comparison spp/src/BamMultiReader.h @ 6:ce08b0efa3fd draft

Uploaded
author zzhou
date Tue, 27 Nov 2012 16:11:40 -0500
parents
children
comparison
equal deleted inserted replaced
5:608a8e0eac56 6:ce08b0efa3fd
1 // ***************************************************************************
2 // BamMultiReader.h (c) 2010 Erik Garrison, Derek Barnett
3 // Marth Lab, Department of Biology, Boston College
4 // All rights reserved.
5 // ---------------------------------------------------------------------------
6 // Last modified: 19 November 2010 (DB)
7 // ---------------------------------------------------------------------------
8 // Functionality for simultaneously reading multiple BAM files
9 // ***************************************************************************
10
11 #ifndef BAMMULTIREADER_H
12 #define BAMMULTIREADER_H
13
14 #include <api_global.h>
15 #include <BamReader.h>
16 #include <map>
17 #include <sstream>
18 #include <string>
19 #include <utility>
20
21 namespace BamTools {
22
23 // index mapping reference/position pairings to bamreaders and their alignments
24 typedef std::multimap<std::pair<int, int>, std::pair<BamReader*, BamAlignment*> > AlignmentIndex;
25
26 class API_EXPORT BamMultiReader {
27
28 // constructor / destructor
29 public:
30 BamMultiReader(void);
31 ~BamMultiReader(void);
32
33 // public interface
34 public:
35
36 // positioning
37 int CurrentRefID;
38 int CurrentLeft;
39
40 // region under analysis, specified using SetRegion
41 BamRegion Region;
42
43 // ----------------------
44 // BAM file operations
45 // ----------------------
46
47 // close BAM files
48 void Close(void);
49
50 // opens BAM files (and optional BAM index files, if provided)
51 // @openIndexes - triggers index opening, useful for suppressing
52 // error messages during merging of files in which we may not have
53 // indexes.
54 // @coreMode - setup our first alignments using GetNextAlignmentCore();
55 // also useful for merging
56 // @preferStandardIndex - look for standard BAM index ".bai" first. If false,
57 // will look for BamTools index ".bti".
58 bool Open(const std::vector<std::string>& filenames, bool openIndexes = true, bool coreMode = false, bool preferStandardIndex = false);
59
60 // returns whether underlying BAM readers ALL have an index loaded
61 // this is useful to indicate whether Jump() or SetRegion() are possible
62 bool IsIndexLoaded(void) const;
63
64 // performs random-access jump to reference, position
65 bool Jump(int refID, int position = 0);
66
67 // sets the target region
68 bool SetRegion(const BamRegion& region);
69 bool SetRegion(const int&, const int&, const int&, const int&); // convenience function to above
70
71 // returns file pointers to beginning of alignments
72 bool Rewind(void);
73
74 // ----------------------
75 // access alignment data
76 // ----------------------
77 // updates the reference id marker to match the lower limit of our readers
78 void UpdateReferenceID(void);
79
80 // retrieves next available alignment (returns success/fail) from all files
81 bool GetNextAlignment(BamAlignment&);
82 // retrieves next available alignment (returns success/fail) from all files
83 // and populates the support data with information about the alignment
84 // *** BUT DOES NOT PARSE CHARACTER DATA FROM THE ALIGNMENT
85 bool GetNextAlignmentCore(BamAlignment&);
86 // ... should this be private?
87 bool HasOpenReaders(void);
88
89 // ----------------------
90 // access auxiliary data
91 // ----------------------
92
93 // returns unified SAM header text for all files
94 const std::string GetHeaderText(void) const;
95 // returns number of reference sequences
96 const int GetReferenceCount(void) const;
97 // returns vector of reference objects
98 const BamTools::RefVector GetReferenceData(void) const;
99 // returns reference id (used for BamMultiReader::Jump()) for the given reference name
100 const int GetReferenceID(const std::string& refName) const;
101 // validates that we have a congruent set of BAM files that are aligned against the same reference sequences
102 void ValidateReaders() const;
103
104 // ----------------------
105 // BAM index operations
106 // ----------------------
107
108 // creates index for BAM files which lack them, saves to files (default = bamFilename + ".bai")
109 bool CreateIndexes(bool useStandardIndex = true);
110
111 // sets the index caching mode for the readers
112 void SetIndexCacheMode(const BamIndex::BamIndexCacheMode mode);
113
114 //const int GetReferenceID(const string& refName) const;
115
116 // utility
117 void PrintFilenames(void);
118 void DumpAlignmentIndex(void);
119 void UpdateAlignments(void); // updates our alignment cache
120
121 // private implementation
122 private:
123
124 // the set of readers and alignments which we operate on, maintained throughout the life of this class
125 std::vector<std::pair<BamReader*, BamAlignment*> > readers;
126
127 // readers and alignments sorted by reference id and position, to keep track of the lowest (next) alignment
128 // when a reader reaches EOF, its entry is removed from this index
129 AlignmentIndex alignments;
130
131 std::vector<std::string> fileNames;
132 };
133
134 } // namespace BamTools
135
136 #endif // BAMMULTIREADER_H