Mercurial > repos > aaronquinlan > multi_intersect

comparison BEDTools-Version-2.14.3/src/multiIntersectBed/multiIntersectBed.h @ 0:dfcd8b6c1bda

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Uploaded
author aaronquinlan
date Thu, 03 Nov 2011 10:25:04 -0400
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:dfcd8b6c1bda
1 /*****************************************************************************
2 multiIntersectBed.h
3
4 (c) 2010 - Aaron Quinlan, UVA
5 - Assaf Gordon, CSHL
6 Quinlan Laboratory
7 Department of Public Health Sciences
8 Center for Public Health Genomics
9 University of Virginia
10 aaronquinlan@gmail.com
11
12 Licenced under the GNU General Public License 2.0 license.
13 ******************************************************************************/
14 #ifndef MULTIINTERSECTBED_H
15 #define MULTIINTERSECTBED_H
16
17 #include <vector>
18 #include <string>
19 #include "bedFile.h"
20 #include "genomeFile.h"
21 #include "intervalItem.h"
22
23 class MultiIntersectBed
24 {
25 private:
26
27 vector<string> filenames;
28 vector<string> titles;
29
30 vector<BedFile*> input_files;
31 vector<int> current_depth;
32 vector<BED> current_item;
33
34 std::ostream &output;
35
36 INTERVALS_PRIORITY_QUEUE queue;
37 std::string current_chrom;
38 map<int, bool> files_with_coverage;
39 int current_non_zero_inputs;
40 bool print_empty_regions;
41 bool haveTitles;
42
43 GenomeFile* genome_sizes;
44
45 std::string no_coverage_value;
46
47 public:
48 MultiIntersectBed(std::ostream& _output,
49 const vector<string>& _filenames,
50 const vector<string>& _titles,
51 bool _print_empty_regions,
52 const std::string& _genomeFileName,
53 const std::string& _no_coverage_value);
54
55 virtual ~MultiIntersectBed();
56
57 // Combines all interval files
58 void MultiIntersect();
59
60 // Print the header line: chrom/start/end + name of each bedgraph file.
61 void PrintHeader();
62
63
64 private:
65
66 // Open all input files, initialize "current_XXX" vectors
67 void OpenFiles();
68
69 // Close the input files.
70 void CloseFiles();
71
72 /*
73 Add an interval from BedGraph file 'index' into the queue.
74 will only be added if it belongs to the current chromosome.
75
76 If the interval was added (=consumed), the next interval will be read from the file
77 using 'LoadNextItem'
78 */
79 void AddInterval(int index);
80
81 /*
82 Loads the next interval from Bed file 'index'.
83 Stores it in 'current_bed_item' vector.
84 */
85 void LoadNextItem(int index);
86
87 /*
88 Scans the 'current_bedgraph_item' vector,
89 find the 'first' chromosome to use (different BedGraph files can start with different chromosomes).
90 */
91 std::string DetermineNextChrom();
92
93 /*
94 Returns 'true' if ALL intervals from ALL BedGraph files were used
95 */
96 bool AllFilesDone();
97
98 /*
99 Extract the next coordinate from the queue, and updates the current coverage information.
100 If multiple interval share the same coordinate values, all of them are handled.
101 If an END coordinate is consumed, the next interval (from the corresponding file) is read.
102 */
103 CHRPOS ConsumeNextCoordinate();
104
105 /*
106 Updates the coverage information based on the given item.
107 Item can be a START coordinate or an END coordiante.
108 */
109 void UpdateInformation(const IntervalItem &item);
110
111 /*
112 prints chrom/start/end and the current depth coverage values of all the files.
113 */
114 void PrintCoverage(CHRPOS start, CHRPOS end);
115
116 /*
117 prints chrom/start/end and the ZERO depth coverage values of all the files.
118 */
119 void PrintEmptyCoverage(CHRPOS start, CHRPOS end);
120
121 void DebugPrintQueue();
122 };
123
124
125 #endif