private NsiReader(GenomeAssembly assembly, IDataSourceVersion version, string jsonKey, ReportFor reportFor, Dictionary <ushort, IntervalArray <string> > intervalArrays)
{
Assembly = assembly;
Version = version;
JsonKey = jsonKey;
ReportFor = reportFor;
_intervalArrays = intervalArrays;
}
private NsiReader(GenomeAssembly assembly, IDataSourceVersion version, string jsonKey, ReportFor reportFor, IntervalArray <string>[] intervalArrays)
{
Assembly = assembly;
Version = version;
JsonKey = jsonKey;
ReportFor = reportFor;
_intervalForest = new IntervalForest <string>(intervalArrays);
}
public IntervalTsvWriter(string outputPath, DataSourceVersion dataSourceVersion, string assembly, int dataVersion, string keyName,
ReportFor reportingFor)
{
var fileName = keyName + "_" + dataSourceVersion.Version.Replace(" ", "_") + ".interval.tsv.gz";
_bgzipTextWriter = new BgzipTextWriter(Path.Combine(outputPath, fileName));
_bgzipTextWriter.Write(GetHeader(dataSourceVersion, dataVersion, assembly, keyName, reportingFor));
_tsvIndex = new TsvIndex(Path.Combine(outputPath, fileName) + ".tvi");
}
public SupplementaryInterval(string keyName, string refName, int start, int end, string jsonString,
ReportFor reportingFor)
{
KeyName = keyName;
ReferenceName = refName;
Start = start;
End = end;
JsonString = jsonString;
ReportingFor = reportingFor;
}
public NsiWriter(Stream stream, DataSourceVersion version,
GenomeAssembly assembly, string jsonKey, ReportFor reportFor, int schemaVersion,
bool leaveOpen = false)
{
_stream = stream;
_leaveOpen = leaveOpen;
WriteHeader(version, assembly, jsonKey, reportFor, schemaVersion);
var blockStream = new BlockStream(new Zstandard(), stream, CompressionMode.Compress);
_writer = new ExtendedBinaryWriter(blockStream, Encoding.UTF8, leaveOpen);
}
public NsiWriter(BinaryWriter writer, DataSourceVersion version,
GenomeAssembly assembly, string jsonKey, ReportFor reportFor, int schemaVersion)
{
_writer = writer;
_memStream = new MemoryStream();
_memWriter = new ExtendedBinaryWriter(_memStream);
version.Write(_memWriter);
_memWriter.Write((byte)assembly);
_memWriter.WriteOptAscii(jsonKey);
_memWriter.Write((byte)reportFor);
_memWriter.WriteOpt(schemaVersion);
}
private void ParseHeaderLine(string line)
{
var words = line.Split('=');
if (words.Length < 2)
{
return;
}
var key = words[0];
var value = words[1];
switch (key)
{
case "#name":
_name = value;
break;
case "#assembly":
_genomeAssembly = value;
break;
case "#version":
_version = value;
break;
case "#releaseDate":
_releaseDate = value;
break;
case "#description":
_description = value;
break;
case "#keyName":
_keyName = value;
break;
case "#reportFor":
_reportFor = GetReportFor(value);
break;
case "#schemaVerstion":
var schemaVersion = int.Parse(value);
if (schemaVersion != JsonCommon.SupplementarySchemaVersion)
{
throw new InvalidDataException($"Expected Schema version:{JsonCommon.SupplementarySchemaVersion}, oberved: {value}");
}
break;
}
}
public NsiReader(Stream stream)
{
_stream = stream;
var compressData = new byte[MaxStreamLength];
int length = stream.Read(compressData, 0, MaxStreamLength);
//uncompress
var zstd = new Zstandard();
var decompressedLength = zstd.GetDecompressedLength(compressData, length);
var decompressedData = new byte[decompressedLength];
zstd.Decompress(compressData, length, decompressedData, decompressedLength);
using (var memStream = new MemoryStream(decompressedData, 0, decompressedLength))
using (var memReader = new ExtendedBinaryReader(memStream))
{
Version = DataSourceVersion.Read(memReader);
Assembly = (GenomeAssembly)memReader.ReadByte();
JsonKey = memReader.ReadAsciiString();
ReportFor = (ReportFor)memReader.ReadByte();
int schemaVersion = memReader.ReadOptInt32();
if (schemaVersion != SaCommon.SchemaVersion)
{
throw new UserErrorException($"Schema version mismatch!! Expected {SaCommon.SchemaVersion}, observed {schemaVersion} for {JsonKey}");
