public void Convert(string outputPath, string description, GlobalImportCommon.FileType fileType)
{
var inputPath = outputPath.Replace(".dat", ".dat.tmp");
using (var reader = new TempPredictionReader(inputPath, description, fileType))
{
Console.Write($"- loading {description}... ");
var tempPredictions = Load(reader);
Console.WriteLine("finished.");
Console.Write($"- creating {description} LUT... ");
var oldLut = TempPrediction.CreateLookupTable(tempPredictions);
var newLut = TempPrediction.ConvertLookupTable(oldLut);
Console.WriteLine("finished.");
Console.Write($"- converting {description} matrices... ");
var predictionsPerRef = TempPrediction.ConvertMatrices(tempPredictions, oldLut, newLut, _numReferenceSeqs);
Console.WriteLine("finished.");
tempPredictions.Clear();
var header = PredictionCacheHeader.GetHeader(CurrentTimeTicks, reader.Header.GenomeAssembly, _numReferenceSeqs);
Console.Write($"- writing to {Path.GetFileName(outputPath)}... ");
using (var writer = new PredictionCacheWriter(outputPath, header))
{
writer.Write(newLut, predictionsPerRef);
}
Console.WriteLine("finished.");
}
}
/// <summary>
/// returns the file header
/// </summary>
public static GlobalImportHeader GetHeader(string description, string filePath,
GlobalImportCommon.FileType expectedFileType, StreamReader reader)
{
string line = reader.ReadLine();
if (!IsValidFile(line, expectedFileType))
{
throw new GeneralException($"The {description} file ({filePath}) has an invalid header.");
}
line = reader.ReadLine();
if (line == null)
{
throw new GeneralException($"The {description} file ({filePath}) has an invalid header.");
}
var cols = line.Split('\t');
if (cols.Length != GlobalImportCommon.NumHeaderColumns)
{
throw new GeneralException($"Expected {GlobalImportCommon.NumHeaderColumns} columns in the header but found {cols.Length}");
}
var vepVersion = ushort.Parse(cols[0]);
var vepReleaseTicks = long.Parse(cols[1]);
var transcriptSource = (TranscriptDataSource)byte.Parse(cols[2]);
var genomeAssembly = (GenomeAssembly)byte.Parse(cols[3]);
return(new GlobalImportHeader(vepVersion, vepReleaseTicks, transcriptSource, genomeAssembly));
}
/// <summary>
/// writes the header to our output file
/// </summary>
private static void WriteHeader(StreamWriter writer, GlobalImportCommon.FileType fileType,
TranscriptDataSource transcriptSource, GenomeAssembly genomeAssembly)
{
var vepReleaseTicks = DateTime.Parse(ConfigurationSettings.VepReleaseDate).Ticks;
writer.WriteLine("{0}\t{1}", GlobalImportCommon.Header, (byte)fileType);
writer.WriteLine("{0}\t{1}\t{2}\t{3}", ConfigurationSettings.VepVersion, vepReleaseTicks, (byte)transcriptSource, (byte)genomeAssembly);
}
/// <summary>
/// constructor
/// </summary>
public VepSimpleIntervalReader(string filePath, string description, GlobalImportCommon.FileType fileType)
{
// sanity check
if (!File.Exists(filePath))
{
throw new FileNotFoundException($"The specified intron file ({filePath}) does not exist.");
}
// open the vcf file and parse the header
_reader = GZipUtilities.GetAppropriateStreamReader(filePath);
VepReaderCommon.GetHeader(description, filePath, fileType, _reader);
}
/// <summary>
/// constructor
/// </summary>
public TempPredictionReader(string filePath, string description, GlobalImportCommon.FileType fileType)
{
// sanity check
if (!File.Exists(filePath))
{
throw new FileNotFoundException($"The specified protein function prediction file ({filePath}) does not exist.");
}
// open the vcf file and parse the header
_reader = GZipUtilities.GetAppropriateBinaryReader(filePath);
Header = GetHeader(description, filePath, fileType, _reader);
}
/// <summary>
/// writes the header to our output file
/// </summary>
private static void WriteHeader(BinaryWriter writer, GlobalImportCommon.FileType fileType,
TranscriptDataSource transcriptSource, GenomeAssembly genomeAssembly)
{
var vepReleaseTicks = DateTime.Parse(ConfigurationSettings.VepReleaseDate).Ticks;
writer.Write(GlobalImportCommon.Header);
writer.Write((byte)fileType);
writer.Write(ConfigurationSettings.VepVersion);
writer.Write(vepReleaseTicks);
writer.Write((byte)transcriptSource);
writer.Write((byte)genomeAssembly);
writer.Write(CacheConstants.GuardInt);
}
/// <summary>
/// returns the file header
/// </summary>
private static GlobalImportHeader GetHeader(string description, string filePath,
GlobalImportCommon.FileType expectedFileType, BinaryReader reader)
{
var header = reader.ReadString();
var fileType = (GlobalImportCommon.FileType)reader.ReadByte();
if (!IsValidFile(header, fileType, expectedFileType))
{
throw new GeneralException($"The {description} file ({filePath}) has an invalid header.");
}
var vepVersion = reader.ReadUInt16();
var vepReleaseTicks = reader.ReadInt64();
var transcriptSource = (TranscriptDataSource)reader.ReadByte();
var genomeAssembly = (GenomeAssembly)reader.ReadByte();
var guardInt = reader.ReadUInt32();
if (guardInt != CacheConstants.GuardInt)
{
throw new GeneralException($"The {description} file ({filePath}) has an invalid header.");
}
return(new GlobalImportHeader(vepVersion, vepReleaseTicks, transcriptSource, genomeAssembly));
}
/// <summary>
/// returns true if this is a valid VEP reader file (binary)
/// </summary>
private static bool IsValidFile(string header, GlobalImportCommon.FileType fileType,
GlobalImportCommon.FileType expectedFileType)
{
return(header == GlobalImportCommon.Header && fileType == expectedFileType);
}
/// <summary>
/// returns true if this is a valid VEP reader file (text)
/// </summary>
private static bool IsValidFile(string line, GlobalImportCommon.FileType fileType)
{
string expectedString = $"{GlobalImportCommon.Header}\t{(byte)fileType}";
return(line == expectedString);
}