/// <summary>
/// The execution method for the activity.
/// </summary>
/// <param name="executionContext">The execution context.</param>
/// <returns>The execution status.</returns>
protected override ActivityExecutionStatus Execute(ActivityExecutionContext executionContext)
{
FastAFormatter formatter = new FastAFormatter();
formatter.Open(OutputFile);
if ((Sequence == null) && (SequenceList != null))
{
foreach (ISequence sequence in SequenceList)
{
formatter.Write(sequence);
}
}
else if ((Sequence != null) && (SequenceList == null))
{
formatter.Write(Sequence);
}
else if ((Sequence != null) && (SequenceList != null))
{
foreach (ISequence sequence in SequenceList)
{
formatter.Write(sequence);
}
formatter.Write(Sequence);
}
formatter.Close();
return(ActivityExecutionStatus.Closed);
}
/// <summary>
/// It writes Contigs to the file.
/// </summary>
/// <param name="scaffolds">The list of scaffolds sequence.</param>
private void WriteContigs(IList <ISequence> scaffolds)
{
if (scaffolds.Count == 0)
{
Output.WriteLine(OutputLevel.Information, "No Scaffolds generated.");
return;
}
EnsureContigNames(scaffolds);
if (!string.IsNullOrEmpty(this.OutputFile))
{
FastAFormatter formatter = new FastAFormatter {
AutoFlush = true
};
using (formatter.Open(this.OutputFile))
{
formatter.Format(scaffolds);
}
Output.WriteLine(OutputLevel.Information, "Wrote {0} scaffolds to {1}", scaffolds.Count, this.OutputFile);
}
else
{
Output.WriteLine(OutputLevel.Information, "Scaffold Results: {0} sequences", scaffolds.Count);
FastAFormatter formatter = new FastAFormatter {
MaxSymbolsAllowedPerLine = Math.Min(80, Console.WindowWidth - 2),
AutoFlush = true
};
formatter.Format(Console.OpenStandardOutput(), scaffolds);
}
}
/// <summary>
/// Writes the contigs to the file.
/// </summary>
/// <param name="assembly">IDeNovoAssembly parameter is the result of running De Novo Assembly on a set of two or more sequences. </param>
protected void writeContigs(PadenaAssembly assembly)
{
if (assembly.AssembledSequences.Count == 0)
{
Output.WriteLine(OutputLevel.Results, "\tNo sequences assembled.");
return;
}
ensureContigNames(assembly.AssembledSequences);
if (!string.IsNullOrEmpty(this.DiagnosticFilePrefix))
{
using (FastAFormatter formatter = new FastAFormatter(ContigFileName)) {
formatter.AutoFlush = true;
foreach (ISequence seq in assembly.AssembledSequences)
{
formatter.Write(seq);
}
}
Output.WriteLine(OutputLevel.Information, "\tWrote {0} sequences to {1}", assembly.AssembledSequences.Count, ContigFileName);
}
else
{
Output.WriteLine(OutputLevel.Information, "\tAssembled Sequence Results: {0} sequences", assembly.AssembledSequences.Count);
using (FastAFormatter formatter = new FastAFormatter()) {
formatter.Open(new StreamWriter(Console.OpenStandardOutput()));
formatter.MaxSymbolsAllowedPerLine = decideOutputWidth();
formatter.AutoFlush = true;
foreach (ISequence seq in assembly.AssembledSequences)
{
formatter.Write(seq);
}
}
}
}
/// <summary>
/// Write sequences to the file
/// </summary>
/// <param name="sequences"></param>
private void WriteSequences(IEnumerable <ISequence> sequences)
{
if (!string.IsNullOrEmpty(this.OutputFile))
{
int count = 0;
var formatter = new FastAFormatter {
AutoFlush = true
};
using (formatter.Open(this.OutputFile))
{
foreach (ISequence sequence in sequences)
{
count++;
formatter.Format(sequence);
}
}
Output.WriteLine(OutputLevel.Information, "Wrote {0} sequences to {1}.", count, this.OutputFile);
}
else
{
Output.WriteLine(OutputLevel.Information, "Results:");
foreach (ISequence seq in sequences)
{
Output.WriteLine(OutputLevel.Results, seq.ID);
Output.WriteLine(OutputLevel.Results, new string(seq.Select(a => (char)a).ToArray()));
}
}
}
public void FastAFormatterValidateWriteWithStream()
{
string actualSequence = string.Empty;
using (var formatter = new FastAFormatter())
{
using (var writer = new StreamWriter(Constants.FastaTempFileName))
{
formatter.Open(writer);
// Gets the actual sequence and the alphabet from the Xml
actualSequence = utilityObj.xmlUtil.GetTextValue(Constants.SimpleFastaNodeName,
Constants.ExpectedSequenceNode);
string alpName = utilityObj.xmlUtil.GetTextValue(Constants.SimpleFastaNodeName,
Constants.AlphabetNameNode);
// Logs information to the log file
ApplicationLog.WriteLine(string.Format(null,
"FastA Formatter BVT: Validating with Sequence '{0}' and Alphabet '{1}'.",
actualSequence, alpName));
var seqOriginal = new Sequence(Utility.GetAlphabet(alpName),
actualSequence);
seqOriginal.ID = "";
Assert.IsNotNull(seqOriginal);
// Use the formatter to write the original sequences to a stream.
