/// <summary>
/// executes the program
/// </summary>
protected override void ProgramExecution()
{
var converter = new FastaToCompressedConverter();
var genomeAssembly = GenomeAssemblyUtilities.Convert(ConfigurationSettings.GenomeAssembly);
converter.Convert(ConfigurationSettings.InputFastaPath, ConfigurationSettings.InputCytobandPath,
ConfigurationSettings.InputChromosomeNamesPath, ConfigurationSettings.OutputCompressedPath,
genomeAssembly);
}
public void GenomeAssemblyConvertTest()
{
Assert.Equal(GenomeAssembly.GRCh37, GenomeAssemblyUtilities.Convert("grCh37"));
Assert.Equal(GenomeAssembly.GRCh38, GenomeAssemblyUtilities.Convert("GrCh38"));
Assert.Equal(GenomeAssembly.hg19, GenomeAssemblyUtilities.Convert("HG19"));
var error = false;
try
{
GenomeAssemblyUtilities.Convert("asldkjf");
}
catch (Exception)
{
error = true;
}
Assert.True(error);
}
/// <summary>
/// executes the program
/// </summary>
protected override void ProgramExecution()
{
var transcriptSource = ConfigurationSettings.ImportRefSeqTranscripts
? TranscriptDataSource.RefSeq
: TranscriptDataSource.Ensembl;
var referenceIndex = new ReferenceIndex(ConfigurationSettings.InputReferencePath);
var vepDirectories = referenceIndex.GetUcscKaryotypeOrder(ConfigurationSettings.InputVepDirectory);
var converter = new VepCacheParser(transcriptSource);
var genomeAssembly = GenomeAssemblyUtilities.Convert(ConfigurationSettings.GenomeAssembly);
// =========================
// create the pre-cache file
// =========================
// process each VEP directory
int numDirectoriesProcessed = 0;
var transcriptPath = ConfigurationSettings.OutputStub + ".transcripts.gz";
var regulatoryPath = ConfigurationSettings.OutputStub + ".regulatory.gz";
var genePath = ConfigurationSettings.OutputStub + ".genes.gz";
var intronPath = ConfigurationSettings.OutputStub + ".introns.gz";
var exonPath = ConfigurationSettings.OutputStub + ".exons.gz";
var mirnaPath = ConfigurationSettings.OutputStub + ".mirnas.gz";
var siftPath = ConfigurationSettings.OutputStub + ".sift.dat";
var polyphenPath = ConfigurationSettings.OutputStub + ".polyphen.dat";
var cdnaPath = ConfigurationSettings.OutputStub + ".cdnas.gz";
var peptidePath = ConfigurationSettings.OutputStub + ".peptides.gz";
using (var transcriptWriter = GZipUtilities.GetStreamWriter(transcriptPath))
using (var regulatoryWriter = GZipUtilities.GetStreamWriter(regulatoryPath))
using (var geneWriter = GZipUtilities.GetStreamWriter(genePath))
using (var intronWriter = GZipUtilities.GetStreamWriter(intronPath))
using (var exonWriter = GZipUtilities.GetStreamWriter(exonPath))
using (var mirnaWriter = GZipUtilities.GetStreamWriter(mirnaPath))
using (var siftWriter = GZipUtilities.GetBinaryWriter(siftPath + ".tmp"))
using (var polyphenWriter = GZipUtilities.GetBinaryWriter(polyphenPath + ".tmp"))
using (var cdnaWriter = GZipUtilities.GetStreamWriter(cdnaPath))
using (var peptideWriter = GZipUtilities.GetStreamWriter(peptidePath))
{
transcriptWriter.NewLine = "\n";
regulatoryWriter.NewLine = "\n";
geneWriter.NewLine = "\n";
intronWriter.NewLine = "\n";
exonWriter.NewLine = "\n";
mirnaWriter.NewLine = "\n";
cdnaWriter.NewLine = "\n";
peptideWriter.NewLine = "\n";
WriteHeader(transcriptWriter, GlobalImportCommon.FileType.Transcript, transcriptSource, genomeAssembly);
WriteHeader(regulatoryWriter, GlobalImportCommon.FileType.Regulatory, transcriptSource, genomeAssembly);
WriteHeader(geneWriter, GlobalImportCommon.FileType.Gene, transcriptSource, genomeAssembly);
WriteHeader(intronWriter, GlobalImportCommon.FileType.Intron, transcriptSource, genomeAssembly);
WriteHeader(exonWriter, GlobalImportCommon.FileType.Exon, transcriptSource, genomeAssembly);
WriteHeader(mirnaWriter, GlobalImportCommon.FileType.MicroRna, transcriptSource, genomeAssembly);
WriteHeader(siftWriter, GlobalImportCommon.FileType.Sift, transcriptSource, genomeAssembly);
WriteHeader(polyphenWriter, GlobalImportCommon.FileType.PolyPhen, transcriptSource, genomeAssembly);
WriteHeader(cdnaWriter, GlobalImportCommon.FileType.CDna, transcriptSource, genomeAssembly);
WriteHeader(peptideWriter, GlobalImportCommon.FileType.Peptide, transcriptSource, genomeAssembly);
foreach (var refTuple in vepDirectories)
{
// DEBUG
//if (refTuple.Item1 != "chr7") continue;
Console.WriteLine("Parsing reference sequence [{0}]:", refTuple.Item1);
numDirectoriesProcessed++;
var refIndex = referenceIndex.GetIndex(refTuple.Item1);
converter.ParseDumpDirectory(refIndex, refTuple.Item2, transcriptWriter, regulatoryWriter, geneWriter,
intronWriter, exonWriter, mirnaWriter, siftWriter, polyphenWriter, cdnaWriter, peptideWriter);
}
}
Console.WriteLine("\n{0} directories processed.", numDirectoriesProcessed);
converter.DumpStatistics();
Console.WriteLine();
// convert our protein function predictions
var predictionConverter = new PredictionConverter(referenceIndex.NumReferenceSeqs);
predictionConverter.Convert(siftPath, "SIFT", GlobalImportCommon.FileType.Sift);
predictionConverter.Convert(polyphenPath, "PolyPhen", GlobalImportCommon.FileType.PolyPhen);
}
/// <summary>
/// executes the program
/// </summary>
protected override void ProgramExecution()
{
Console.WriteLine("Reading file: {0}", ConfigurationSettings.InputWigFixFile);
var timer = new Benchmark();
var version = GetDataVersion();
using (var nirvanaPhylopDatabaseCreator = new PhylopWriter(ConfigurationSettings.InputWigFixFile, version, GenomeAssemblyUtilities.Convert(ConfigurationSettings.GenomeAssembly), ConfigurationSettings.OutputNirvanaDirectory))
{
nirvanaPhylopDatabaseCreator.ExtractPhylopScores();
}
Console.WriteLine("Time:{0}", timer.GetElapsedTime());
}