BamWriter.h

   1 // ***************************************************************************
   2 // BamWriter.h (c) 2009 Michael Strömberg, Derek Barnett
   3 // Marth Lab, Department of Biology, Boston College
   4 // All rights reserved.
   5 // ---------------------------------------------------------------------------
   6 // Last modified: 24 June 2009 (DB)
   7 // ---------------------------------------------------------------------------
   8 // The BGZF routines were adapted from the bgzf.c code developed at the Broad
   9 // Institute.
  10 // ---------------------------------------------------------------------------
  11 // Provides the basic functionality for producing BAM files
  12 // ***************************************************************************
  13
  14 /*!
  15         \file BamWriter.h
  16         \brief API for writing BAM files.
  17 */
  18
  19 #pragma once
  20
  21 // C++ includes
  22 #include <string>
  23 #include <vector>
  24
  25 // zlib includes
  26 #include <zlib.h>
  27
  28 // BamTools includes
  29 #include "BamAux.h"
  30
  31 namespace BamTools {
  32
  33         //! API for writing BAM files.
  34         class BamWriter {
  35
  36                 public:
  37
  38                         //! Constructor.
  39                         BamWriter(void);
  40
  41                         //! Destructor.
  42                         ~BamWriter(void);
  43
  44                 public:
  45
  46                         /*!
  47                                 \brief Closes the alignment archive
  48                                 \sa Open()
  49                         */
  50                         void Close(void);
  51
  52                         /*!
  53                                 \brief Opens the alignment archive
  54                                 \param filename output BAM file
  55                                 \param samHeader SAM-format header text
  56                                 \param referenceSequences Reference sequence data
  57                                 \sa Close()
  58                         */
  59                         void Open(const std::string& filename, const std::string& samHeader, const RefVector& referenceSequences);
  60
  61                         /*!
  62                                 \brief Saves an alignment to the archive
  63                                 \param al The BamAlignment to be saved
  64                         */
  65                         void SaveAlignment(const BamAlignment& al);
  66
  67                 // --------------------------------------------------------------------------------------
  68                 // internal methods
  69                 private:
  70                         // closes the BAM file
  71                         void BgzfClose(void);
  72                         // compresses the current block
  73                         int BgzfDeflateBlock(void);
  74                         // flushes the data in the BGZF block
  75                         void BgzfFlushBlock(void);
  76                         // opens the BAM file for writing
  77                         void BgzfOpen(const std::string& filename);
  78                         // packs an unsigned integer into the specified buffer
  79                         static inline void BgzfPackUnsignedInt(char* buffer, unsigned int value);
  80                         // packs an unsigned short into the specified buffer
  81                         static inline void BgzfPackUnsignedShort(char* buffer, unsigned short value);
  82                         // writes the supplied data into the BGZF buffer
  83                         unsigned int BgzfWrite(const char* data, const unsigned int dataLen);
  84                         // calculates the minimum bin that contains a region [begin, end)
  85                         static inline unsigned int CalculateMinimumBin(unsigned int begin, unsigned int end);
  86                         // creates a packed cigar string from the supplied alignment
  87                         static void CreatePackedCigar(const std::vector<CigarOp>& cigarOperations, std::string& packedCigar);
  88                         // encodes the supplied query sequence into 4-bit notation
  89                         static void EncodeQuerySequence(const std::string& query, std::string& encodedQuery);
  90                         // our BGZF output object
  91                         BgzfData mBGZF;
  92         };
  93
  94         //! \cond
  95         // --------------------------------------------------------------------------------------
  96         // static inline methods (internal - can exclude from main documentation)
  97
  98         // packs an unsigned integer into the specified buffer
  99         inline void BamWriter::BgzfPackUnsignedInt(char* buffer, unsigned int value) {
 100                 buffer[0] = (char)value;
 101                 buffer[1] = (char)(value >> 8);
 102                 buffer[2] = (char)(value >> 16);
 103                 buffer[3] = (char)(value >> 24);
 104         }
 105
 106         // packs an unsigned short into the specified buffer
 107         inline void BamWriter::BgzfPackUnsignedShort(char* buffer, unsigned short value) {
 108                 buffer[0] = (char)value;
 109                 buffer[1] = (char)(value >> 8);
 110         }
 111         // --------------------------------------------------------------------------------------
 112         //! \endcond
 113
 114 } // end BamTools namespace