ApplicationLog.WriteLine(string.Format(null,
"FastA Formatter BVT: Creating the Temp file '{0}'.",
Constants.FastaTempFileName));
formatter.Write(seqOriginal);
formatter.Close();
}
IEnumerable <ISequence> seq = null;
using (var reader = new StreamReader(Constants.FastaTempFileName))
{
// Read the new file, then compare the sequences
using (var parser = new FastAParser())
{
parser.Alphabet = Alphabets.Protein;
seq = parser.Parse(reader);
//Create a list of sequences.
List <ISequence> seqsList = seq.ToList();
Assert.IsNotNull(seqsList);
var seqString = new string(seqsList[0].Select(a => (char)a).ToArray());
Assert.AreEqual(actualSequence, seqString);
}
}
// Passed all the tests, delete the tmp file. If we failed an Assert,
// the tmp file will still be there in case we need it for debugging.
File.Delete(Constants.FastaTempFileName);
ApplicationLog.WriteLine("Deleted the temp file created.");
}
}
public static async void AssemblySequences(string fastqFileName)
{
var parser = new FastQParser();
List <IQualitativeSequence> sequences = new List <IQualitativeSequence>();
using (var fileStream = new FileStream(fastqFileName, FileMode.Open))
{
sequences = parser.Parse(fileStream).ToList();
}
OverlapDeNovoAssembler assembler = new OverlapDeNovoAssembler();
IDeNovoAssembly assembly = assembler.Assemble(sequences);
FastAFormatter outputFormatter = new FastAFormatter();
outputFormatter.Open("assembled_sequences.fasta");
outputFormatter.Format(assembly.AssembledSequences);
outputFormatter.Close();
}
/// <summary>
/// Writes ambiguous reads that are filtered out to the specified file.
/// </summary>
/// <param name="ambiguousReads">Reads with ambiguous symbols.</param>
/// <param name="ambiguousFilename">File to write.</param>
private static void WriteAmbiguousReads(BlockingCollection <ISequence> ambiguousReads, string ambiguousFilename)
{
FastAFormatter formatter = new FastAFormatter()
{
AutoFlush = true
};
using (formatter.Open(ambiguousFilename))
{
while (!ambiguousReads.IsCompleted)
{
ISequence seq;
if (ambiguousReads.TryTake(out seq, -1))
{
formatter.Format(seq);
}
}
}
}
/// <summary>
/// It Writes the contigs to the file.
/// </summary>
/// <param name="assembly">IDeNovoAssembly parameter is the result of running De Novo Assembly on a set of two or more sequences. </param>
protected void WriteContigs(IDeNovoAssembly assembly)
{
if (assembly.AssembledSequences.Count == 0)
{
Output.WriteLine(OutputLevel.Results, "No sequences assembled.");
return;
}
EnsureContigNames(assembly.AssembledSequences);
if (!string.IsNullOrEmpty(this.OutputFile))
{
using (FastAFormatter formatter = new FastAFormatter(this.OutputFile))
{
formatter.AutoFlush = true;
foreach (ISequence seq in assembly.AssembledSequences)
{
formatter.Write(seq);
}
}
Output.WriteLine(OutputLevel.Information, "Wrote {0} sequences to {1}", assembly.AssembledSequences.Count, this.OutputFile);
}
else
{
Output.WriteLine(OutputLevel.Information, "Assembled Sequence Results: {0} sequences", assembly.AssembledSequences.Count);
using (FastAFormatter formatter = new FastAFormatter())
{
formatter.Open(new StreamWriter(Console.OpenStandardOutput()));
formatter.MaxSymbolsAllowedPerLine = Math.Min(80, Console.WindowWidth - 2);
formatter.AutoFlush = true;
foreach (ISequence seq in assembly.AssembledSequences)
{
formatter.Write(seq);
}
}
}
}
/// <summary>
/// It Writes the contigs to the file.
