internal static bool IsRefCall([NotNull] this IVcfVariant variant, [CanBeNull] string sampleName)
{
//refsite is a refcall for sure
if (variant.IsRefSite())
{
return(true);
}
//if not refsite and no sample field, not a refcall
if (variant.Samples.Count == 0)
{
return(false);
}
var sample = sampleName == null ? variant.Samples[0] : variant.Samples[sampleName];
var isCn = sample.SampleDictionary.TryGetValue(VcfConstants.CnSampleFieldKey, out var cnString);
var isGt = sample.SampleDictionary.TryGetValue(VcfConstants.GenotypeKey, out var gt);
if (isGt)
{
//todo: refining how to deal with ploidy. Also here we don't deal with LOH. assuming CN = ploidy is ref
var gtArray = gt.Split('/', '|');
if (isCn && int.TryParse(cnString, out var intCn))
{
return(intCn == gtArray.Length);
}
return(gtArray.All(alleleIndex => alleleIndex == "0"));
}
return(isCn && cnString == "2");
}
public static FailedReason?ParseFromVariant([NotNull] IVcfVariant variant, bool isCrossTypeOn,
[CanBeNull] string sampleName, [CanBeNull] out WittyerType svType)
{
if (variant.IsRefSite() || IsRefCall(out var ploidy, out var cn, out var hasCn))
{
svType = CopyNumberReference;
return(null);
}
var hasSvTypeKey = variant.Info.TryGetValue(VcfConstants.SvTypeKey, out var svTypeStr);
if (!hasSvTypeKey)
{
// todo: maybe we can allow small variants, which would not have SVTYPE
throw new InvalidDataException(
$"Following variant does not have {VcfConstants.SvTypeKey} info key:\n{variant}");
}
svType = null;
if (svTypeStr == SvTypeStrings.TranslocationBreakend)
{
// breakends can be IntraChromosomeBreakend and TranslocationBreakend, so can't tell from SVTYPE.
var mate = variant is IBreakEnd cast
? cast.Mate
: SimpleBreakEnd.TryParse(variant.GetSingleAlt()).GetOrThrow();
svType = variant.Contig.Equals(mate.Contig)
? IntraChromosomeBreakend
: TranslocationBreakend;
return(null);
}
if (!TryParseSvType(svTypeStr, out svType))
{
// Not BND because of check above, and if not parsable and not CNV, it's something we don't know.
if (svTypeStr != SvTypeStrings.Cnv)
{
throw new InvalidDataException($"Cannot recognize SVTYPE of {svTypeStr}");
}
}
else if (!svType.HasBaseLevelStats)
{
// If INV or INS or whatever that doesn't need to look for CN, return.
return(null);
}
if (!hasCn)
{
return(svType == null
? FailedReason.CnvWithoutCn
: default(FailedReason?)); // DEL or DUP without CN
}
// At this point, it is CNV with CN or DEL/DUP with CN, which are also considered CNV
if (cn == null)
{
// has CN, but can't parse.
svType = null; // clear out SVTYPE=DEL/DUP
return(FailedReason.UndeterminedCn);
}
svType = GetSvType(cn.Value);
return(null);
WittyerType GetSvType(int cnValue)
=> cnValue < ploidy
? (isCrossTypeOn ? Deletion : CopyNumberLoss)
: (isCrossTypeOn ? Duplication : CopyNumberGain);
bool IsRefCall(out int ploidyP, out int?cnP, out bool hasCnP)
{
ploidyP = 2;
cnP = null;
hasCnP = false;
//if not refsite and no sample field, not a refcall
if (variant.Samples.Count == 0)
{
return(false);
}
var sample = sampleName == null ? variant.Samples[0] : variant.Samples[sampleName];
hasCnP = sample.SampleDictionary.TryGetValue(VcfConstants.CnSampleFieldKey, out var cnString);
var isGt = sample.SampleDictionary.TryGetValue(VcfConstants.GenotypeKey, out var gt);
if (hasCnP && int.TryParse(cnString, out var i))
{
cnP = i;
}
if (!isGt)
{
return(hasCnP && cnString == "2");
}
//todo: refining how to deal with ploidy. Also here we don't deal with LOH. assuming CN = ploidy is ref
var gtArray = gt.Split(VcfConstants.GtPhasedValueDelimiter[0],
VcfConstants.GtUnphasedValueDelimiter[0]);
ploidyP = gtArray.Length;
return(cnP == null?gtArray.All(alleleIndex => alleleIndex == "0") : cnP.Value == ploidyP);
}
}
