Class for reading/validating a fastq file. More...

#include <FastQFile.h>

Collaboration diagram for FastQFile:

Public Member Functions
	FastQFile (int minReadLength=10, int numPrintableErrors=20)
	Constructor.
void	disableMessages ()
	Disable messages - do not write to cout.
void	enableMessages ()
	Enable messages - write to cout.
void	disableSeqIDCheck ()
	Disable Unique Sequence ID checking (Unique Sequence ID checking is enabled by default).
void	enableSeqIDCheck ()
	Enable Unique Sequence ID checking.
void	setMaxErrors (int maxErrors)
	Set the number of errors after which to quit reading/validating a file, defaults to -1.
FastQStatus::Status	openFile (const char *fileName, BaseAsciiMap::SPACE_TYPE spaceType=BaseAsciiMap::UNKNOWN)
	Open a FastQFile.
FastQStatus::Status	closeFile ()
	Close a FastQFile.
bool	isOpen ()
	Check to see if the file is open.
bool	isEof ()
	Check to see if the file is at the end of the file.
bool	keepReadingFile ()
	Returns whether or not to keep reading the file, it stops reading (false) if eof or there is a problem reading the file.
FastQStatus::Status	validateFastQFile (const String &filename, bool printBaseComp, BaseAsciiMap::SPACE_TYPE spaceType, bool printQualAvg=false)
	Validate the specified fastq file.
FastQStatus::Status	readFastQSequence ()
	Read 1 FastQSequence, validating it.
BaseAsciiMap::SPACE_TYPE	getSpaceType ()
	Get the space type used for this file.
Public Attributes
Public Sequence Line variables.
Keep public variables for a sequence's line so they can be accessed without having to do string copies.
String	myRawSequence
String	mySequenceIdLine
String	mySequenceIdentifier
String	myPlusLine
String	myQualityString

Detailed Description

Class for reading/validating a fastq file.

Definition at line 29 of file FastQFile.h.

Constructor & Destructor Documentation

FastQFile::FastQFile	(	int	minReadLength = `10`,
		int	numPrintableErrors = `20`
	)

Constructor.

/param minReadLength The minimum length that a base sequence must be for it to be valid.

Parameters:

numPrintableErrors The maximum number of errors that should be reported in detail before suppressing the errors.

Definition at line 30 of file FastQFile.cpp.

   : myFile(NULL),
     myBaseComposition(),
     myQualPerCycle(),
     myCountPerCycle(),
     myCheckSeqID(true),
     myMinReadLength(minReadLength),
     myNumPrintableErrors(numPrintableErrors),
     myMaxErrors(-1),
     myDisableMessages(false),
     myFileProblem(false)
{
   // Reset the member data.
   reset();
}

Member Function Documentation

void FastQFile::disableSeqIDCheck ( )

Disable Unique Sequence ID checking (Unique Sequence ID checking is enabled by default).

Definition at line 61 of file FastQFile.cpp.

{
    myCheckSeqID = false;
}

void FastQFile::enableSeqIDCheck ( )

Enable Unique Sequence ID checking.

(Unique Sequence ID checking is enabled by default).

Definition at line 69 of file FastQFile.cpp.

{
    myCheckSeqID = true;
}

bool FastQFile::keepReadingFile ( )

Returns whether or not to keep reading the file, it stops reading (false) if eof or there is a problem reading the file.

Definition at line 184 of file FastQFile.cpp.

References isEof().

Referenced by validateFastQFile().

{
   if(isEof() || myFileProblem)
   {
      return(false);
   }
   return(true);
}

FastQStatus::Status FastQFile::openFile	(	const char *	fileName,
		BaseAsciiMap::SPACE_TYPE	spaceType = `BaseAsciiMap::UNKNOWN`
	)

Open a FastQFile.

Use the specified SPACE_TYPE to determine BASE, COLOR, or UNKNOWN.

Definition at line 83 of file FastQFile.cpp.

References closeFile(), FastQStatus::FASTQ_OPEN_ERROR, FastQStatus::FASTQ_SUCCESS, ifopen(), BaseComposition::resetBaseMapType(), and BaseComposition::setBaseMapType().

Referenced by validateFastQFile().

{
   // reset the member data.
   reset();

   myBaseComposition.resetBaseMapType();
   myBaseComposition.setBaseMapType(spaceType);
   myQualPerCycle.clear();
   myCountPerCycle.clear();

   FastQStatus::Status status = FastQStatus::FASTQ_SUCCESS;

   // Close the file if there is already one open - checked by close.
   status = closeFile();
   if(status == FastQStatus::FASTQ_SUCCESS)
   {
      // Successfully closed a previously opened file if there was one.
      
