annotate BEDTools-Version-2.14.3/src/fjoin/fjoin.h @ 0:dfcd8b6c1bda

Uploaded
author aaronquinlan
date Thu, 03 Nov 2011 10:25:04 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
1 /*****************************************************************************
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
2 intersectBed.h
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
3
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
4 (c) 2009 - Aaron Quinlan
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
5 Hall Laboratory
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
6 Department of Biochemistry and Molecular Genetics
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
7 University of Virginia
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
8 aaronquinlan@gmail.com
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
9
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
10 Licenced under the GNU General Public License 2.0 license.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
11 ******************************************************************************/
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
12 #ifndef INTERSECTBED_H
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
13 #define INTERSECTBED_H
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
14
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
15 #include "bedFile.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
16 // #include "BamReader.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
17 // #include "BamWriter.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
18 // #include "BamAncillary.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
19 // #include "BamAux.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
20 // using namespace BamTools;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
21
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
22
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
23 #include <vector>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
24 #include <queue>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
25 #include <iostream>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
26 #include <fstream>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
27 #include <stdlib.h>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
28 using namespace std;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
29
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
30
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
31
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
32 class BedIntersect {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
33
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34 public:
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36 // constructor
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 BedIntersect(string bedAFile, string bedBFile, bool anyHit,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 bool writeA, bool writeB, bool writeOverlap, bool writeAllOverlap,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 float overlapFraction, bool noHit, bool writeCount, bool forceStrand,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool reciprocal, bool obeySplits, bool bamInput, bool bamOutput);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 // destructor
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 ~BedIntersect(void);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 private:
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 //------------------------------------------------
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 // private attributes
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49 //------------------------------------------------
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 string _bedAFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51 string _bedBFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 bool _writeA; // should the original A feature be reported?
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 bool _writeB; // should the original B feature be reported?
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55 bool _writeOverlap;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 bool _writeAllOverlap;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58 bool _forceStrand;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 bool _reciprocal;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60 float _overlapFraction;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62 bool _anyHit;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63 bool _noHit;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 bool _writeCount; // do we want a count of the number of overlaps in B?
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 bool _obeySplits;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 bool _bamInput;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 bool _bamOutput;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 bool _printable;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 queue<BED*> _outputBuffer;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 bool _lastPick;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 map<string, vector<BED*> > _windowA;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 map<string, vector<BED*> > _windowB;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 // instance of a bed file class.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 BedFile *_bedA, *_bedB;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 //------------------------------------------------
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 // private methods
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 //------------------------------------------------
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 void IntersectBed(istream &bedInput);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 void Scan(BED *x, vector<BED *> *windowX, BedLineStatus xStatus,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 const BED &y, vector<BED *> *windowY, BedLineStatus yStatus);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 void AddHits(BED *x, const BED &y);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 void FlushOutputBuffer(bool final = false);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 vector<BED*>* GetWindow(const string &chrom, bool isA);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 void ChromSwitch(const string &chrom);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 void IntersectBed();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 void IntersectBam(string bamFile);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 bool processHits(BED &a, vector<BED> &hits);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 bool FindOverlaps(const BED &a, vector<BED> &hits);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 bool FindOneOrMoreOverlap(const BED &a);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 void ReportOverlapDetail(const int &overlapBases, const BED &a, const BED &b,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 const CHRPOS &s, const CHRPOS &e);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 void ReportOverlapSummary(const BED &a, const int &numOverlapsFound);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110 void ReportHits(set<BED> &A, set<BED> &B);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 };
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 #endif /* INTERSECTBED_H */