/// <summary>
/// Find the last position before 'pos' within an exon
/// </summary>
/// <param name="pos"></param>
/// <returns></returns>
private long LastExonPositionBefore(long pos)
{
long last = -1;
foreach (Exon ex in Exons.OrderBy(x => x.OneBasedStart))
{
if (pos < ex.OneBasedStart)
{
// Nothing found?
if (last < 0)
{
Console.WriteLine("WARNING: Cannot find last exonic position before " + pos + " for transcript '" + ID + "'");
return(-1);
}
return(last);
}
else if (pos <= ex.OneBasedEnd)
{
return(pos);
}
last = ex.OneBasedEnd;
}
if (last < 0)
{
Console.WriteLine("WARNING: Cannot find last exonic position before " + pos + " for transcript '" + ID + "'");
}
return(pos);
}
/// <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>
/// Find the first position after 'pos' within an exon
/// </summary>
/// <param name="pos"></param>
/// <returns></returns>
private long FirstExonPositionAfter(long pos)
{
foreach (Exon ex in Exons.OrderBy(x => x.OneBasedStart))
{
if (pos <= ex.OneBasedStart)
{
return(ex.OneBasedStart);
}
if (pos <= ex.OneBasedEnd)
{
return(pos);
}
}
Console.WriteLine("WARNING: Cannot find first exonic position after " + pos + " for transcript '" + ID + "'");
return(-1);
}
public List <MetadataListItem <List <string> > > GetFeatures()
{
var features = new List <MetadataListItem <List <string> > >();
var geneMetadata = GetGtfFeatureMetadata();
features.Add(geneMetadata);
List <Interval> exonsAndCds = Exons.OfType <Interval>().Concat(CodingDomainSequences).OrderBy(t => t.OneBasedStart).ToList(); // exons before cds; exons should come up first after stable sorting
Exon currentExon = null;
foreach (Interval xc in exonsAndCds)
{
Exon x = xc as Exon;
if (x != null)
{
currentExon = x;
features.Add(x.GetGtfFeatureMetadata());
}
else
{
features.Add(CDSFeatureMetadata(xc as CDS, currentExon));
}
}
return(features);
}
/// <summary>
/// Creates coding domains based on another annotated transcript
/// </summary>
/// <param name="withCDS"></param>
/// <returns>true if this transcript was annotated; false if the transcript with CDS did not lead to an annotation</returns>
public bool CreateCDSFromAnnotatedStartCodons(Transcript withCDS)
{
// Nothing to do if null input
if (withCDS == null)
{
return(false);
}
// Figure out the start position
CDS firstCds = withCDS.CdsSortedStrand.First();
long cdsStartInChrom = IsStrandPlus() ? firstCds.OneBasedStart : firstCds.OneBasedEnd;
long cdsStartInMrna = BaseNumber2MRnaPos(cdsStartInChrom);
if (cdsStartInMrna < 0)
{
return(false);
} // the coding start wasn't within any of the exons of this transcript
// Figure out the stop codon from translation
ISequence spliced = SplicedRNA();
ISequence translateThis = spliced.GetSubSequence(cdsStartInMrna, spliced.Count - cdsStartInMrna);
ISequence proteinSequence = Translation.OneFrameTranslation(translateThis, Gene.Chromosome.Mitochondrial);
int stopIdx = proteinSequence.Select(x => x).ToList().IndexOf(Alphabets.Protein.Ter);
if (stopIdx < 0)
{
return(false);
} // no stop codon in sight
long endInMrna = cdsStartInMrna + (stopIdx + 1) * GeneModel.CODON_SIZE - 1; // include the stop codon in CDS
long lengthInMrna = endInMrna - cdsStartInMrna + 1;
// Figure out the stop index on the chromosome
long utr5ishstart = IsStrandPlus() ? Exons.Min(x => x.OneBasedStart) : cdsStartInChrom + 1;
long utr5ishend = IsStrandPlus() ? cdsStartInChrom - 1 : Exons.Max(x => x.OneBasedEnd);
Interval utr5ish = new Interval(null, "", Source, Strand, utr5ishstart, utr5ishend);
var intervals = SortedStrand(Exons.SelectMany(x => x.Minus(utr5ish)).ToList());
long lengthSoFar = 0;
foreach (Interval y in intervals)
{
long lengthSum = lengthSoFar + y.Length();
if (lengthSum <= lengthInMrna) // add this whole interval
{
var toAdd = new CDS(this, ChromosomeID, Source, Strand, y.OneBasedStart, y.OneBasedEnd, 0);
CodingDomainSequences.Add(toAdd);
lengthSoFar += toAdd.Length();
}
else if (lengthSoFar < lengthInMrna) // chop off part of this interval
{
long chopLength = lengthSum - lengthInMrna;
long start = IsStrandPlus() ?
y.OneBasedStart :
y.OneBasedStart + chopLength;
long end = IsStrandPlus() ?
y.OneBasedEnd - chopLength :
y.OneBasedEnd;
var toAdd = new CDS(this, ChromosomeID, Source, Strand, start, end, 0);
CodingDomainSequences.Add(toAdd);
lengthSoFar += toAdd.Length();
}
}
SetRegions(this);
return(true);
}
/// <summary>
/// Return an exon intersecting 'marker' (first exon found)
/// </summary>
/// <param name="marker"></param>
/// <returns></returns>
public Exon FindExon(Interval marker)
{
return(Exons.FirstOrDefault(x => x.Intersects(marker)));
}
/// <summary>
/// Return the an exon that intersects 'pos'
/// </summary>
/// <param name="pos"></param>
/// <returns></returns>
public Exon FindExon(int pos)
{
return(Exons.FirstOrDefault(x => x.Includes(pos)));
}
