/// <summary>
/// For each group in the grouping, this method traverses through each range
/// in the group and normalizes the ranges down to the minimal spanning set
/// required to still show the same range spans.
///
/// For instance if you had in group 'Chr1' the following ranges:
///
/// -> 10 to 100
/// -> 200 to 250
/// -> 35 to 45
/// -> 90 to 150
///
/// The result of MergeOverlaps would reduce the ranges in the 'Chr1' group to:
/// For minOverlap = 0
///
/// -> 10 to 150
/// -> 200 to 250
///
/// for minOverlap = -50
///
/// -> 10 to 250
///
/// Running this method creates all new ISequenceRange objects and adds them
/// to the newly created SequenceRangeGrouping returned here.
/// </summary>
/// <param name="minOverlap">Minmum length of bases pairs should be overlapped.</param>
/// <param name="isParentSeqRangesRequired">If this flag is set to true then the sequence ranges from
/// which the new sequence range is created are added to the ParentSeqRanges property of the
/// new sequence range.</param>
public SequenceRangeGrouping MergeOverlaps(long minOverlap = 0, bool isParentSeqRangesRequired = false)
{
SequenceRangeGrouping seqRangeGroup = new SequenceRangeGrouping();
List <ISequenceRange> sortedRanges = new List <ISequenceRange>();
foreach (List <ISequenceRange> rangeList in this._groups.Values)
{
sortedRanges.AddRange(rangeList);
sortedRanges.Sort();
while (sortedRanges.Count > 0)
{
ISequenceRange seqRange = new SequenceRange(sortedRanges[0].ID,
sortedRanges[0].Start,
sortedRanges[0].End);
if (isParentSeqRangesRequired)
{
AddParent(seqRange, sortedRanges[0]);
}
sortedRanges.RemoveAt(0);
seqRangeGroup.Add(seqRange);
if (sortedRanges.Count > 0)
{
while (sortedRanges.Count > 0 && (seqRange.End - sortedRanges[0].Start) >= minOverlap)
{
seqRange.End = Math.Max(seqRange.End, sortedRanges[0].End);
if (isParentSeqRangesRequired)
{
AddParent(seqRange, sortedRanges[0]);
}
sortedRanges.RemoveAt(0);
}
}
}
sortedRanges.Clear();
}
return(seqRangeGroup);
}
/// <summary>
/// Subtracts the query SequenceRangeGrouping from this SequenceRangeGrouping.
///
/// For example,
///
/// Ranges in this instance Ranges in the query
/// 1 to 4 2 to 6
/// 4 to 8 3 to 6
/// 8 to 12 9 to 14
/// 25 to 35
///
/// Result for minOverlap set to 1
/// 1. If outputType is IntervalsWithNoOverlap
/// 25 to 35
///
/// 2. If outputType is NonOverlappingPiecesOfIntervals
/// 1 to 2
/// 6 to 8
/// 8 to 9
/// 25 to 35
///
/// Running this method creates all new ISequenceRange objects and adds them
/// to the newly created SequenceRangeGrouping returned here.
/// </summary>
/// <param name="query">Query grouping.</param>
/// <param name="minOverlap">Minmum length of overlap. By default this will be set to 1</param>
/// <param name="outputType">
/// Type of output required, IntervalsWithNoOverlap or NonOverlappingPiecesOfIntervals.
/// By default this will be set to NonOverlappingPiecesOfIntervals that is non overlapping
/// pieces of intervels along with non overlapping ranges from this instance
/// will be returned.
/// </param>
/// <param name="isParentSeqRangesRequired">If this flag is set to true then the sequence ranges from
/// which the new sequence range is created are added to the ParentSeqRanges property of the
/// new sequence range.</param>
public SequenceRangeGrouping Subtract(SequenceRangeGrouping query,
long minOverlap = 1,
SubtractOutputType outputType = SubtractOutputType.NonOverlappingPiecesOfIntervals,
bool isParentSeqRangesRequired = false)
{
SequenceRangeGrouping result = new SequenceRangeGrouping();
List <ISequenceRange> refSeqRanges = new List <ISequenceRange>();
List <ISequenceRange> querySeqRanges = new List <ISequenceRange>();
List <ISequenceRange> previousSeqRanges = new List <ISequenceRange>();
SequenceRange range = null;
// merge the query sequence ranges.
