public void BuildGraph_GenerateDotFile_WriteContigs(string fastaPath)
{
var assemblyName = "IntegrationTests";
var projectPath = Environment.CurrentDirectory.Substring(0, Environment.CurrentDirectory.IndexOf(assemblyName) + assemblyName.Length);
fastaPath = Path.Combine(projectPath, fastaPath);
var fileService = new FileService();
var kmerLength = 19;
var errorCorrector = new ErrorCorrector(kmerLength);
var fastaReader = new FastaReader(fileService);
var reads = fastaReader.ParseFastaFile(fastaPath);
errorCorrector.BuildHistogram(reads);
var graphBuilder = new DeBruijnGraphBuilder(kmerLength, errorCorrector);
var graph = graphBuilder.Build(reads);
errorCorrector.PrintResult();
graph.CleanUp();
var dotFileDirectory = Path.Combine(Path.GetDirectoryName(fastaPath), "graphs");
Directory.CreateDirectory(dotFileDirectory);
graphBuilder.ToDot(fileService, Path.Combine(dotFileDirectory, Path.GetFileNameWithoutExtension(fastaPath) + ".dot"), graph);
var contigs = graph.GetContigs();
var contigsDirectory = Path.Combine(Path.GetDirectoryName(fastaPath), "contigs");
Directory.CreateDirectory(contigsDirectory);
fastaReader.WriteFastaFile(Path.Combine(contigsDirectory, Path.GetFileNameWithoutExtension(fastaPath) + ".contigs.fasta"), contigs);
}
private void Run()
{
var fileService = new FileService();
var kmerLength = K ?? 19;
var errorCorrector = new ErrorCorrector(kmerLength, true);
var fastaReader = new FastaReader(fileService);
var reads = fastaReader.ParseFastaFile(ReadsPath);
// errorCorrector.BuildHistogram(reads);
var graphBuilder = new DeBruijnGraphBuilder(kmerLength, errorCorrector);
var graph = graphBuilder.Build(reads);
errorCorrector.PrintResult();
graph.CleanUp();
if (!string.IsNullOrWhiteSpace(DotFilePath))
{
Directory.CreateDirectory(Path.GetDirectoryName(DotFilePath));
graphBuilder.ToDot(fileService, DotFilePath, graph);
}
var contigs = graph.GetContigs();
Directory.CreateDirectory(Path.GetDirectoryName(ContigsPath));
fastaReader.WriteFastaFile(ContigsPath, contigs);
}
public void ParseFastaFile_EmptyFile_ThrowsArgumentException()
{
var fileReader = new TestFileService();
fileReader.fileLines = new List <string>();
var fastaReader = new FastaReader(fileReader);
Assert.Throws <ArgumentException>(() =>
fastaReader.ParseFastaFile("path").ToList());
}
public void ParseFastaFile_ReturnsCorrectSequences()
{
var fileReader = new TestFileService();
var fastaReader = new FastaReader(fileReader);
var sequences = fastaReader.ParseFastaFile("path").ToList();
Assert.Equal(10, sequences.Count);
Assert.All(sequences, s => Assert.Equal(80, s.Length));
}
public void ParseFastaFile_NoSequencesInFile_ThrowsArgumentException()
{
var fileReader = new TestFileService();
fileReader.fileLines = new List <string>()
{
"this is definitely not a fasta file",
"ATCGCTGJGJGNVBCACABJP",
"dvdfvdf fsdfs dfsdfsgdf"
};
var fastaReader = new FastaReader(fileReader);
Assert.Throws <ArgumentException>(() =>
fastaReader.ParseFastaFile("path").ToList());
}
public void Correct_RealData(string fastaPath)
{
var assemblyName = "IntegrationTests";
var projectPath = Environment.CurrentDirectory.Substring(0, Environment.CurrentDirectory.IndexOf(assemblyName) + assemblyName.Length);
fastaPath = Path.Combine(projectPath, fastaPath);
var kmerLength = 19;
var corrector = new ErrorCorrector(kmerLength);
var fastaReader = new FastaReader(new FileService());
var reads = fastaReader.ParseFastaFile(fastaPath);
var histogram = corrector.BuildHistogram(reads);
Console.WriteLine("Before correction");
WriteLowCountDistinctKmersCount(histogram, 12);
var kmersCorrected = corrector.CorrectReadsAndSplitToKmers(reads);
var histogramCorrected = corrector.BuildHistogram(kmersCorrected.Select(k => k.ToString()));
Console.WriteLine("After correction");
WriteLowCountDistinctKmersCount(histogramCorrected, 12);
corrector.PrintResult();
}