public static void WriteBases(DataBundle bundle, List <TranscriptPacket> packets, string outputStub,
List <string> outputFiles)
{
var refPath = outputStub + ".bases" + TempCacheExt;
outputFiles.Add(refPath);
const int flankingLength = 500;
using (var writer = new CompressedSequenceWriter(refPath, bundle.SequenceReader.Metadata,
bundle.Sequence.CytogeneticBands, bundle.Cache.Header.GenomeAssembly))
{
for (ushort refIndex = 0; refIndex < bundle.Sequence.Renamer.NumRefSeqs; refIndex++)
{
var interval = GetBoundingInterval(packets, refIndex, flankingLength);
if (interval == null)
{
continue;
}
LoadSequence(bundle, refIndex);
var bases = bundle.Sequence.Substring(interval.Start, interval.End - interval.Start + 1);
var ensemblRefName = bundle.Sequence.Renamer.EnsemblReferenceNames[refIndex];
writer.Write(ensemblRefName, bases, interval.Start);
}
}
}
private static void WriteReference(ILogger logger, string outputPath, CompressedSequenceReader reader,
IChromosome chromosome, string referenceBases, int offset)
{
logger.Write("- writing reference bases... ");
var cytogeneticBands = new CytogeneticBands(reader.CytogeneticBands);
using (var writer = new CompressedSequenceWriter(FileUtilities.GetCreateStream(outputPath),
reader.ReferenceMetadataList, cytogeneticBands, reader.Assembly))
{
writer.Write(chromosome.EnsemblName, referenceBases, offset);
}
logger.WriteLine("finished.");
}
/// <summary>
/// this function is used when we want to populate the chromosome renamer and the cytogenetic bands
/// </summary>
public static void WriteEmptyBases(DataBundle bundle, HashSet <ushort> refIndices, string outputStub,
List <string> outputFiles)
{
var refPath = outputStub + ".bases" + TempCacheExt;
outputFiles.Add(refPath);
using (var writer = new CompressedSequenceWriter(refPath, bundle.SequenceReader.Metadata,
bundle.Sequence.CytogeneticBands, bundle.Cache.Header.GenomeAssembly))
{
foreach (var refIndex in refIndices.OrderBy(x => x))
{
var ensemblRefName = bundle.Sequence.Renamer.EnsemblReferenceNames[refIndex];
writer.Write(ensemblRefName, "A");
}
}
}
/// <summary>
/// converts the FASTA file to a compressed reference file
/// </summary>
public void Convert(string inputFastaPath, string inputCytogeneticBandpath, string inputChromosomeNamesPath,
string outputCompressedPath, GenomeAssembly genomeAssembly)
{
Console.Write("- getting reference metadata... ");
var referenceMetaDataList = GetReferenceMetadata(inputChromosomeNamesPath);
Console.WriteLine("{0} references found.", referenceMetaDataList.Count);
var renamer = new ChromosomeRenamer();
renamer.AddReferenceMetadata(referenceMetaDataList);
// pre-allocate the cytogenetic bands
Console.Write("- getting cytogenetic bands... ");
var cytogeneticBands = GetCytogeneticBands(inputCytogeneticBandpath, renamer);
Console.WriteLine("finished.\n");
// parse the reference
using (var fastaReader = new FastaReader(inputFastaPath))
{
using (var writer = new CompressedSequenceWriter(outputCompressedPath, referenceMetaDataList, cytogeneticBands, genomeAssembly))
{
Console.WriteLine("Converting the following reference sequences:");
while (true)
{
var referenceSequence = fastaReader.GetReferenceSequence();
if (referenceSequence == null)
{
break;
}
Console.WriteLine("- {0} ({1:n0} bytes)", referenceSequence.Name, referenceSequence.Bases.Length);
writer.Write(referenceSequence.Name, referenceSequence.Bases);
}
}
}
Console.WriteLine("\nFile size: {0}", new FileInfo(outputCompressedPath).Length);
}