public static (Dictionary <string, string> EntrezGeneIdToSymbol, Dictionary <string, string> EnsemblIdToSymbol) ParseUniversalGeneArchive(string inputReferencePath, string universalGeneArchivePath)
{
IDictionary <string, IChromosome> refNameToChromosome;
if (inputReferencePath == null)
{
refNameToChromosome = null;
}
else
{
(_, refNameToChromosome, _) = SequenceHelper.GetDictionaries(inputReferencePath);
}
UgaGene[] genes;
using (var reader = new UgaGeneReader(GZipUtilities.GetAppropriateReadStream(universalGeneArchivePath),
refNameToChromosome))
{
genes = reader.GetGenes();
}
var entrezGeneIdToSymbol = genes.GetGeneIdToSymbol(x => x.EntrezGeneId);
var ensemblIdToSymbol = genes.GetGeneIdToSymbol(x => x.EnsemblId);
return(entrezGeneIdToSymbol, ensemblIdToSymbol);
}
private static ExitCodes ProgramExecution()
{
var logger = new ConsoleLogger();
string transcriptPath = _inputPrefix + ".transcripts.gz";
string siftPath = _inputPrefix + ".sift.gz";
string polyphenPath = _inputPrefix + ".polyphen.gz";
string regulatoryPath = _inputPrefix + ".regulatory.gz";
(var refIndexToChromosome, var refNameToChromosome, int numRefSeqs) = SequenceHelper.GetDictionaries(_inputReferencePath);
using (var transcriptReader = new MutableTranscriptReader(GZipUtilities.GetAppropriateReadStream(transcriptPath), refIndexToChromosome))
using (var regulatoryReader = new RegulatoryRegionReader(GZipUtilities.GetAppropriateReadStream(regulatoryPath), refIndexToChromosome))
using (var siftReader = new PredictionReader(GZipUtilities.GetAppropriateReadStream(siftPath), refIndexToChromosome, IntermediateIoCommon.FileType.Sift))
using (var polyphenReader = new PredictionReader(GZipUtilities.GetAppropriateReadStream(polyphenPath), refIndexToChromosome, IntermediateIoCommon.FileType.Polyphen))
using (var geneReader = new UgaGeneReader(GZipUtilities.GetAppropriateReadStream(ExternalFiles.UniversalGeneFilePath), refNameToChromosome))
{
var genomeAssembly = transcriptReader.Header.Assembly;
var source = transcriptReader.Header.Source;
long vepReleaseTicks = transcriptReader.Header.VepReleaseTicks;
ushort vepVersion = transcriptReader.Header.VepVersion;
logger.Write("- loading universal gene archive file... ");
var genes = geneReader.GetGenes();
var geneForest = CreateGeneForest(genes, numRefSeqs, genomeAssembly);
logger.WriteLine($"{genes.Length:N0} loaded.");
logger.Write("- loading regulatory region file... ");
var regulatoryRegions = regulatoryReader.GetRegulatoryRegions();
logger.WriteLine($"{regulatoryRegions.Length:N0} loaded.");
logger.Write("- loading transcript file... ");
var transcripts = transcriptReader.GetTranscripts();
var transcriptsByRefIndex = transcripts.GetMultiValueDict(x => x.Chromosome.Index);
logger.WriteLine($"{transcripts.Length:N0} loaded.");
MarkCanonicalTranscripts(logger, transcripts);
var predictionBuilder = new PredictionCacheBuilder(logger, genomeAssembly);
var predictionCaches = predictionBuilder.CreatePredictionCaches(transcriptsByRefIndex, siftReader, polyphenReader, numRefSeqs);
logger.Write("- writing SIFT prediction cache... ");
predictionCaches.Sift.Write(FileUtilities.GetCreateStream(CacheConstants.SiftPath(_outputCacheFilePrefix)));
logger.WriteLine("finished.");
logger.Write("- writing PolyPhen prediction cache... ");
predictionCaches.PolyPhen.Write(FileUtilities.GetCreateStream(CacheConstants.PolyPhenPath(_outputCacheFilePrefix)));
logger.WriteLine("finished.");
var transcriptBuilder = new TranscriptCacheBuilder(logger, genomeAssembly, source, vepReleaseTicks, vepVersion);
var transcriptStaging = transcriptBuilder.CreateTranscriptCache(transcripts, regulatoryRegions, geneForest, numRefSeqs);
logger.Write("- writing transcript cache... ");
transcriptStaging.Write(FileUtilities.GetCreateStream(CacheConstants.TranscriptPath(_outputCacheFilePrefix)));
logger.WriteLine("finished.");
}
return(ExitCodes.Success);
}
EnsemblIdToSymbol) ParseUniversalGeneArchive()
{
var(_, refNameToChromosome, _) = SequenceHelper.GetDictionaries(_inputReferencePath);
UgaGene[] genes;
using (var reader = new UgaGeneReader(GZipUtilities.GetAppropriateReadStream(_universalGeneArchivePath),
refNameToChromosome))
{
genes = reader.GetGenes();
}
var entrezGeneIdToSymbol = genes.GetGeneIdToSymbol(x => x.EntrezGeneId);
var ensemblIdToSymbol = genes.GetGeneIdToSymbol(x => x.EnsemblId);
return(entrezGeneIdToSymbol, ensemblIdToSymbol);
}