IList <ISequenceRange> queryList = null;
if (isParentSeqRangesRequired)
{
queryList = query.Flatten();
}
query = query.MergeOverlaps(0, isParentSeqRangesRequired);
foreach (string id in _groups.Keys)
{
refSeqRanges.Clear();
querySeqRanges.Clear();
refSeqRanges.AddRange(_groups[id]);
if (query._groups.ContainsKey(id))
{
querySeqRanges.AddRange(query._groups[id]);
querySeqRanges.Sort();
}
if (querySeqRanges.Count > 0)
{
foreach (ISequenceRange refRange in refSeqRanges)
{
previousSeqRanges.Clear();
IList <ISequenceRange> overlappingQueryRanges = GetOverlappingRenges(refRange,
querySeqRanges,
minOverlap);
if (overlappingQueryRanges == null || overlappingQueryRanges.Count == 0)
{
if (minOverlap > 0 || outputType == SubtractOutputType.NonOverlappingPiecesOfIntervals)
{
range = new SequenceRange(refRange.ID, refRange.Start, refRange.End);
CopyOfMetadata(range, refRange);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
result.Add(range);
}
continue;
}
// no need to proceed if only non overlapping intervels needed.
if (outputType == SubtractOutputType.IntervalsWithNoOverlap)
{
continue;
}
ISequenceRange previousOverlappingRange = null;
foreach (ISequenceRange queryRange in overlappingQueryRanges)
{
// in case of non overlapping pieces of intervals get the non overlapping
// ranges from reference sequence range.
if (refRange.Start < queryRange.Start)
{
if (previousSeqRanges.Count > 0 && previousSeqRanges[0].Start < queryRange.Start)
{
// if the previous overlapping range's start and end are equal then no need to change the metadataSeqRanges.
if (previousOverlappingRange == null || previousOverlappingRange.Start != queryRange.Start && previousOverlappingRange.End != queryRange.End)
{
for (int i = previousSeqRanges.Count - 1; i >= 0; i--)
{
if (previousSeqRanges[i].End > queryRange.Start)
{
previousSeqRanges[i].End = queryRange.Start;
}
else if (previousSeqRanges[i].End < queryRange.Start)
{
previousSeqRanges.RemoveAt(i);
}
}
}
if (isParentSeqRangesRequired)
{
if (queryList.Contains(queryRange))
{
AddParent(previousSeqRanges[0], queryRange);
}
else
{
if (queryRange.ParentSeqRanges.Count > 0)
{
AddParent(previousSeqRanges[0], queryRange.ParentSeqRanges);
}
}
}
}
else
{
if (previousOverlappingRange == null || previousOverlappingRange.Start != queryRange.Start && previousOverlappingRange.End != queryRange.End)
{
range = new SequenceRange(refRange.ID, refRange.Start, queryRange.Start);
result.Add(range);
CopyOfMetadata(range, refRange);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
}
if (isParentSeqRangesRequired)
{
if (queryList.Contains(queryRange))
{
AddParent(range, queryRange);
}
else
{
if (queryRange.ParentSeqRanges.Count > 0)
{
AddParent(range, queryRange.ParentSeqRanges);
}
}
}
}
}
if (queryRange.End < refRange.End)
{
if (previousOverlappingRange == null || previousOverlappingRange.Start != queryRange.Start && previousOverlappingRange.End != queryRange.End)
{
range = new SequenceRange(refRange.ID, queryRange.End, refRange.End);
CopyOfMetadata(range, refRange);
result.Add(range);
previousSeqRanges.Add(range);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
}
if (isParentSeqRangesRequired)
{
if (queryList.Contains(queryRange))
{
AddParent(range, queryRange);
}
else
{
if (queryRange.ParentSeqRanges.Count > 0)
{
AddParent(range, queryRange.ParentSeqRanges);
}
}
}
}
previousOverlappingRange = queryRange;
}
}
}
else
{
foreach (SequenceRange refRange in refSeqRanges)
{
range = new SequenceRange(refRange.ID, refRange.Start, refRange.End);
CopyOfMetadata(range, refRange);
result.Add(range);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
}
}
}
return(result);
}
/// <summary>
/// Returns overlapping sequence ranges from this and specified SequenceRangeGroup for each group in this grouping.
///
/// For instance if you had in group 'Chr1' the following ranges:
///
/// Ranges in this instance Ranges in the query
/// 0 to 10 20 to 40
/// 30 to 50 70 to 100
/// 60 to 80 400 to 800
/// 300 to 500 850 to 900
/// 600 to 700 900 to 1200
/// 800 to 1000
///
/// Result for minOverlap set to 1
/// 1. If outputType is OverlappingPiecesOfIntervals.
