private static ExitCodes ProgramExecution()
{
var referenceProvider = new ReferenceSequenceProvider(FileUtilities.GetReadStream(_compressedReference));
TranscriptCacheData transcriptData;
using (var transcriptCacheReader = new TranscriptCacheReader(FileUtilities.GetReadStream(CacheConstants.TranscriptPath(_transcriptCachePrefix))))
{
transcriptData = transcriptCacheReader.Read(referenceProvider.RefIndexToChromosome);
}
var spliceIntervals = SpliceUtilities.GetSpliceIntervals(referenceProvider, transcriptData);
var nirEnstToGeneSymbols = SpliceUtilities.GetEnstToGeneSymbols(referenceProvider, transcriptData);
Dictionary <string, string> spliceAiEnstToGeneSymbols;
using (var reader = new StreamReader(GZipUtilities.GetAppropriateReadStream(_geneInfoFile)))
{
spliceAiEnstToGeneSymbols = SpliceUtilities.GetSpliceAiGeneSymbols(reader);
}
var spliceAiToNirvanaGeneSymbols =
SpliceUtilities.GetSymbolMapping(spliceAiEnstToGeneSymbols, nirEnstToGeneSymbols);
Console.WriteLine($"Mapped {spliceAiToNirvanaGeneSymbols.Count} spliceAI gene symbols to Nirvana gene symbols (out of {spliceAiEnstToGeneSymbols.Count})");
var version = DataSourceVersionReader.GetSourceVersion(_inputFile + ".version");
string outFileName = $"{version.Name}_{version.Version}";
using (var spliceAiParser = new SpliceAiParser(
GZipUtilities.GetAppropriateReadStream(_inputFile),
referenceProvider, spliceIntervals, spliceAiToNirvanaGeneSymbols))
using (var nsaStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.SaFileSuffix)))
using (var indexStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.SaFileSuffix + SaCommon.IndexSufix)))
using (var nsaWriter = new NsaWriter(nsaStream, indexStream, version, referenceProvider, SaCommon.SpliceAiTag, true, true, SaCommon.SchemaVersion, false))
{
nsaWriter.Write(spliceAiParser.GetItems());
}
Console.WriteLine($"Total number of entries from Splice AI: {SpliceAiParser.Count}");
return(ExitCodes.Success);
}
private static ExitCodes ProgramExecution()
{
var referenceProvider = new ReferenceSequenceProvider(FileUtilities.GetReadStream(_compressedReference));
TranscriptCacheData transcriptData;
using (var transcriptCacheReader = new TranscriptCacheReader(FileUtilities.GetReadStream(CacheConstants.TranscriptPath(_transcriptCachePrefix))))
{
transcriptData = transcriptCacheReader.Read(referenceProvider.RefIndexToChromosome);
}
var spliceIntervals = SpliceUtilities.GetSpliceIntervals(referenceProvider, transcriptData);
var nirvanaGeneForest = SpliceUtilities.GetGeneForest(transcriptData);
Console.WriteLine("Loaded transcripts and generated splice intervals.");
Dictionary <string, List <string> > geneSymbolSynonyms;
using (var geneInfoParser = new GeneInfoParser(GZipUtilities.GetAppropriateStreamReader(_geneInfoFile)))
{
geneSymbolSynonyms = geneInfoParser.GetGeneSymbolSynonyms();
}
Console.WriteLine("Loaded gene symbol synonyms");
var version = DataSourceVersionReader.GetSourceVersion(_inputFile + ".version");
string outFileName = $"{version.Name}_{version.Version}";
using (var spliceAiParser = new SpliceAiParser(
GZipUtilities.GetAppropriateReadStream(_inputFile),
referenceProvider, spliceIntervals, nirvanaGeneForest, geneSymbolSynonyms))
using (var nsaStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.SaFileSuffix)))
using (var indexStream = FileUtilities.GetCreateStream(Path.Combine(_outputDirectory, outFileName + SaCommon.SaFileSuffix + SaCommon.IndexSufix)))
{
var nsaWriter = new NsaWriter(new ExtendedBinaryWriter(nsaStream), new ExtendedBinaryWriter(indexStream), version, referenceProvider, SaCommon.SpliceAiTag, true, true, SaCommon.SchemaVersion, false);
nsaWriter.Write(spliceAiParser.GetItems());
}
Console.WriteLine($"Total number of entries from Splice AI: {SpliceAiParser.Count}");
return(ExitCodes.Success);
}