public int Extract()
{
var count = 0;
using (_reader)
using (_writer)
{
for (var i = _begin; i <= _end; i++)
{
var sa = _reader.GetAnnotation(i) as SupplementaryAnnotationPosition;
if (sa != null)
{
count++;
_writer.Write(new SupplementaryPositionCreator(sa), i);
}
}
var miniSaInterval = new AnnotationInterval(_begin, _end);
// get the supplementary intervals overlapping the mini SA interval
var suppIntervals = new List <SupplementaryInterval>();
var readerSuppInterval = _reader.GetSupplementaryIntervals(_renamer);
if (readerSuppInterval != null)
{
foreach (var interval in readerSuppInterval)
{
if (miniSaInterval.Overlaps(interval.Start, interval.End))
{
suppIntervals.Add(interval as SupplementaryInterval);
}
}
}
Console.WriteLine("Found {0} supplementary intervals.", suppIntervals.Count);
_writer.SetIntervalList(suppIntervals);
}
return(count);
}
public void ReadWriteWithSuppIntervals()
{
// NIR-1359
var randomPath = Path.Combine(Path.GetTempPath(), Path.GetRandomFileName());
// create our expected data source versions
var dbSnpVersion = new DataSourceVersion("dbSNP", "142", DateTime.Parse("2015-01-02").Ticks);
var expectedDataSourceVersions = new List <DataSourceVersion> {
dbSnpVersion
};
// This is the case where Nirvana throws an error: Too many bytes in what should have been a 7 bit encoded Int32.
var sa = new SupplementaryAnnotationPosition(5073770);
var saCreator = new SupplementaryPositionCreator(sa);
// adding a supplementary interval
var intValues = new Dictionary <string, int>();
var doubleValues = new Dictionary <string, double>();
var freqValues = new Dictionary <string, double>();
var stringValues = new Dictionary <string, string>();
var boolValues = new List <string>();
var suppInterval = new SupplementaryInterval(5073770, 5073970, "chr1", "<DUP>", VariantType.duplication, "ClinVar", _renamer, intValues,
doubleValues, freqValues, stringValues, boolValues);
suppInterval.AddStringValue("ID", "RandomClin001");
// the above code was unit tested in MergeDbSnpClinVar()
using (var writer = new SupplementaryAnnotationWriter(randomPath, "chr9", expectedDataSourceVersions))
{
writer.SetIntervalList(new List <SupplementaryInterval> {
suppInterval
});
writer.Write(saCreator, sa.ReferencePosition);
}
// read the supplementary annotation file
using (var reader = new SupplementaryAnnotationReader(randomPath))
{
// read the stored intervals
var suppIntervals = reader.GetSupplementaryIntervals(_renamer).ToList();
Assert.Equal(1, suppIntervals.Count);
foreach (var interval in suppIntervals)
{
Assert.Equal(5073770, interval.Start);
Assert.Equal(5073970, interval.End);
Assert.Equal("<DUP>", interval.AlternateAllele);
Assert.Equal("ClinVar", interval.Source);
Assert.Equal("duplication", interval.VariantType.ToString());
foreach (var keyValuePair in interval.StringValues)
{
if (keyValuePair.Key == "ID")
{
Assert.Equal("RandomClin001", keyValuePair.Value);
}
if (keyValuePair.Key == "vid")
{
Assert.Equal("1:5073770:5073970", keyValuePair.Value);
}
}
}
}
File.Delete(randomPath);
File.Delete(randomPath + ".idx");
}
public void ReadAndWrite()
{
// read the supplementary annotation file
using (var reader = new SupplementaryAnnotationReader(_randomPath))
{
var observedDataSourceVersions = reader.Header.DataSourceVersions;
var refSeq = reader.Header.ReferenceSequenceName;
var dataVersion = reader.Header.DataVersion;
var creationTime = reader.Header.CreationTimeTicks;
var genomeAssembly = reader.Header.GenomeAssembly;
// check the data source versions
Assert.Equal(observedDataSourceVersions.Count, 1);
var observedDataSourceVersion = observedDataSourceVersions[0];
Assert.Equal(_expectedDataSourceVersion.Name, observedDataSourceVersion.Name);
Assert.Equal(_expectedDataSourceVersion.Version, observedDataSourceVersion.Version);
Assert.Equal(_expectedDataSourceVersion.ReleaseDateTicks, observedDataSourceVersion.ReleaseDateTicks);
Assert.NotNull(refSeq);
Assert.Equal(SupplementaryAnnotationCommon.DataVersion, dataVersion);
Assert.True(DateTime.MinValue.Ticks != creationTime);
Assert.True(genomeAssembly == GenomeAssembly.Unknown);
var expDbsnp1 =
((DbSnpAnnotation)
_expectedAnnotation1.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
var expDbsnp2 =
((DbSnpAnnotation)
_expectedAnnotation2.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
var expDbsnp3 =
((DbSnpAnnotation)
_expectedAnnotation3.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
// extract the three annotations
var observedAnnotation1 = reader.GetAnnotation(100) as SupplementaryAnnotationPosition;
var observedAnnotation2 = reader.GetAnnotation(101) as SupplementaryAnnotationPosition;
var observedAnnotation3 = reader.GetAnnotation(102) as SupplementaryAnnotationPosition;
var obsDbsnp1 =
((DbSnpAnnotation)
observedAnnotation1.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
var obsDbsnp2 =
((DbSnpAnnotation)
observedAnnotation2.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
var obsDbsnp3 =
((DbSnpAnnotation)
observedAnnotation3.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
Assert.Equal(expDbsnp1, obsDbsnp1);
Assert.Equal(expDbsnp2, obsDbsnp2);
Assert.Equal(expDbsnp3, obsDbsnp3);
// jump around the file
var observedJumpAnnotation2 = reader.GetAnnotation(_expectedAnnotation2.ReferencePosition) as SupplementaryAnnotationPosition;
var observedJumpAnnotation1 = reader.GetAnnotation(_expectedAnnotation1.ReferencePosition) as SupplementaryAnnotationPosition;
var observedJumpAnnotation3 = reader.GetAnnotation(_expectedAnnotation3.ReferencePosition) as SupplementaryAnnotationPosition;
var obsJumpDbsnp1 =
((DbSnpAnnotation)
observedJumpAnnotation1.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
var obsJumpDbsnp2 =
((DbSnpAnnotation)
observedJumpAnnotation2.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
var obsJumpDbsnp3 =
((DbSnpAnnotation)
observedJumpAnnotation3.AlleleSpecificAnnotations[AltAllele].Annotations[
DataSourceCommon.GetIndex(DataSourceCommon.DataSource.DbSnp)]).DbSnp;
Assert.Equal(expDbsnp1, obsJumpDbsnp1);
Assert.Equal(expDbsnp2, obsJumpDbsnp2);
Assert.Equal(expDbsnp3, obsJumpDbsnp3);
var observedInterval = reader.GetSupplementaryIntervals(_renamer);
Assert.Equal(_expectedInterval, observedInterval.First());
}
}