}
int count = memReader.ReadOptInt32();
var suppIntervals = new Dictionary <ushort, List <Interval <string> > >();
for (var i = 0; i < count; i++)
{
var saInterval = new SuppInterval(memReader);
if (suppIntervals.TryGetValue(saInterval.Chromosome.Index, out var intervals))
{
intervals.Add(new Interval <string>(saInterval.Start, saInterval.End, saInterval.GetJsonString()));
}
else
{
suppIntervals[saInterval.Chromosome.Index] = new List <Interval <string> > {
new Interval <string>(saInterval.Start, saInterval.End, saInterval.GetJsonString())
}
};
}
_intervalArrays = new Dictionary <ushort, IntervalArray <string> >(suppIntervals.Count);
foreach ((ushort chromIndex, List <Interval <string> > intervals) in suppIntervals)
{
_intervalArrays[chromIndex] = new IntervalArray <string>(intervals.ToArray());
}
}
}
public static (bool MatchByAllele, bool IsArray, SaJsonValueType PrimaryType, ReportFor reportFor) ParseMatchVariantsBy(string line)
{
line = line.Trim();
CheckPrefix(line, "#matchVariantsBy");
string firstCol = line.OptimizedSplit('\t')[0];
(_, string matchBy) = firstCol.OptimizedKeyValue();
bool matchByAllele;
bool isArray;
SaJsonValueType primaryType;
ReportFor reportFor = ReportFor.AllVariants;
switch (matchBy)
{
case null:
throw new UserErrorException("Please provide the annotation reporting criteria in the format: #matchVariantsBy=allele.");
case "allele":
matchByAllele = true;
isArray = false;
primaryType = SaJsonValueType.Object;
break;
case "position":
primaryType = SaJsonValueType.ObjectArray;
matchByAllele = false;
isArray = true;
break;
case "sv":
primaryType = SaJsonValueType.ObjectArray;
matchByAllele = false;
isArray = true;
reportFor = ReportFor.StructuralVariants;
break;
default:
throw new UserErrorException("matchVariantsBy tag has to be either \'allele\' or \'position\'");
}
return(matchByAllele, isArray, primaryType, reportFor);
}
public IntervalAnnotationHeader(string name, string assembly, string version, string releaseDate, string description,
ReportFor reportingFor) : base(name, assembly, version, releaseDate, description)
{
_reportingFor = reportingFor;
}
public static List <ISupplementaryInterval> GetSpecificIntervals(ReportFor reportFor, IEnumerable <ISupplementaryInterval> intervals)
{
return(intervals.Where(interval => interval.ReportingFor == reportFor).ToList());
}
public static NsiWriter GetNsiWriter(Stream nsiStream, DataSourceVersion version, GenomeAssembly assembly, string jsonTag, ReportFor reportFor) => new NsiWriter(nsiStream, version, assembly, jsonTag, reportFor, SaCommon.SchemaVersion);
private static string GetHeader(DataSourceVersion dataSourceVersion, int dataVersion, string assembly, string keyName, ReportFor reportingFor)
{
var sb = StringBuilderCache.Acquire();
sb.Append($"#name={dataSourceVersion.Name}\n");
sb.Append($"#assembly={assembly}\n");
sb.Append($"#version={dataSourceVersion.Version}\n");
sb.Append($"#description={dataSourceVersion.Description}\n");
var releaseDate = new DateTime(dataSourceVersion.ReleaseDateTicks, DateTimeKind.Utc);
sb.Append($"#releaseDate={releaseDate:yyyy-MM-dd}\n");
sb.Append($"#dataVersion={dataVersion}\n");
sb.Append($"#schemaVersion={JsonCommon.SupplementarySchemaVersion}\n");
sb.Append($"#reportFor={reportingFor}\n");
sb.Append($"#keyName={keyName}\n");
sb.Append("#CHROM\tSTART\tEND\tJSON\n");
return(StringBuilderCache.GetStringAndRelease(sb));
}
private void WriteHeader(DataSourceVersion version, GenomeAssembly assembly, string jsonKey, ReportFor reportFor, int schemaVersion)
{
using (var writer = new ExtendedBinaryWriter(_stream, Encoding.UTF8, true))
{
writer.WriteOptAscii(SaCommon.NsiIdentifier);
version.Write(writer);
writer.Write((byte)assembly);
writer.WriteOptAscii(jsonKey);
writer.Write((byte)reportFor);
writer.Write(schemaVersion);
writer.Write(SaCommon.GuardInt);
}
}
