(segment.MeanCount, 0, MidpointRounding.AwayFromZero), segment.BinCount, segment.CopyNumber);
if (segment.MajorChromosomeCount.HasValue)
{
writer.Write(":{0}", segment.MajorChromosomeCount);
}
writer.WriteLine();
}
}
}
}
/// <summary>
/// Outputs the copy number calls to a text file.
/// </summary>
public static void WriteSegments(string outVcfPath, List <CanvasSegment> segments, string wholeGenomeFastaDirectory, string sampleName,
List <string> extraHeaders, PloidyInfo ploidy, int qualityThreshold = 10)
{
using (BgzipOrStreamWriter writer = new BgzipOrStreamWriter(outVcfPath))
{
// Write the VCF header:
writer.WriteLine("##fileformat=VCFv4.1");
writer.WriteLine($"##source={CanvasVersionInfo.NameString} {CanvasVersionInfo.VersionString}");
writer.WriteLine($"##reference={Path.Combine(wholeGenomeFastaDirectory, "genome.fa")}");
foreach (string header in extraHeaders ?? new List <string>())
{
writer.WriteLine(header);
}
GenomeMetadata genome = new GenomeMetadata();
genome.Deserialize(Path.Combine(wholeGenomeFastaDirectory, "GenomeSize.xml"));
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
writer.WriteLine($"##contig=<ID={chromosome.Name},length={chromosome.Length}>");
}
string qualityFilter = $"q{qualityThreshold}";
writer.WriteLine("##ALT=<ID=CNV,Description=\"Copy number variable region\">");
writer.WriteLine($"##FILTER=<ID={qualityFilter},Description=\"Quality below {qualityThreshold}\">");
writer.WriteLine("##FILTER=<ID=L10kb,Description=\"Length shorter than 10kb\">");
writer.WriteLine("##INFO=<ID=SVTYPE,Number=1,Type=String,Description=\"Type of structural variant\">");
writer.WriteLine("##INFO=<ID=END,Number=1,Type=Integer,Description=\"End position of the variant described in this record\">");
writer.WriteLine("##INFO=<ID=CNVLEN,Number=1,Type=Integer,Description=\"Number of reference positions spanned by this CNV\">");
writer.WriteLine("##FORMAT=<ID=RC,Number=1,Type=Float,Description=\"Mean counts per bin in the region\">");
writer.WriteLine("##FORMAT=<ID=BC,Number=1,Type=Float,Description=\"Number of bins in the region\">");
writer.WriteLine("##FORMAT=<ID=CN,Number=1,Type=Integer,Description=\"Copy number genotype for imprecise events\">");
writer.WriteLine("##FORMAT=<ID=MCC,Number=1,Type=Integer,Description=\"Major chromosome count (equal to copy number for LOH regions)\">");
writer.WriteLine("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\t" + sampleName);
SanityCheckChromosomeNames(genome, segments);
foreach (GenomeMetadata.SequenceMetadata chromosome in genome.Sequences)
{
foreach (CanvasSegment segment in segments)
{
if (!segment.Chr.Equals(chromosome.Name, StringComparison.OrdinalIgnoreCase))
{
continue;
}
int referenceCopyNumber = ploidy?.GetReferenceCopyNumber(segment) ?? 2;
CnvType cnvType = segment.GetCnvType(referenceCopyNumber);
// From vcf 4.1 spec:
// If any of the ALT alleles is a symbolic allele (an angle-bracketed ID String “<ID>”) then the padding base is required and POS denotes the
// coordinate of the base preceding the polymorphism.
string alternateAllele = cnvType.ToAltId();
int position = (alternateAllele.StartsWith("<") && alternateAllele.EndsWith(">")) ? segment.Begin : segment.Begin + 1;
writer.Write($"{segment.Chr}\t{position}\tCanvas:{cnvType.ToVcfId()}:{segment.Chr}:{segment.Begin + 1}-{segment.End}\t");
writer.Write($"N\t{alternateAllele}\t{segment.QScore}\t{segment.Filter}\t", alternateAllele, segment.QScore, segment.Filter);
if (cnvType != CnvType.Reference)
{
writer.Write($"SVTYPE={cnvType.ToSvType()};");
}
writer.Write($"END={segment.End}");
if (cnvType != CnvType.Reference)
{
writer.Write($";CNVLEN={segment.End - segment.Begin}");
}
// FORMAT field
writer.Write("\tRC:BC:CN", segment.End);
if (segment.MajorChromosomeCount.HasValue)
{
writer.Write(":MCC");
}
writer.Write("\t{1}:{2}:{3}", segment.End, Math.Round(segment.MeanCount, 0, MidpointRounding.AwayFromZero), segment.BinCount, segment.CopyNumber);
if (segment.MajorChromosomeCount.HasValue)
{
writer.Write(":{0}", segment.MajorChromosomeCount);
}
writer.WriteLine();
}
}
}
}