internal static IWittyerVariant Create([NotNull] IVcfVariant baseVariant,
string sample, double percentageDistance, uint basepairDistance,
IReadOnlyList <uint> bins, WittyerVariantType svType)
{
if (svType == WittyerVariantType.Invalid)
{
throw new InvalidDataException(
$"Invalid {VcfConstants.SvTypeKey} in variant: \n{baseVariant}\nNot sure why you got here though. Check with a witty.er developer!");
}
var end = baseVariant.Position + baseVariant.GetSvLength();
var baseInterval = BedInterval.Create(baseVariant.Position, end);
var borderInterval =
baseVariant.Position.CalculateBorderInterval(baseInterval,
baseVariant.ParseCi(WittyerConstants.Cipos),
percentageDistance, basepairDistance);
// wittyerVariant should all have end border, it's a matter of how to find it,
// either END key in INFO field, sort out through SVLEN or other ways, details can be defined in FindEndBorder() later
var endInterval = (end - 1).CalculateBorderInterval(baseInterval,
baseVariant.ParseCi(WittyerConstants.Ciend),
percentageDistance, basepairDistance);
var posContigAndInterval =
ContigAndInterval.Create(baseVariant.Contig, borderInterval.Start, borderInterval.Stop + 1);
var endContigAndInterval =
ContigAndInterval.Create(baseVariant.Contig, endInterval.Start, endInterval.Stop + 1);
return(Create(baseVariant, baseInterval, svType, posContigAndInterval,
Winner.Create(svType, baseInterval, bins), new List <OverlapAnnotation>(),
WittyerSample.CreateOverall(baseVariant, sample, svType == WittyerVariantType.CopyNumberReference),
endContigAndInterval));
}
internal static IWittyerBnd Create([NotNull] IVcfVariant variant,
IVcfVariant secondVariant, [CanBeNull] string sampleName,
double percentageDistance, uint basepairDistance, IReadOnlyList <uint> bins)
{
var(first, second) = MiscUtils.FindBndEntriesOrder(variant, secondVariant);
var posInterval = first.CalculateBndBorderInterval(second,
first.ParseCi(WittyerConstants.Cipos), percentageDistance, basepairDistance);
var endInterval = second.CalculateBndBorderInterval(first,
second.ParseCi(WittyerConstants.Cipos), percentageDistance,
basepairDistance);
var svType = variant.ParseWittyerVariantType(sampleName);
var winner = GetWinner();
var overlapInfo = new List <OverlapAnnotation>();
var sample = WittyerSample.CreateOverall(variant, sampleName, false);
return(Create(first, posInterval, winner, endInterval, overlapInfo, sample, second, svType));
Winner GetWinner()
{
if (svType == WittyerVariantType.TranslocationBreakend)
{
return(Winner.Create(svType));
}
if (svType != WittyerVariantType.Insertion)
{
return(Winner.Create(svType, BedInterval.Create(first.Position, second.Position + 1), bins));
}
uint?end = null;
// insertion, try sequences first
if (variant.IsSimpleSequence(out var length))
{
end = length;
}
// try svlength, but if not, assume unknown length.
else if (variant.TryGetSvLength(out length) == null)
{
end = length;
}
return(Winner.Create(svType, end == null ? null : BedInterval.Create(variant.Position, variant.Position + end.Value), bins));
}
}
internal static IWittyerVariant Create([NotNull] IVcfVariant baseVariant,
[CanBeNull] IVcfSample sample, [NotNull] WittyerType svType,
[NotNull] IReadOnlyList <uint> bins, [CanBeNull] double?percentageDistance,
uint basepairDistance)
{
// originalInterval is needed to adjust CIPOS and CIEND against for PD/BPD, but it won't be used for actual reflen and binning.