      // Open the file
      myFile = ifopen(fileName, "rt");
      myFileName = fileName;
      
      if(myFile == NULL)
      {
         // Failed to open the file.
         status = FastQStatus::FASTQ_OPEN_ERROR;
      }
   }

   if(status != FastQStatus::FASTQ_SUCCESS)
   {
      // Failed to open the file.
      std::string errorMessage = "ERROR: Failed to open file: ";
      errorMessage += fileName;
      logMessage(errorMessage.c_str());
   }
   return(status);
}

void FastQFile::setMaxErrors ( int maxErrors )

Set the number of errors after which to quit reading/validating a file, defaults to -1.

Parameters:

maxErrors # of errors before quitting, -1 indicates to not quit until the entire file has been read/validated (default), 0 indicates to quit without reading/validating anything.

Definition at line 76 of file FastQFile.cpp.

{
   myMaxErrors = maxErrors;
}

FastQStatus::Status FastQFile::validateFastQFile	(	const String &	filename,
		bool	printBaseComp,
		BaseAsciiMap::SPACE_TYPE	spaceType,
		bool	printQualAvg = `false`
	)

Validate the specified fastq file.

Parameters:

filename	fastq file to be validated.
printBaseComp	whether or not to print the base composition for the file. true means print it, false means do not.
spaceType	the spaceType to use for validation - BASE_SPACE, COLOR_SPACE, or UNKNOWN (UNKNOWN means to determine the spaceType to validate against from the first character of the first sequence).
printQualAvg	whether or not to print the quality averages for the file. true means to print it, false (default) means do not.

Returns:: the fastq validation status, SUCCESS on a successfully validated fastq file.

Definition at line 195 of file FastQFile.cpp.

References closeFile(), FastQStatus::FASTQ_INVALID, FastQStatus::FASTQ_NO_SEQUENCE_ERROR, FastQStatus::FASTQ_OPEN_ERROR, FastQStatus::FASTQ_SUCCESS, keepReadingFile(), openFile(), BaseComposition::print(), and readFastQSequence().

{
   // Open the fastqfile.
   if(openFile(filename, spaceType) != FastQStatus::FASTQ_SUCCESS)
   {
      // Failed to open the specified file.
      return(FastQStatus::FASTQ_OPEN_ERROR);
   }

   // Track the total number of sequences that were validated.
   int numSequences = 0;

   // Keep reading the file until there are no more fastq sequences to process
   // and not configured to quit after a certain number of errors or there
   // has not yet been that many errors.
   // Or exit if there is a problem reading the file.
   FastQStatus::Status status = FastQStatus::FASTQ_SUCCESS;
   while (keepReadingFile() &&
          ((myMaxErrors == -1) || (myMaxErrors > myNumErrors)))
   {
      // Validate one sequence.  This call will read all the lines for 
      // one sequence.
      status = readFastQSequence();
      if((status == FastQStatus::FASTQ_SUCCESS) || (status == FastQStatus::FASTQ_INVALID))
      {
         // Read a sequence and it is either valid or invalid, but
         // either way, a sequence was read, so increment the sequence count.
         ++numSequences;
      }
      else
      {
         // Other error, so break out of processing.
         break;
      }
   }
   
   // Report Base Composition Statistics.
   if(printBaseComp)
   {
      myBaseComposition.print();
   }

   if(printQualAvg)
   {
      printAvgQual();
   }

   std::string finishMessage = "Finished processing ";
   finishMessage += myFileName.c_str();
   char buffer[100];
   if(sprintf(buffer, 
              " with %u lines containing %d sequences.", 
              myLineNum, numSequences) > 0)
   {
      finishMessage += buffer;
      logMessage(finishMessage.c_str());
   }
   if(sprintf(buffer, 
              "There were a total of %d errors.", 
              myNumErrors) > 0)
   {
      logMessage(buffer);
   }

   // Close the input file.
   FastQStatus::Status closeStatus = closeFile();

   if((status != FastQStatus::FASTQ_SUCCESS) && (status != FastQStatus::FASTQ_INVALID))
   {
      // Stopped validating due to some error other than invalid, so
      // return that error.
      return(status);
   }
   else if(myNumErrors == 0)
   {
      // No errors, check to see if there were any sequences.
      // Finished processing all of the sequences in the file.
      // If there are no sequences, report an error.
      if(numSequences == 0)
      {
         // Empty file, return error.
         logMessage("ERROR: No FastQSequences in the file.");
         return(FastQStatus::FASTQ_NO_SEQUENCE_ERROR);
      }
      return(FastQStatus::FASTQ_SUCCESS);
   }
   else
   {
      // The file is invalid.  But check the close status.  If the close
      // failed, it means there is a problem with the file itself not just
      // with validation, so the close failure should be returned.
      if(closeStatus != FastQStatus::FASTQ_SUCCESS)
      {
         return(closeStatus);
      }
      return(FastQStatus::FASTQ_INVALID);
   }
}

The documentation for this class was generated from the following files:

fastq/FastQFile.h
fastq/FastQFile.cpp

Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

Member Function Documentation