private static ExitCodes ProgramExecution()
{
var genomeAssembly = GenomeAssemblyHelper.Convert(_genomeAssembly);
Console.Write("- loading previous reference names... ");
List <IChromosome> oldChromosomes = ReferenceNamesReader.GetReferenceNames(FileUtilities.GetReadStream(_referenceNamesPath));
Console.WriteLine("finished.");
IDictionary <string, IChromosome> oldRefNameToChromosome = ReferenceDictionaryUtils.GetRefNameToChromosome(oldChromosomes);
Console.Write("- reading the genome assembly report... ");
List <IChromosome> chromosomes = AssemblyReader.GetChromosomes(FileUtilities.GetReadStream(_genomeAssemblyReportPath), oldRefNameToChromosome, oldChromosomes.Count);
int numRefSeqs = chromosomes.Count;
Console.WriteLine($"{numRefSeqs} references found.");
Console.Write("- checking reference index contiguity... ");
CheckReferenceIndexContiguity(chromosomes, oldChromosomes);
Console.WriteLine("contiguous.");
IDictionary <string, IChromosome> refNameToChromosome = ReferenceDictionaryUtils.GetRefNameToChromosome(chromosomes);
Console.Write("- reading cytogenetic bands... ");
List <Band>[] cytogeneticBandsByRef = CytogeneticBandsReader.GetCytogeneticBands(FileUtilities.GetReadStream(_cytogeneticBandPath),
numRefSeqs, refNameToChromosome);
Console.WriteLine("finished.");
Console.WriteLine("- reading FASTA files:");
List <FastaSequence> fastaSequences = GetFastaSequences(_fastaPrefix, refNameToChromosome);
long genomeLength = GetGenomeLength(fastaSequences);
Console.WriteLine($"- genome length: {genomeLength:N0}");
Console.Write("- check if chrY has PAR masking... ");
CheckChrYPadding(fastaSequences);
Console.WriteLine("unmasked.");
Console.Write("- applying 2-bit compression... ");
List <Creation.ReferenceSequence> referenceSequences = CreateReferenceSequences(fastaSequences, cytogeneticBandsByRef);
Console.WriteLine("finished.");
Console.Write("- creating reference sequence file... ");
CreateReferenceSequenceFile(genomeAssembly, _patchLevel, chromosomes, referenceSequences);
long fileSize = new FileInfo(_outputCompressedPath).Length;
Console.WriteLine($"{fileSize:N0} bytes");
return(ExitCodes.Success);
}
private static ExitCodes ProgramExecution()
{
var genomeAssembly = GenomeAssemblyHelper.Convert(_genomeAssembly);
Console.Write("- reading the genome assembly report... ");
var dummyRefNameToChromosome = new Dictionary <string, IChromosome>();
List <IChromosome> chromosomes = AssemblyReader.GetChromosomes(FileUtilities.GetReadStream(_genomeAssemblyReportPath), dummyRefNameToChromosome, 0);
int numRefSeqs = chromosomes.Count;
Console.WriteLine($"{numRefSeqs} references found.");
IDictionary <string, IChromosome> refNameToChromosome = ReferenceDictionaryUtils.GetRefNameToChromosome(chromosomes);
Console.Write("- reading FASTA file... ");
var fastaSequence = GetFastaSequence(_fastaPath, refNameToChromosome);
Console.WriteLine($"- sequence length: {fastaSequence.Bases.Length:N0}");
Console.Write("- reading cytogenetic bands... ");
List <Band> cytogeneticBands = GetCytogeneticBands(fastaSequence.Chromosome.Index, numRefSeqs, refNameToChromosome);
Console.WriteLine("finished.");
Console.Write("- applying 2-bit compression... ");
var referenceSequence = CreateReferenceSequence(fastaSequence, cytogeneticBands);
Console.WriteLine("finished.");
Console.Write("- creating reference sequence file... ");
var minimalChromosomes = new List <IChromosome> {
fastaSequence.Chromosome
};
CreateReferenceSequenceFile(genomeAssembly, minimalChromosomes, referenceSequence);
long fileSize = new FileInfo(_outputCompressedPath).Length;
Console.WriteLine($"{fileSize:N0} bytes");
return(ExitCodes.Success);
}