/// </summary>
/// <param name="assembly">IDeNovoAssembly parameter is the result of running De Novo Assembly on a set of two or more sequences. </param>
protected void WriteContigs(IEnumerable <ISequence> assembly)
{
int counter = 1;
if (!string.IsNullOrEmpty(this.OutputFile))
{
FastAFormatter formatter = new FastAFormatter {
AutoFlush = true
};
using (formatter.Open(this.OutputFile))
{
foreach (ISequence seq in assembly)
{
if (string.IsNullOrEmpty(seq.ID))
{
seq.ID = GenerateSequenceId(counter);
}
formatter.Format(seq);
counter++;
}
}
Output.WriteLine(OutputLevel.Information, Resources.OutPutWrittenToFileSpecified);
}
else
{
Output.WriteLine(OutputLevel.Information, "Assembled Sequence Results:");
foreach (ISequence seq in assembly)
{
if (string.IsNullOrEmpty(seq.ID))
{
seq.ID = GenerateSequenceId(counter);
}
Output.WriteLine(OutputLevel.Results, seq.ID);
Output.WriteLine(OutputLevel.Results, new string(seq.Select(a => (char)a).ToArray()));
counter++;
}
}
}
/// <summary>
/// Filters the test data for the input file
/// </summary>
/// <param name="inputFile">Input File</param>
/// <param name="outputFile">Output File</param>
/// <param name="repeatLength">Repeat Length</param>
static void FilterTestData(string inputFile, string outputFile, int repeatLength)
{
if (File.Exists(inputFile))
{
Console.WriteLine("Processing the file '{0}'.", inputFile);
// Read the inputfile with the help of FastA Parser
FastAParser parserObj = new FastAParser();
FastAFormatter outputWriter = new FastAFormatter();
using (parserObj.Open(inputFile))
using (outputWriter.Open(outputFile))
{
IEnumerable <ISequence> inputReads = parserObj.Parse();
// Going through read by read in a given file
foreach (ISequence seq in inputReads)
{
// Get the First read in the file
byte[] actualRead = seq.ToArray();
// Assign the temporary local variables required
byte previousChar = actualRead[0];
int repeatLenCount = 0;
bool ignoreRead = false;
// Go through each and every character/byte in the read
for (int j = 1; j < actualRead.Length; j++)
{
// Check if the previous character is same as current.
if (previousChar == actualRead[j])
{
repeatLenCount++;
// if repeat length exceeds, skip this read and continue with other read
if (repeatLenCount == repeatLength)
{
Console.WriteLine("Character '{0}' repeated more than '{1}' times and read '{2}' is skipped",
(char)previousChar, repeatLength, seq.ID);
ignoreRead = true;
break;
}
continue;
}
repeatLenCount = 0;
previousChar = actualRead[j];
}
Console.WriteLine("Read '{0}' Processed.", seq.ID);
// Check if the length exceeds the max length and write it to the output file
if (!ignoreRead)
{
outputWriter.Format(seq);
}
}
}
Console.WriteLine();
Console.WriteLine("Filtering Completed!!");
}
else
{
Console.WriteLine("Enter Valid File Path.");
}
}
/// <summary>
/// Does the logic behind the sequence simulation
/// </summary>
public void DoSimulation(string outputFileName, Action <long, long> updateSimulationStats, Action <string> simulationComplete)
{
const string filePostfix = "_{0}.fa";
FileInfo file = new FileInfo(outputFileName);
if (file.Directory == null || !file.Directory.Exists)
{
throw new ArgumentException("Could not write to the output directory for " + outputFileName);
}
if (Settings.OutputSequenceCount <= 0)
{
throw new ArgumentException("'Max Output Sequences Per File' should be greater than zero.");
}
if (Settings.SequenceLength <= 0)
{
throw new ArgumentException("'Mean Output Length' should be greater than zero.");
}
string filePrefix = String.IsNullOrEmpty(file.Extension) ? file.FullName : file.FullName.Substring(0, file.FullName.IndexOf(file.Extension));
long seqCount = (Settings.DepthOfCoverage * SequenceToSplit.Count) / Settings.SequenceLength;
long fileCount = seqCount / Settings.OutputSequenceCount;
if (seqCount % Settings.OutputSequenceCount != 0)
{
fileCount++;
}
// Update the UI
updateSimulationStats(seqCount, fileCount);
int fileIndex = 1;
FastAFormatter formatter = null;
List <ISequence> generatedSequenceList = new List <ISequence>();
for (long i = 0; i < seqCount; i++)
{
generatedSequenceList.Add(CreateSubsequence(i, SequenceToSplit, Settings));
if (generatedSequenceList.Count >= Settings.OutputSequenceCount)
{
FileInfo outFile = new FileInfo(filePrefix + string.Format(filePostfix, fileIndex++));
formatter = new FastAFormatter();
using (formatter.Open(outFile.FullName))
{
formatter.Format(generatedSequenceList);
}
generatedSequenceList.Clear();
}
}
// Pick off any remaining sequences into the final file.
