private MutableTranscript GetNextTranscript()
{
string line = _reader.ReadLine();
if (line == null)
{
return(null);
}
var transcriptInfo = ReadTranscriptInfo(line);
var gene = ReadGene(transcriptInfo.Chromosome);
var translation = ReadTranslation();
var exons = ReadExons(transcriptInfo.Chromosome);
var introns = ReadIntervals("Introns");
var cdnaMaps = ReadCdnaMaps();
var mirnas = ReadIntervals("miRNAs");
var selenocysteines = ReadSelenocysteines();
var rnaEdits = ReadRnaEdits();
var transcript = new MutableTranscript(transcriptInfo.Chromosome, transcriptInfo.Start, transcriptInfo.End,
transcriptInfo.Id, transcriptInfo.Version, transcriptInfo.CcdsId, transcriptInfo.RefSeqId,
transcriptInfo.BioType, transcriptInfo.IsCanonical, translation.CodingRegion, translation.Id,
translation.Version, translation.PeptideSeq, transcriptInfo.Source, gene, exons,
transcriptInfo.StartExonPhase, transcriptInfo.TotalExonLength, introns, cdnaMaps, null, null,
transcriptInfo.TranslateableSequence, mirnas, transcriptInfo.CdsStartNotFound,
transcriptInfo.CdsEndNotFound, selenocysteines, rnaEdits, transcriptInfo.BamEditStatus);
AddMutableContents(transcript);
return(transcript);
}
internal void Write(MutableTranscript transcript)
{
WriteTranscriptInfo(transcript);
WriteGene(_writer, transcript.Gene);
WriteTranslation(transcript.CodingRegion, transcript.ProteinId, transcript.ProteinVersion, transcript.PeptideSequence);
WriteExons(transcript.Exons);
WriteIntervals(transcript.Introns, "Introns");
WriteCdnaMaps(transcript.CdnaMaps);
WriteIntervals(transcript.MicroRnas, "miRNAs");
WriteSelenocysteines(transcript.SelenocysteinePositions);
WriteRnaEdits(transcript.RnaEdits);
}
private static ITranscript ToTranscript(this MutableTranscript mt)
{
var translation = mt.CodingRegion == null
? null
: GetTranslation(mt.CodingRegion, mt.CdsLength, CompactId.Convert(mt.ProteinId, mt.ProteinVersion),
mt.PeptideSequence);
var sortedMicroRnas = mt.MicroRnas?.OrderBy(x => x.Start).ToArray();
return(new Transcript(mt.Chromosome, mt.Start, mt.End, CompactId.Convert(mt.Id, mt.Version), translation,
mt.BioType, mt.UpdatedGene, mt.TotalExonLength, mt.NewStartExonPhase, mt.IsCanonical,
mt.TranscriptRegions, (ushort)mt.Exons.Length, sortedMicroRnas, mt.SiftIndex, mt.PolyPhenIndex,
mt.Source, mt.CdsStartNotFound, mt.CdsEndNotFound, mt.SelenocysteinePositions, mt.RnaEdits));
}
private void AddMutableContents(MutableTranscript mt)
{
mt.TranscriptRegions = TranscriptRegionMerger.GetTranscriptRegions(mt.CdnaMaps, mt.Exons, mt.Introns, mt.Gene.OnReverseStrand);
TranscriptRegionValidater.Validate(mt.Id, mt.CdnaMaps, mt.Exons, mt.Introns, mt.TranscriptRegions);
mt.NewStartExonPhase = mt.StartExonPhase < 0 ? (byte)0 : (byte)mt.StartExonPhase;
if (mt.CodingRegion == null)
{
return;
}
var codingSequence = new CodingSequence(_sequence, mt.CodingRegion, mt.TranscriptRegions,
mt.Gene.OnReverseStrand, mt.NewStartExonPhase, mt.RnaEdits);
mt.CdsLength = codingSequence.GetCodingSequence().Length;
mt.CodingRegion = new CodingRegion(mt.CodingRegion.Start, mt.CodingRegion.End,
mt.CodingRegion.CdnaStart, mt.CodingRegion.CdnaEnd, mt.CdsLength);
}
private void WriteTranscriptInfo(MutableTranscript transcript)
{
_writer.WriteLine($"Transcript\t{transcript.Id}\t{transcript.Version}\t{transcript.Chromosome.UcscName}\t{transcript.Chromosome.Index}\t{transcript.Start}\t{transcript.End}\t{transcript.BioType}\t{(byte)transcript.BioType}\t{BoolToChar(transcript.IsCanonical)}\t{transcript.TotalExonLength}\t{transcript.CcdsId}\t{transcript.RefSeqId}\t{(byte)transcript.Source}\t{BoolToChar(transcript.CdsStartNotFound)}\t{BoolToChar(transcript.CdsEndNotFound)}\t{transcript.StartExonPhase}\t{transcript.BamEditStatus}");
_writer.WriteLine(transcript.TranslateableSequence);
}
public bool Pass(MutableTranscript transcript) => _whitelist.Any(prefix => transcript.Id.StartsWith(prefix)) && !transcript.Id.Contains("dupl");