/// <summary>
/// executes the program
/// </summary>
protected override void ProgramExecution()
{
var transcriptPath = ConfigurationSettings.InputPrefix + ".transcripts.gz";
var regulatoryPath = ConfigurationSettings.InputPrefix + ".regulatory.gz";
var genePath = ConfigurationSettings.InputPrefix + ".genes.gz";
var intronPath = ConfigurationSettings.InputPrefix + ".introns.gz";
var mirnaPath = ConfigurationSettings.InputPrefix + ".mirnas.gz";
var siftPath = ConfigurationSettings.InputPrefix + ".sift.dat";
var polyphenPath = ConfigurationSettings.InputPrefix + ".polyphen.dat";
var peptidePath = ConfigurationSettings.InputPrefix + ".peptides.gz";
var renamer = ChromosomeRenamer.GetChromosomeRenamer(FileUtilities.GetReadStream(ConfigurationSettings.InputReferencePath));
using (var transcriptReader = new VepTranscriptReader(transcriptPath))
using (var regulatoryReader = new VepRegulatoryReader(regulatoryPath))
using (var geneReader = new VepGeneReader(genePath))
using (var mergedGeneReader = new VepCombinedGeneReader(ConfigurationSettings.InputMergedGenesPath))
using (var intronReader = new VepSimpleIntervalReader(intronPath, "intron", GlobalImportCommon.FileType.Intron))
using (var mirnaReader = new VepSimpleIntervalReader(mirnaPath, "miRNA", GlobalImportCommon.FileType.MicroRna))
using (var peptideReader = new VepSequenceReader(peptidePath, "peptide", GlobalImportCommon.FileType.Peptide))
{
var converter = new NirvanaDatabaseCreator(transcriptReader, regulatoryReader, geneReader,
mergedGeneReader, intronReader, mirnaReader, peptideReader, renamer);
converter.LoadData();
converter.MarkCanonicalTranscripts(ConfigurationSettings.InputLrgPath);
converter.CreateTranscriptCacheFile(ConfigurationSettings.OutputCacheFilePrefix);
converter.CopyPredictionCacheFile("SIFT", siftPath, CacheConstants.SiftPath(ConfigurationSettings.OutputCacheFilePrefix));
converter.CopyPredictionCacheFile("PolyPhen", polyphenPath, CacheConstants.PolyPhenPath(ConfigurationSettings.OutputCacheFilePrefix));
}
}
/// <summary>
/// loads all the genes in the specified file
/// </summary>
private List <MutableGene> LoadGenes(string genesPath, string description)
{
var genes = new List <MutableGene>();
using (var reader = new VepGeneReader(genesPath))
{
if (_header == null)
{
_header = reader.Header;
}
while (true)
{
var gene = reader.Next();
if (gene == null)
{
break;
}
genes.Add(gene);
}
}
var transcriptDataSource = genes.First().TranscriptDataSource;
int numGenesWithoutSymbol = GetNumGenesWithoutSymbol(description, genes);
if (numGenesWithoutSymbol > 0 && transcriptDataSource == TranscriptDataSource.RefSeq)
{
ResolveMissingRefSeqGeneSymbols(description, genes);
}
return(genes);
}
/// <summary>
/// constructor
/// </summary>
public NirvanaDatabaseCreator(VepTranscriptReader transcriptReader, VepRegulatoryReader regulatoryReader,
VepGeneReader geneReader, VepCombinedGeneReader mergedGeneReader, VepSimpleIntervalReader intronReader,
VepSimpleIntervalReader mirnaReader, VepSequenceReader peptideReader, ChromosomeRenamer renamer)
{
_transcriptReader = transcriptReader;
_regulatoryReader = regulatoryReader;
_geneReader = geneReader;
_mergedGeneReader = mergedGeneReader;
_intronReader = intronReader;
_microRnaReader = mirnaReader;
_peptideReader = peptideReader;
_renamer = renamer;
_transcripts = new List <VD.Transcript>();
_regulatoryElements = new List <VD.RegulatoryElement>();
_genes = new List <MutableGene>();
_mergedGenes = new List <MutableGene>();
_introns = new List <VD.SimpleInterval>();
_microRnas = new List <VD.SimpleInterval>();
_peptideSeqs = new List <string>();
}