private static string ExtractDbId(IAnnotatedVariant annotatedVariant, string idField)
{
var dbSnp = new VcfField();
var nonDbsnpIds = GetNonDbsnpIds(idField);
if (nonDbsnpIds != null)
{
foreach (var nonDbsnpId in nonDbsnpIds)
{
dbSnp.Add(nonDbsnpId);
}
}
foreach (var altAllele in annotatedVariant.AnnotatedAlternateAlleles)
{
if (altAllele.DbSnpIds == null)
{
continue;
}
foreach (var id in altAllele.DbSnpIds)
{
dbSnp.Add(id);
}
}
return(dbSnp.GetString(""));
}
private void AddInfoField(string infoField, IAnnotatedVariant annotatedVariant, StringBuilder sb)
{
var infoEntries = new VcfField();
if (!string.IsNullOrEmpty(infoField))
{
infoEntries.Add(infoField);
}
ExtractInfo(annotatedVariant as UnifiedJson, infoEntries);
infoField = infoEntries.GetString("");
// remove .
if (infoField == ".")
{
infoField = "";
}
sb.Append(infoField);
var csqs = new List <CsqEntry>();
ExtractCsqs(annotatedVariant as UnifiedJson, csqs);
if (csqs.Count != 0)
{
if (infoField.Length > 0)
{
sb.Append(";");
}
}
// append CSQ tags using delegate from annotator
sb.Append(GetCsqtAndCsqrVcfInfo(csqs));
if (csqs.Count == 0 && infoField.Length == 0)
{
sb.Append(".");
}
}
private static void ExtractInfo(UnifiedJson unifiedJson, VcfField infoEntries)
{
var alleleFreq1000G = new VcfInfoKeyValue("AF1000G");
var ancestralAllele = new VcfInfoKeyValue("AA");
var clinVar = new VcfInfoKeyValue("clinvar");
var cosmic = new VcfInfoKeyValue("cosmic");
var evs = new VcfInfoKeyValue("EVS");
var globalMinorAllele = new VcfInfoKeyValue("GMAF");
var phyloP = new VcfInfoKeyValue("phyloP");
foreach (var jsonVariant in unifiedJson.AnnotatedAlternateAlleles)
{
if (jsonVariant.IsReferenceMinor)
{
infoEntries.Add("RefMinor");
}
phyloP.Add(jsonVariant.PhylopScore);
ancestralAllele.Add(jsonVariant.AncestralAllele);
foreach (var cosmicEntry in jsonVariant.CosmicEntries)
{
if (cosmicEntry.ID == null)
{
continue;
}
if (cosmicEntry.SaAltAllele != jsonVariant.SaAltAllele)
{
continue;
}
cosmic.Add(jsonVariant.GenotypeIndex.ToString(CultureInfo.InvariantCulture) + '|' + cosmicEntry.ID);
}
foreach (var clinVarEntry in jsonVariant.ClinVarEntries)
{
if (clinVarEntry.Significance == null)
{
continue;
}
if (clinVarEntry.SaAltAllele != jsonVariant.SaAltAllele)
{
continue;
}
clinVar.Add(jsonVariant.GenotypeIndex.ToString(CultureInfo.InvariantCulture) + '|' + RemoveWhiteSpaceAndComma(clinVarEntry.Significance));
}
if (jsonVariant.GlobalMinorAllele != null || jsonVariant.GlobalMinorAlleleFrequency != null)
{
globalMinorAllele.Add(jsonVariant.GlobalMinorAllele + '|' + jsonVariant.GlobalMinorAlleleFrequency);
}
else
{
// for multi allelic variants, we need to add a . for the entries that do not have a Global minor allele.
globalMinorAllele.Add(null);
}
alleleFreq1000G.Add(jsonVariant.AlleleFrequencyAll);
if (jsonVariant.EvsAlleleFrequencyAll != null || jsonVariant.EvsCoverage != null || jsonVariant.EvsSamples != null)
{
evs.Add(jsonVariant.EvsAlleleFrequencyAll + '|' + jsonVariant.EvsCoverage + '|' + jsonVariant.EvsSamples);
}
else
{
evs.Add(null);
}
}
infoEntries.Add(ancestralAllele.GetString());
infoEntries.Add(globalMinorAllele.GetString());
infoEntries.Add(alleleFreq1000G.GetString());
infoEntries.Add(evs.GetString());
infoEntries.Add(phyloP.GetString());
infoEntries.Add(cosmic.GetString());
infoEntries.Add(clinVar.GetString());
}