public void DisableOriginalOnFailedLookup()
{
var observedReferenceName = _renamer.GetUcscReferenceName("dummy", false);
Assert.Equal(null, observedReferenceName);
observedReferenceName = _renamer.GetEnsemblReferenceName("dummy", false);
Assert.Equal(null, observedReferenceName);
}
public void BeforeInitialization()
{
var emptyChromosomeNamer = new ChromosomeRenamer();
Assert.Throws <InvalidOperationException>(() => emptyChromosomeNamer.GetUcscReferenceName("1"));
Assert.Throws <InvalidOperationException>(() => emptyChromosomeNamer.GetEnsemblReferenceName("chr1"));
}
public void GetUnknownDesiredReferenceName()
{
const string expectedChromosomeName = "O";
var observedUcscReferenceName = _renamer.GetUcscReferenceName(expectedChromosomeName);
Assert.Equal(expectedChromosomeName, observedUcscReferenceName);
var observedEnsemblReferenceName = _renamer.GetEnsemblReferenceName(expectedChromosomeName);
Assert.Equal(expectedChromosomeName, observedEnsemblReferenceName);
}
public static bool IsDesiredChromosome(string chromosome, ChromosomeRenamer renamer)
{
if (ChromosomeWhiteList == null)
{
return(true);
}
if (ChromosomeWhiteList.Count == 0)
{
return(true);
}
return(ChromosomeWhiteList.Contains(renamer.GetEnsemblReferenceName(chromosome)));
}
public void AddReferenceNameEnsemblEmpty()
{
const string ucscReferenceName = "chr1";
var emptyChromosomeNamer = new ChromosomeRenamer();
var referenceMetadata = new List <ReferenceMetadata>
{
new ReferenceMetadata(null, ucscReferenceName, true)
};
emptyChromosomeNamer.AddReferenceMetadata(referenceMetadata);
var observedUcscReferenceName = emptyChromosomeNamer.GetUcscReferenceName(null);
var observedEnsemblReferenceName = emptyChromosomeNamer.GetEnsemblReferenceName(ucscReferenceName);
Assert.Equal(ucscReferenceName, observedEnsemblReferenceName);
Assert.Null(observedUcscReferenceName);
}
private void OpenNewSaWriter()
{
var currentEnsemblRefName = _renamer.GetEnsemblReferenceName(_currentRefName);
var currentUcscRefName = _renamer.GetUcscReferenceName(_currentRefName);
if (InputFileParserUtilities.ProcessedReferences.Contains(currentEnsemblRefName))
{
throw new Exception($"usorted file, for chromsome {_currentRefName}, SA will be rewritten");
}
InputFileParserUtilities.ProcessedReferences.Add(currentEnsemblRefName);
var saPath = Path.Combine(_nsdBaseFileName, currentUcscRefName + ".nsa");
_saWriter = new SupplementaryAnnotationWriter(saPath, _currentRefName, _dataSources, _compressedSequence.GenomeAssembly);
Console.WriteLine("Populating {0} data...", currentUcscRefName);
_creationBench.Reset();
_numSaWritten = 0;
}
private IEnumerable <ClinGenItem> GetClinGenItems()
{
using (var reader = GZipUtilities.GetAppropriateStreamReader(_clinGenFileInfo.FullName))
{
string line;
while ((line = reader.ReadLine()) != null)
{
if (IsClinGenHeader(line))
{
continue;
}
var cols = line.Split('\t');
string id = cols[0];
string ucscChrom = cols[1];
string chrom = _renamer.GetEnsemblReferenceName(ucscChrom);
if (!InputFileParserUtilities.IsDesiredChromosome(chrom, _renamer))
{
continue;
}
int start = int.Parse(cols[2]);
int end = int.Parse(cols[3]);
int observedGains = int.Parse(cols[4]);
int observedLosses = int.Parse(cols[5]);
var variantType = GetVariantType(cols[6]);
var clinInterpretation = GetClinInterpretation(cols[7]);
bool validated = cols[8].Equals("True");
