/// <summary>
/// Fix coding domain sequences that have end frames
/// </summary>
/// <returns></returns>
public void FrameCorrection()
{
// No coding domains? Nothing to do
if (CdsSortedStrand == null || CdsSortedStrand.Count == 0)
{
return;
}
CDS cdsFirst = CdsSortedStrand.First();
if (cdsFirst != null)
{
UTR5Prime utr = cdsFirst.StartFrameCorrection();
if (utr != null)
{
UTRs.Add(utr);
}
}
CDS cdsLast = CdsSortedStrand.Last();
if (cdsLast != null)
{
UTR3Prime utr = cdsLast.EndFrameCorrection(RetrieveCodingSequence().Count);
if (utr != null)
{
UTRs.Add(utr);
}
}
_CodingSequence = null; // update this later after this frame update
}
/// <summary>
/// Create UTR regions for this transcript
/// </summary>
public List <UTR> CreateUTRs()
{
if (CodingDomainSequences.Count == 0)
{
return(UTRs);
}
List <Interval> missing = Exons.OfType <Interval>().ToList();
foreach (Interval interval in UTRs.Concat(CodingDomainSequences.OfType <Interval>().ToList()))
{
missing = missing.SelectMany(i => i.Minus(interval)).ToList();
}
long codingMin = CodingDomainSequences.Select(c => c.OneBasedStart).Min();
long codingMax = CodingDomainSequences.Select(c => c.OneBasedEnd).Max();
foreach (Interval interval in missing)
{
Exon x = FindExon(interval);
if (x == null)
{
throw new ArgumentException("Cannot find exon for UTR: " + interval.ToString());
}
UTR toAdd = null;
if (IsStrandPlus())
{
if (interval.OneBasedEnd <= codingMin)
{
toAdd = new UTR5Prime(x, x.ChromosomeID, x.Source, x.Strand, interval.OneBasedStart, interval.OneBasedEnd);
}
else if (interval.OneBasedStart >= codingMax)
{
toAdd = new UTR3Prime(x, x.ChromosomeID, x.Source, x.Strand, interval.OneBasedStart, interval.OneBasedEnd);
}
}
else
{
if (interval.OneBasedStart >= codingMax)
{
toAdd = new UTR5Prime(x, x.ChromosomeID, x.Source, x.Strand, interval.OneBasedStart, interval.OneBasedEnd);
}
else if (interval.OneBasedEnd <= codingMin)
{
toAdd = new UTR3Prime(x, x.ChromosomeID, x.Source, x.Strand, interval.OneBasedStart, interval.OneBasedEnd);
}
}
// OK?
if (toAdd != null)
{
UTRs.Add(toAdd);
}
}
return(UTRs);
}
/// <summary>
/// Corrrects coordinates of end of coding sequence based on coding sequence length
/// </summary>
/// <param name="codingSequenceLength"></param>
/// <returns></returns>
public UTR3Prime EndFrameCorrection(long codingSequenceLength)
{
long endFrame = codingSequenceLength % 3;
if (endFrame <= 0)
{
return(null);
} // nothing to do
// First exon is not zero? => Create a UTR5 prime to compensate
Exon parent = (FindParent(typeof(Transcript)) as Transcript).GetLastCodingExon();
UTR3Prime utr3 = null;
if (IsStrandPlus())
{
long start = OneBasedEnd - (endFrame - 1);
utr3 = new UTR3Prime(parent, parent.ChromosomeID, parent.Source, parent.Strand, start, OneBasedEnd);
}
else
{
long end = OneBasedStart + (endFrame - 1);
utr3 = new UTR3Prime(parent, parent.ChromosomeID, parent.Source, parent.Strand, OneBasedStart, end);
}
// correct start or end coordinates
if (IsStrandPlus())
{
OneBasedEnd -= endFrame;
}
else
{
OneBasedStart += endFrame;
}
return(utr3);
}
public UTR3Prime(UTR3Prime utr)
: base(utr)
{
}