if (generatedSequenceList.Count > 0)
{
FileInfo outFile = new FileInfo(filePrefix + string.Format(filePostfix, fileIndex++));
formatter = new FastAFormatter();
using (formatter.Open(outFile.FullName))
{
formatter.Format(generatedSequenceList);
}
simulationComplete(formatter.Name);
}
// Either we ended exactly on the boundary with no additional sequences
// generated, OR we never generated any files.
else
{
simulationComplete(formatter != null ? formatter.Name : string.Empty);
}
}
public void FastAFormatterValidateWrite1()
{
var formatter = new FastAFormatter();
using (formatter.Open(Constants.FastaTempFileName))
{
// Gets the actual sequence and the alphabet from the Xml
string actualSequence = utilityObj.xmlUtil.GetTextValue(Constants.SimpleFastaNodeName,
Constants.ExpectedSequenceNode);
string alpName = utilityObj.xmlUtil.GetTextValue(Constants.SimpleFastaNodeName,
Constants.AlphabetNameNode);
// Logs information to the log file
ApplicationLog.WriteLine(string.Format("FastA Formatter BVT: Validating with Sequence '{0}' and Alphabet '{1}'.",
actualSequence, alpName));
var seqOriginal = new Sequence(Utility.GetAlphabet(alpName),
actualSequence)
{
ID = ""
};
Assert.IsNotNull(seqOriginal);
// Use the formatter to write the original sequences to a temp file
ApplicationLog.WriteLine(string.Format("FastA Formatter BVT: Creating the Temp file '{0}'.",
Constants.FastaTempFileName));
var seqList = new List <ISequence> {
seqOriginal, seqOriginal, seqOriginal
};
formatter.Format(seqList);
formatter.Close();
IEnumerable <ISequence> seqsNew = null;
// Read the new file, then compare the sequences
var parser = new FastAParser();
{
parser.Alphabet = Alphabets.Protein;
seqsNew = parser.Parse(Constants.FastaTempFileName);
char[] seqString = seqsNew.ElementAt(0).Select(a => (char)a).ToArray();
var newSequence = new string(seqString);
Assert.IsNotNull(seqsNew);
ApplicationLog.WriteLine(string.Format(null,
"FastA Formatter BVT: New Sequence is '{0}'.",
newSequence));
// Now compare the sequences.
int countNew = seqsNew.Count();
Assert.AreEqual(3, countNew);
ApplicationLog.WriteLine("The Number of sequences are matching.");
Assert.AreEqual(seqOriginal.ID, seqsNew.ElementAt(0).ID);
Assert.AreEqual(new string(seqsNew.ElementAt(0).Select(a => (char)a).ToArray()), newSequence);
ApplicationLog.WriteLine(string.Format(null,
"FastA Formatter BVT: The FASTA sequences '{0}' are matching with Format() method.",
newSequence));
// Passed all the tests, delete the tmp file. If we failed an Assert,
// the tmp file will still be there in case we need it for debugging.
File.Delete(Constants.FastaTempFileName);
ApplicationLog.WriteLine("Deleted the temp file created.");
}
}
}