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);
}
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();
}
public void Correct(string read, string expectedRead, int numberOfErrors)
{
var corrector = new ErrorCorrector(3);
var reads = new List <string> {
read
};
var expectedReads = new List <string> {
expectedRead
};
corrector.BuildHistogram(reads);
var expectedKmers = new TestErrorCorrector(3).CorrectReadsAndSplitToKmers(expectedReads).ToList();
var correctedKmers = corrector.CorrectReadsAndSplitToKmers(reads).ToList();
Assert.Equal(expectedKmers, correctedKmers);
Assert.Equal(numberOfErrors, corrector.CorrectedKmersCount);
}
public void Histogram()
{
var corrector = new ErrorCorrector(3);
var reads = new List <string>
{
string.Join("", Enumerable.Repeat("CAT", 10))
};
var histogram = corrector.BuildHistogram(reads);
var expected = new Dictionary <string, int>
{
{ "ATC", 9 },
{ "CAT", 10 },
{ "TCA", 9 }
};
Assert.Equal(expected, histogram);
}