/// 30 to 40
/// 70 to 80
/// 400 to 500
/// 600 o 700
/// 850 to 900
/// 900 to 1000
/// 2. If outputType is OverlappingIntervals
/// 30 to 50
/// 60 to 80
/// 300 to 500
/// 600 to 700
/// 800 to 1000
///
/// Running this method creates all new ISequenceRange objects and adds them
/// to the newly created SequenceRangeGrouping returned here.
/// </summary>
/// <param name="query">Query grouping.</param>
/// <param name="minOverlap">Minmum length of bases pairs should be overlapped.
/// By default this will be set to 1.</param>
/// <param name="outputType">
/// Type of output required, OverlappingPiecesOfIntervals or OverlappingIntervals.
/// By default this will be set to OverlappingPiecesOfIntervals that is only the base pairs that overlaps with
/// query ranges will be returned.</param>
/// <param name="isParentSeqRangesRequired">If this flag is set to true then the sequence ranges from
/// which the new sequence range is created are added to the ParentSeqRanges property of the
/// new sequence ranges.</param>
public SequenceRangeGrouping Intersect(SequenceRangeGrouping query, long minOverlap = 1, IntersectOutputType outputType = IntersectOutputType.OverlappingPiecesOfIntervals, bool isParentSeqRangesRequired = false)
{
SequenceRangeGrouping result = new SequenceRangeGrouping();
List <ISequenceRange> refSeqRanges = new List <ISequenceRange>();
List <ISequenceRange> querySeqRanges = new List <ISequenceRange>();
SequenceRange range = null;
// merge the query sequence ranges.
IList <ISequenceRange> queryList = null;
if (isParentSeqRangesRequired)
{
queryList = query.Flatten();
}
query = query.MergeOverlaps(0, isParentSeqRangesRequired);
foreach (string id in _groups.Keys)
{
refSeqRanges.Clear();
querySeqRanges.Clear();
refSeqRanges.AddRange(_groups[id]);
if (query._groups.ContainsKey(id))
{
querySeqRanges.AddRange(query._groups[id]);
querySeqRanges.Sort();
}
if (querySeqRanges.Count > 0)
{
foreach (ISequenceRange refRange in refSeqRanges)
{
IList <ISequenceRange> overlappingQueryRanges = GetOverlappingRenges(refRange, querySeqRanges, minOverlap);
if (overlappingQueryRanges == null || overlappingQueryRanges.Count == 0)
{
// If the minOverlap is lessthan or equal to zero and overlapping intervals are required.
// then add the ref seq to result.
if (minOverlap <= 0 && outputType == IntersectOutputType.OverlappingIntervals)
{
range = new SequenceRange(refRange.ID, refRange.Start, refRange.End);
CopyOfMetadata(range, refRange);
result.Add(range);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
}
continue;
}
ISequenceRange previousOverlappingRange = null;
foreach (ISequenceRange queryRange in overlappingQueryRanges)
{
if (outputType == IntersectOutputType.OverlappingPiecesOfIntervals)
{
// Add ref sequence only once for query ranges having same start and end.
if (previousOverlappingRange == null || (previousOverlappingRange.Start != queryRange.Start && previousOverlappingRange.End != queryRange.End))
{
range = new SequenceRange(
refRange.ID,
Math.Max(queryRange.Start, refRange.Start),
Math.Min(queryRange.End, refRange.End));
result.Add(range);
CopyOfMetadata(range, refRange);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
}
if (isParentSeqRangesRequired)
{
if (queryList.Contains(queryRange))
{
AddParent(range, queryRange);
}
else
{
if (queryRange.ParentSeqRanges.Count > 0)
{
AddParent(range, queryRange.ParentSeqRanges);
}
}
}
}
else
{
// Add ref sequence only once.
if (previousOverlappingRange == null)
{
range = new SequenceRange(refRange.ID, refRange.Start, refRange.End);
CopyOfMetadata(range, refRange);
result.Add(range);
if (isParentSeqRangesRequired)
{
AddParent(range, refRange);
}
else
{
break;
}
}
if (isParentSeqRangesRequired)
{
if (queryList.Contains(queryRange))
{
AddParent(range, queryRange);
}
else
{
if (queryRange.ParentSeqRanges.Count > 0)
{
AddParent(range, queryRange.ParentSeqRanges);
}
}
}
}
previousOverlappingRange = queryRange;
}
}
}
}
return(result);
}