protected IEnumerable<CNVCall> GetCnvCallsFromVcf(string vcfPath)
{
using (VcfReader reader = new VcfReader(vcfPath, false))
{
foreach (VcfVariant variant in reader.GetVariants())
{
int end;
int CN = GetCopyNumber(variant, out end);
yield return new CNVCall(variant.ReferenceName, variant.ReferencePosition, end, CN);
}
}
}
/// <summary>
/// Load the somatic SNV output (from Strelka), and use it to dervie an estimate of overall purity.
/// Reference: https://ukch-confluence.illumina.com/display/collab/Estimate+purity+of+tumour+samples+using+somatic+SNVs
/// </summary>
protected double EstimatePurityFromSomaticSNVs()
{
Dictionary<string, List<CanvasSegment>> segmentsByChromosome = CanvasSegment.GetSegmentsByChromosome(Segments);
int recordCount = 0;
List<float> variantFrequencies = new List<float>();
using (VcfReader reader = new VcfReader(this.SomaticVCFPath))
{
foreach (VcfVariant variant in reader.GetVariants())
{
recordCount++;
// Skip non-PF variants:
if (variant.Filters != "PASS") continue;
// Skip everything but SNVs:
if (variant.ReferenceAllele.Length > 1 || variant.VariantAlleles.Length != 1 || variant.VariantAlleles[0].Length != 1
|| variant.VariantAlleles[0] == ".")
{
continue;
}
string refTagName = string.Format("{0}U", variant.ReferenceAllele[0]);
string altTagName = string.Format("{0}U", variant.VariantAlleles[0][0]);
string[] counts = variant.Genotypes.Last()[refTagName].Split(',');
int refCount = 0;
foreach (string bit in counts) refCount += int.Parse(bit);
int altCount = 0;
counts = variant.Genotypes.Last()[altTagName].Split(',');
foreach (string bit in counts) altCount += int.Parse(bit);
float VF = altCount / (float)(altCount + refCount);
if (VF >= 0.5) continue;
variantFrequencies.Add(VF);
}
}
Console.WriteLine(">>>Loaded {0} somatic variants; saved {1} somatic SNV frequencies", recordCount, variantFrequencies.Count);
// Don't bother providing an estimate if we have very few events:
if (variantFrequencies.Count < 100)
{
return double.NaN;
}
double mean = variantFrequencies.Average();
double estimatedPurity = Math.Min(1, mean * 2);
Console.WriteLine(">>>Estimated tumor purity of {0} from somatic SNV calls", estimatedPurity);
return estimatedPurity;
}
protected IEnumerable<CNVCall> GetCnvCallsFromVcf(string vcfPath, bool includePassingOnly)
{
using (VcfReader reader = new VcfReader(vcfPath, false))
{
foreach (VcfVariant variant in reader.GetVariants())
{
int end;
int CN = GetCopyNumber(variant, out end);
if (includePassingOnly && variant.Filters != "PASS") continue;
yield return new CNVCall(variant.ReferenceName, variant.ReferencePosition, end, CN);
}
}
}
/// <summary>
/// Load a list of all variants in a file. This is memory-intensive; don't do this for whole-genome vcf files!
/// </summary>
public static List<VcfVariant> GetAllVariantsInFile(string vcfPath)
{
List<VcfVariant> allVariants = new List<VcfVariant>();
using (VcfReader reader = new VcfReader(vcfPath))
{
foreach (VcfVariant variant in reader.GetVariants())
{
allVariants.Add(variant);
}
}
return allVariants;
}