private void AddIupacVariants(ClinvarVariant variant, List <ClinvarVariant> variantList)
{
foreach (char altAllele in _iupacBases[variant.AltAllele[0]])
{
variantList.Add(new ClinvarVariant(variant.Chromosome, variant.Start, variant.Stop, variant.VariantId, variant.RefAllele, altAllele.ToString()));
}
}
private void AddIupacVariants(ClinvarVariant variant)
{
foreach (var altAllele in _iupacBases[variant.AltAllele[0]])
{
_variantList.Add(new ClinvarVariant(variant.Chromosome, variant.Start, variant.Stop, variant.ReferenceAllele, altAllele.ToString()));
}
}
private static bool IsInvalidVariant(ClinvarVariant variant)
{
if (variant.VariantType == "ALU")
{
return(true);
}
return(variant.Chromosome == null ||
(variant.VariantType == "Microsatellite" || variant.VariantType == "Variation") &&
string.IsNullOrEmpty(variant.AltAllele));
}
private static ClinvarVariant GenerateRefAllele(ClinvarVariant variant, ICompressedSequence compressedSequence)
{
if (variant == null)
{
return(null);
}
var extractedRef = compressedSequence.Substring(variant.Start - 1, variant.Stop - variant.Start + 1);
return(new ClinvarVariant(variant.Chromosome, variant.Start, variant.Stop, extractedRef, variant.AltAllele ?? ""));
}
private HashSet <string> GetOmimIds(ClinvarVariant variant)
{
var extendedOmimIds = new HashSet <string>(_omimIDs);
foreach (var omimId in variant.AllelicOmimIds)
{
extendedOmimIds.Add(omimId);
}
return(extendedOmimIds);
}
private ClinvarVariant LeftShift(ClinvarVariant variant)
{
if (variant.ReferenceAllele == null || variant.AltAllele == null)
{
return(variant);
}
var alignedVariant = _aligner.LeftAlign(variant.Start, variant.ReferenceAllele, variant.AltAllele);
if (alignedVariant == null)
{
return(variant);
}
return(new ClinvarVariant(variant.Chromosome, alignedVariant.Item1, variant.Stop, alignedVariant.Item2, alignedVariant.Item3));
}
private static void UpdateVariantType(ClinvarVariant variant)
{
var refAllele = variant.RefAllele;
var altAllele = variant.AltAllele;
if (refAllele == null || altAllele == null)
{
return;
}
var variantType = SmallVariantCreator.GetVariantType(refAllele, altAllele);
switch (variantType)
{
case VariantType.deletion:
variant.VariantType = "Deletion";
break;
case VariantType.insertion:
variant.VariantType = "Insertion";
break;
case VariantType.indel:
variant.VariantType = "Indel";
break;
case VariantType.duplication:
variant.VariantType = "Duplication";
break;
case VariantType.SNV:
variant.VariantType = "SNV";
break;
case VariantType.MNV:
variant.VariantType = "MNV";
break;
}
}
private List <ClinVarItem> ExtractClinVarItems(LiteXmlElement xmlElement)
{
ClearClinvarFields();
if (xmlElement == null)
{
return(null);
}
if (xmlElement.IsEmpty())
{
return(null);
}
foreach (var child in xmlElement.Children)
{
switch (child.Name)
{
case "ReferenceClinVarAssertion":
ParseRefClinVarAssertion(child);
break;
case "ClinVarAssertion":
ParseScv(child);
break;
}
}
if (_recordStatus != "current")
{
Console.WriteLine($"record status not current: {_recordStatus} for {_id}");
return(null);
}
var clinvarList = new List <ClinVarItem>();
foreach (var variant in _variantList)
{
// in order to match the VCF, we leave out the ones that do not have dbsnp id
if (variant.DbSnp == null)
{
continue;
}
if (!InputFileParserUtilities.IsDesiredChromosome(variant.Chromosome, _compressedSequence.Renamer))
{
continue;
}
if (variant.VariantType == "Microsatellite")
{
continue;
}
var refIndex = _compressedSequence.Renamer.GetReferenceIndex(variant.Chromosome);
if (refIndex == ChromosomeRenamer.UnknownReferenceIndex)
{
throw new GeneralException($"Could not find the reference index for: {variant.Chromosome}");
}
_dataFileManager.LoadReference(refIndex, () => {});
ClinvarVariant shiftedVariant = variant;
//some entries do not have ref allele in the xml file. For those, we extract them from our ref sequence
if (variant.ReferenceAllele == null && variant.VariantType == "Deletion")
{
shiftedVariant = GenerateRefAllele(variant, _compressedSequence);
}
if (variant.AltAllele == null && variant.VariantType == "Duplication")
{
shiftedVariant = GenerateAltAllele(variant, _compressedSequence);
}
//left align the variant
shiftedVariant = LeftShift(shiftedVariant);
if (variant.ReferenceAllele == null && variant.VariantType == "Indel" && variant.AltAllele != null)
{
shiftedVariant = GenerateRefAllele(variant, _compressedSequence);
}
_pubMedIds.Sort();
if (string.IsNullOrEmpty(shiftedVariant.ReferenceAllele) && string.IsNullOrEmpty(shiftedVariant.AltAllele))
{
continue;
}
clinvarList.Add(
new ClinVarItem(shiftedVariant.Chromosome,
shiftedVariant.Start,
_alleleOrigins.Distinct().ToList(),
shiftedVariant.AltAllele ?? "",
_id,
_reviewStatus,
_medGenIDs.Distinct().ToList(),
_omimIDs.Distinct().ToList(),
_orphanetIDs.Distinct().ToList(),
_prefPhenotypes.Count > 0? _prefPhenotypes.Distinct().ToList(): _altPhenotypes.Distinct().ToList(),
shiftedVariant.ReferenceAllele ?? "",
_significance,
_pubMedIds.Distinct().ToList(),
_lastUpdatedDate));
}
return(clinvarList.Count > 0 ? clinvarList: null);
}