var phenotypes = cols[9] == "" ? null : new HashSet <string>(cols[9].Split(','));
var phenotypeIds = cols[10] == "" ? null : new HashSet <string>(cols[10].Split(','));
var currentItem = new ClinGenItem(id, chrom, start, end, variantType, observedGains, observedLosses,
clinInterpretation, validated, phenotypes, phenotypeIds);
yield return(currentItem);
}
}
}
/// <summary>
/// constructs a VID based on the supplied feature
/// </summary>
public string Create(ChromosomeRenamer renamer, string referenceName, VariantAlternateAllele altAllele)
{
referenceName = renamer.GetEnsemblReferenceName(referenceName);
string vid;
switch (altAllele.NirvanaVariantType)
{
case VariantType.SNV:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.AlternateAllele}";
break;
case VariantType.insertion:
vid = altAllele.IsStructuralVariant
? $"{referenceName}:{altAllele.Start}:{altAllele.End}:INS"
: $"{referenceName}:{altAllele.Start}:{altAllele.End}:{GetInsertedAltAllele(altAllele.AlternateAllele)}";
break;
case VariantType.deletion:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}";
break;
case VariantType.MNV:
case VariantType.indel:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:{GetInsertedAltAllele(altAllele.AlternateAllele)}";
break;
case VariantType.duplication:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:DUP";
break;
case VariantType.tandem_duplication:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:TDUP";
break;
case VariantType.translocation_breakend:
vid = altAllele.BreakEnds?[0].ToString();
break;
case VariantType.inversion:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:Inverse";
break;
case VariantType.mobile_element_insertion:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:MEI";
break;
case VariantType.copy_number_gain:
case VariantType.copy_number_loss:
case VariantType.copy_number_variation:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:{altAllele.CopyNumber}";
break;
case VariantType.reference_no_call:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}:NC";
break;
default:
vid = $"{referenceName}:{altAllele.Start}:{altAllele.End}";
break;
}
return(vid);
}
// constructor
public CreateSupplementaryDatabase(
string compressedReferencePath,
string nsdBaseFileName,
string dbSnpFileName = null,
string cosmicVcfFile = null,
string cosmicTsvFile = null,
string clinVarFileName = null,
string oneKGenomeAfFileName = null,
string evsFileName = null,
string exacFileName = null,
List <string> customFiles = null,
string dgvFileName = null,
string oneKSvFileName = null,
string clinGenFileName = null,
string chrWhiteList = null)
{
_nsdBaseFileName = nsdBaseFileName;
_dataSources = new List <DataSourceVersion>();
_iSupplementaryDataItemList = new List <IEnumerator <SupplementaryDataItem> >();
_supplementaryIntervalList = new List <SupplementaryInterval>();
Console.WriteLine("Creating supplementary annotation files... Data version: {0}, schema version: {1}", SupplementaryAnnotationCommon.DataVersion, SupplementaryAnnotationCommon.SchemaVersion);
_compressedSequence = new CompressedSequence();
var compressedSequenceReader = new CompressedSequenceReader(FileUtilities.GetReadStream(compressedReferencePath), _compressedSequence);
_renamer = _compressedSequence.Renamer;
_dataFileManager = new DataFileManager(compressedSequenceReader, _compressedSequence);
if (!string.IsNullOrEmpty(chrWhiteList))