var baseInterval = baseVariant.ToBedInterval(true, out var originalEnd, out var sharedFirstBase);
if (baseInterval == null)
{
throw new InvalidOperationException(
$"Expected failure of {nameof(WittyerUtils.ToBedInterval)} to throw, but didn't...");
}
// CI intervals are always based on the original POS/END
var posStart = baseVariant.Position;
if (sharedFirstBase)
{
posStart++;
}
var ciPosInterval = posStart.ConvertPositionToCiInterval(baseVariant, WittyerConstants.Cipos);
var ciEndInterval = originalEnd.ConvertPositionToCiInterval(baseVariant, WittyerConstants.Ciend);
var baseStart = sharedFirstBase
? baseInterval.Start
: baseInterval.Start + 1; // not sharing first base (ref site or complex types, etc) need adjustment
// the pd/bpd intervals are based on the trimmed variant's coordinates.
var(posInterval, endInterval) = WittyerUtils.GetPosAndEndInterval(baseVariant.Contig, percentageDistance,
basepairDistance, ciPosInterval, baseStart, ciEndInterval, baseInterval.Stop);
return(new WittyerVariantInternal(svType, baseVariant, baseInterval,
Winner.Create(svType, baseInterval, bins),
posInterval, ciPosInterval, endInterval, ciEndInterval,
WittyerSample.CreateFromVariant(baseVariant, sample, svType == WittyerType.CopyNumberReference)));
}
internal static IWittyerBnd Create([NotNull] IVcfVariant first, [CanBeNull] IVcfSample originalSample,
[NotNull] WittyerType svType, [NotNull] IReadOnlyList <uint> bins, uint bpd, double?percentageDistance,
[NotNull] IVcfVariant second)
{
if (!ReferenceEquals(first, second))
{
(first, second) = FindBndEntriesOrder(in first, in second);
}
var ciPosInterval = first.Position.ConvertPositionToCiInterval(first, WittyerConstants.Cipos);
var ciEndInterval = ReferenceEquals(first, second)
? ciPosInterval // same variant means same intervals.
: second.Position.ConvertPositionToCiInterval(second, WittyerConstants.Cipos);
IContigAndInterval posInterval, endInterval;
if (ReferenceEquals(first, second)) // insertions need trimming and stuff.
{
var trimmed = first.TryNormalizeVariant(VariantNormalizer.TrimCommonBases, 0).GetOrThrow();
var tuple = (bpd, bpd);
var(posStart, posStop) = trimmed.Position.ConvertPositionToCiInterval(tuple);
WittyerUtils.GetBetterInterval(ciPosInterval, ref posStart, ref posStop);
posInterval = endInterval = ContigAndInterval.Create(first.Contig, posStart, posStop);
}
else
{
(posInterval, endInterval) = WittyerUtils.GetPosAndEndInterval(first.Contig,
svType == WittyerType.IntraChromosomeBreakend ? percentageDistance : null, bpd,
ciPosInterval, first.Position, ciEndInterval, second.Position, second.Contig);
}
var winner = GetWinner();
var sample = WittyerSample.CreateFromVariant(first, originalSample, false);
return(new WittyerBndInternal(svType, first, posInterval, ciPosInterval,
second, endInterval, ciEndInterval, winner, sample));
(IVcfVariant first, IVcfVariant second) FindBndEntriesOrder(in IVcfVariant variantA,
in IVcfVariant variantB)
=> ContigAndPositionComparer.Default.Compare(variantA, variantB) > 0
? (variantB, variantA)
: (variantA, variantB);
Winner GetWinner()
{
if (svType == WittyerType.TranslocationBreakend)
{
return(Winner.Create(svType));
}
IInterval <uint> bedInterval;
if (svType == WittyerType.Insertion)
{
bedInterval = GetInsertionInterval(first);
}
else
{
var start = first.Position;
if (start > 0)
{
start--;
}
bedInterval = BedInterval.Create(start, second.Position);
}
return(Winner.Create(svType, bedInterval, bins));
}
}