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 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);
}
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);
}