{
Console.WriteLine("Creating SA for the following chromosomes only:");
foreach (var refSeq in chrWhiteList.Split(','))
{
InputFileParserUtilities.ChromosomeWhiteList.Add(_renamer.GetEnsemblReferenceName(refSeq));
Console.Write(refSeq + ",");
}
Console.WriteLine();
}
else
{
InputFileParserUtilities.ChromosomeWhiteList = null;
}
if (dbSnpFileName != null)
{
AddSourceVersion(dbSnpFileName);
var dbSnpReader = new DbSnpReader(new FileInfo(dbSnpFileName), _renamer);
var dbSnpEnumerator = dbSnpReader.GetEnumerator();
_iSupplementaryDataItemList.Add(dbSnpEnumerator);
}
if (cosmicVcfFile != null && cosmicTsvFile != null)
{
AddSourceVersion(cosmicVcfFile);
var cosmicReader = new MergedCosmicReader(cosmicVcfFile, cosmicTsvFile, _renamer);
var cosmicEnumerator = cosmicReader.GetEnumerator();
_iSupplementaryDataItemList.Add(cosmicEnumerator);
}
if (oneKGenomeAfFileName != null)
{
AddSourceVersion(oneKGenomeAfFileName);
var oneKGenReader = new OneKGenReader(new FileInfo(oneKGenomeAfFileName), _renamer);
var oneKGenEnumerator = oneKGenReader.GetEnumerator();
_iSupplementaryDataItemList.Add(oneKGenEnumerator);
}
if (oneKSvFileName != null)
{
if (oneKGenomeAfFileName == null)
{
AddSourceVersion(oneKSvFileName);
}
var oneKGenSvReader = new OneKGenSvReader(new FileInfo(oneKSvFileName), _renamer);
var oneKGenSvEnumerator = oneKGenSvReader.GetEnumerator();
_iSupplementaryDataItemList.Add(oneKGenSvEnumerator);
}
if (evsFileName != null)
{
AddSourceVersion(evsFileName);
var evsReader = new EvsReader(new FileInfo(evsFileName), _renamer);
var evsEnumerator = evsReader.GetEnumerator();
_iSupplementaryDataItemList.Add(evsEnumerator);
}
if (exacFileName != null)
{
AddSourceVersion(exacFileName);
var exacReader = new ExacReader(new FileInfo(exacFileName), _renamer);
var exacEnumerator = exacReader.GetEnumerator();
_iSupplementaryDataItemList.Add(exacEnumerator);
}
if (clinVarFileName != null)
{
AddSourceVersion(clinVarFileName);
var clinVarReader = new ClinVarXmlReader(new FileInfo(clinVarFileName), compressedSequenceReader, _compressedSequence);
var clinVarList = clinVarReader.ToList();
clinVarList.Sort();
Console.WriteLine($"{clinVarList.Count} clinvar items read form XML file");
IEnumerator <ClinVarItem> clinVarEnumerator = clinVarList.GetEnumerator();
_iSupplementaryDataItemList.Add(clinVarEnumerator);
}
if (dgvFileName != null)
{
AddSourceVersion(dgvFileName);
var dgvReader = new DgvReader(new FileInfo(dgvFileName), _renamer);
var dgvEnumerator = dgvReader.GetEnumerator();
_iSupplementaryDataItemList.Add(dgvEnumerator);
}
if (clinGenFileName != null)
{
AddSourceVersion(clinGenFileName);
var clinGenReader = new ClinGenReader(new FileInfo(clinGenFileName), _renamer);
var clinGenEnumerator = clinGenReader.GetEnumerator();
_iSupplementaryDataItemList.Add(clinGenEnumerator);
}
if (customFiles != null)
{
foreach (var customFile in customFiles)
{
AddSourceVersion(customFile);
var customReader = new CustomAnnotationReader(new FileInfo(customFile), _renamer);
var customEnumerator = customReader.GetEnumerator();
_iSupplementaryDataItemList.Add(customEnumerator);
}
}
// initializing the IEnumerators in the list
foreach (var iDataEnumerator in _iSupplementaryDataItemList)
{
if (!iDataEnumerator.MoveNext())
{
_iSupplementaryDataItemList.Remove(iDataEnumerator);
}
}
_additionalItemsList = new List <SupplementaryDataItem>();
}