private void LoadHeader()
{
// grab the index and masked intervals offsets
_indexOffset = _reader.ReadInt64();
_reader.ReadInt64();
// skip the creation time
_reader.ReadOptInt64();
// grab the reference metadata
NumRefSeqs = (ushort)_reader.ReadOptInt32();
for (ushort i = 0; i < NumRefSeqs; i++)
{
var metadata = ReferenceMetadata.Read(_reader);
ReferenceMetadataList.Add(metadata);
AddReferenceName(metadata.EnsemblName, metadata.UcscName, i);
}
// read the cytogenetic bands
CytogeneticBands = Genome.CytogeneticBands.Read(_reader);
// read the genome assembly
Sequence.Assembly = (GenomeAssembly)_reader.ReadByte();
// grab the data start tag
ulong dataStartTag = _reader.ReadUInt64();
if (dataStartTag != CompressedSequenceCommon.DataStartTag)
{
throw new InvalidFileFormatException($"The data start tag does not match the expected values: Obs: {dataStartTag} vs Exp: {CompressedSequenceCommon.DataStartTag}");
}
}
public NsaIndexBlock(ExtendedBinaryReader reader)
{
Start = reader.ReadOptInt32();
End = reader.ReadOptInt32();
FilePosition = reader.ReadOptInt64();
Length = reader.ReadOptInt32();
}
public Chunk(ExtendedBinaryReader reader)
{
_start = reader.ReadOptInt32();
_end = reader.ReadOptInt32();
FilePosition = reader.ReadOptInt64();
Length = reader.ReadOptInt32();
}
public JasixNode(ExtendedBinaryReader reader)
{
_start = reader.ReadOptInt32();
//on disk we will store the end as an offset to save space
_end = _start + reader.ReadOptInt32();
FileLocation = reader.ReadOptInt64();
}
private static Interval <long> ReadInterval(ExtendedBinaryReader reader)
{
int begin = reader.ReadOptInt32();
int end = reader.ReadOptInt32();
long position = reader.ReadOptInt64();
return(new Interval <long>(begin, end, position));
}
public static IDataSourceVersion Read(ExtendedBinaryReader reader)
{
var name = reader.ReadAsciiString();
var version = reader.ReadAsciiString();
var releaseDateTicks = reader.ReadOptInt64();
var description = reader.ReadAsciiString();
return(new DataSourceVersion(name, version, releaseDateTicks, description));
}
internal DataSourceVersion(ExtendedBinaryReader reader)
{
Name = reader.ReadAsciiString();
Version = reader.ReadAsciiString();
ReleaseDateTicks = reader.ReadOptInt64();
Description = reader.ReadAsciiString();
_hashCode = Name.GetHashCode() ^ Version.GetHashCode() ^ ReleaseDateTicks.GetHashCode();
}
public void Read(ExtendedBinaryReader reader)
{
var count = reader.ReadOptInt32();
for (var i = 0; i < count; i++)
{
DbSnp.Add(reader.ReadOptInt64());
}
}
public static Phenotype ReadPhenotype(ExtendedBinaryReader reader)
{
var mimNumber = reader.ReadOptInt64();
var phenotype = reader.ReadAsciiString();
var mapping = (Mapping)reader.ReadByte();
var comments = (Comments)reader.ReadByte();
var inheritance = reader.ReadOptArray(reader.ReadAsciiString);
var inheritances = inheritance == null ? null : new HashSet <string>(inheritance);
return(new Phenotype(mimNumber, phenotype, mapping, comments, inheritances));
}
private static long GetObservedLong(long expectedLong)
{
long observedLong;
using (var memoryStream = GetMemoryStream(expectedLong))
using (var reader = new ExtendedBinaryReader(memoryStream))
{
observedLong = reader.ReadOptInt64();
}
return(observedLong);
}
public JasixIndex(Stream stream) : this()
{
_stream = stream;
using (var reader = new ExtendedBinaryReader(stream))
{
int version = reader.ReadOptInt32();
if (version != JasixCommons.Version)
{
throw new InvalidDataException($"Invalid Jasix version: Observed {version}, expected{JasixCommons.Version}");
}
int count = reader.ReadOptInt32();
for (var i = 0; i < count; i++)
{
var chrIndex = new JasixChrIndex(reader);
_chrIndices[chrIndex.ReferenceSequence] = chrIndex;
}
int synonymCount = reader.ReadOptInt32();
for (var i = 0; i < synonymCount; i++)
{
string synonym = reader.ReadAsciiString();
string indexName = reader.ReadAsciiString();
_synonymToChrName[synonym] = indexName;
}
int sectionCount = reader.ReadOptInt32();
for (var i = 0; i < sectionCount; i++)
{
string sectionName = reader.ReadAsciiString();
long begin = reader.ReadOptInt64();
long end = reader.ReadOptInt64();
_sectionRanges[sectionName] = new FileRange(begin, end);
}
}
}
public static OmimAnnotation Read(ExtendedBinaryReader reader)
{
var hgnc = reader.ReadAsciiString();
var description = reader.ReadAsciiString();
var mimNumber = reader.ReadOptInt64();
var phenotypeCount = reader.ReadOptInt32();
var phenotypes = new List <Phenotype>();
for (var i = 0; i < phenotypeCount; i++)
{
phenotypes.Add(Phenotype.ReadPhenotype(reader));
}
return(new OmimAnnotation(hgnc, description, mimNumber, phenotypes));
}
public ClinVarItem(ExtendedBinaryReader reader)
{
_alleleOrigins = reader.ReadOptArray(reader.ReadAsciiString)?.ToList();
SaAltAllele = reader.ReadAsciiString();
AltAllele = SaAltAllele != null?SupplementaryAnnotationUtilities.ReverseSaReducedAllele(SaAltAllele) : ""; // A
ReferenceAllele = reader.ReadAsciiString();
ID = reader.ReadAsciiString();
ReviewStatus = (ReviewStatusEnum)reader.ReadByte();
IsAlleleSpecific = reader.ReadAsciiString();
_medgenIds = reader.ReadOptArray(reader.ReadAsciiString)?.ToList();
_omimIds = reader.ReadOptArray(reader.ReadAsciiString)?.ToList();
_orphanetIds = reader.ReadOptArray(reader.ReadAsciiString)?.ToList();
_phenotypes = reader.ReadOptArray(reader.ReadUtf8String)?.ToList();
Significance = reader.ReadAsciiString();
LastUpdatedDate = reader.ReadOptInt64();
_pubmedIds = reader.ReadOptArray(reader.ReadOptInt64)?.ToList();
}
public void Long(long expectedNum)
{
long observedNum;
using (var ms = new MemoryStream())
{
using (var writer = new ExtendedBinaryWriter(ms, Encoding.UTF8, true))
{
writer.WriteOpt(expectedNum);
}
ms.Seek(0, SeekOrigin.Begin);
using (var reader = new ExtendedBinaryReader(ms))
{
observedNum = reader.ReadOptInt64();
}
}
Assert.Equal(expectedNum, observedNum);
}
public void ReadOptInt64_ThrowException_WithCorruptData()
{
Assert.Throws <FormatException>(delegate
{
using (var ms = new MemoryStream())
{
using (var writer = new BinaryWriter(ms, Encoding.UTF8, true))
{
const ulong corruptData = 0xffffffffffffffff;
writer.Write(corruptData);
writer.Write(corruptData);
}
ms.Position = 0;
using (var reader = new ExtendedBinaryReader(ms))
{
reader.ReadOptInt64();
}
}
});
}