1 // ***************************************************************************
\r
2 // BamMultiReader.h (c) 2010 Erik Garrison
\r
3 // Marth Lab, Department of Biology, Boston College
\r
4 // All rights reserved.
\r
5 // ---------------------------------------------------------------------------
\r
6 // Last modified: 22 February 2010 (EG)
\r
7 // ---------------------------------------------------------------------------
\r
8 // Functionality for simultaneously reading multiple BAM files
\r
9 // ***************************************************************************
\r
11 #ifndef BAMMULTIREADER_H
\r
12 #define BAMMULTIREADER_H
\r
17 #include <utility> // for pair
\r
20 using namespace std;
\r
22 // BamTools includes
\r
24 #include "BamReader.h"
\r
26 namespace BamTools {
\r
28 // index mapping reference/position pairings to bamreaders and their alignments
\r
29 typedef multimap<pair<int, int>, pair<BamReader*, BamAlignment*> > AlignmentIndex;
\r
32 class BamMultiReader {
\r
34 // constructor / destructor
\r
36 BamMultiReader(void);
\r
37 ~BamMultiReader(void);
\r
46 // ----------------------
\r
47 // BAM file operations
\r
48 // ----------------------
\r
52 // performs random-access jump to reference, position
\r
53 bool Jump(int refID, int position = 0);
\r
54 // opens BAM files (and optional BAM index files, if provided)
\r
55 //void Open(const vector<std::string&> filenames, const vector<std::string&> indexFilenames);
\r
56 void Open(const vector<string> filenames, bool openIndexes = true);
\r
57 // returns file pointers to beginning of alignments
\r
60 // ----------------------
\r
61 // access alignment data
\r
62 // ----------------------
\r
63 // updates the reference id marker to match the lower limit of our readers
\r
64 void UpdateReferenceID(void);
\r
66 // retrieves next available alignment (returns success/fail) from all files
\r
67 bool GetNextAlignment(BamAlignment&);
\r
68 // ... should this be private?
\r
69 bool HasOpenReaders(void);
\r
71 // ----------------------
\r
72 // access auxiliary data
\r
73 // ----------------------
\r
75 // returns unified SAM header text for all files
\r
76 const string GetHeaderText(void) const;
\r
77 // returns number of reference sequences
\r
78 const int GetReferenceCount(void) const;
\r
79 // returns vector of reference objects
\r
80 const BamTools::RefVector GetReferenceData(void) const;
\r
81 // returns reference id (used for BamMultiReader::Jump()) for the given reference name
\r
82 const int GetReferenceID(const std::string& refName) const;
\r
83 // validates that we have a congruent set of BAM files that are aligned against the same reference sequences
\r
84 void ValidateReaders() const;
\r
86 // ----------------------
\r
87 // BAM index operations
\r
88 // ----------------------
\r
90 // creates index for BAM files which lack them, saves to files (default = bamFilename + ".bai")
\r
91 bool CreateIndexes(void);
\r
93 //const int GetReferenceID(const string& refName) const;
\r
96 void PrintFilenames(void);
\r
97 void DumpAlignmentIndex(void);
\r
99 // private implementation
\r
102 // the set of readers and alignments which we operate on, maintained throughout the life of this class
\r
103 vector<pair<BamReader*, BamAlignment*> > readers;
\r
105 // readers and alignments sorted by reference id and position, to keep track of the lowest (next) alignment
\r
106 // when a reader reaches EOF, its entry is removed from this index
\r
107 AlignmentIndex alignments;
\r
109 vector<string> fileNames;
\r
112 } // namespace BamTools
\r
114 #endif // BAMMULTIREADER_H
\r