Mercurial > repos > aaronquinlan > multi_intersect

comparison BEDTools-Version-2.14.3/src/unionBedGraphs/unionBedGraphs.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 unionBedGraphs.h
3
4 (c) 2010 - Assaf Gordon, CSHL
5 - Aaron Quinlan, UVA
6 Hall Laboratory
7 Department of Biochemistry and Molecular Genetics
8 University of Virginia
9 aaronquinlan@gmail.com
10
11 Licenced under the GNU General Public License 2.0 license.
12 ******************************************************************************/
13 #ifndef UNIONBEDGRAPHS_H
14 #define UNIONBEDGRAPHS_H
15
16 #include <vector>
17 #include <string>
18 #include "bedGraphFile.h"
19 #include "genomeFile.h"
20 #include "intervalItem.h"
21
22 class UnionBedGraphs
23 {
24 private:
25 typedef BEDGRAPH_STR BEDGRAPH_TYPE;
26
27 vector<string> filenames;
28 vector<string> titles;
29
30 vector<BedGraphFile*> bedgraph_files;
31 vector<BEDGRAPH_TYPE::DEPTH_TYPE> current_depth;
32 vector<BEDGRAPH_TYPE> current_bedgraph_item;
33
34 std::ostream &output;
35
36 INTERVALS_PRIORITY_QUEUE queue;
37 std::string current_chrom;
38 int current_non_zero_inputs;
39 bool print_empty_regions;
40
41 GenomeFile* genome_sizes;
42
43 std::string no_coverage_value;
44
45 public:
46 UnionBedGraphs(std::ostream& _output,
47 const vector<string>& _filenames,
48 const vector<string>& _titles,
49 bool _print_empty_regions,
50 const std::string& _genomeFileName,
51 const std::string& _no_coverage_value);
52
53 virtual ~UnionBedGraphs();
54
55 // Combines all bedgraph files
56 void Union();
57
58 // Print the header line: chrom/start/end + name of each bedgraph file.
59 void PrintHeader();
60
61
62 private:
63
64 // Open all BedGraph files, initialize "current_XXX" vectors
65 void OpenBedgraphFiles();
66
67 // Close the BedGraph files.
68 void CloseBedgraphFiles();
69
70 /*
71 Add an interval from BedGraph file 'index' into the queue.
72 will only be added if it belongs to the current chromosome.
73
74 If the interval was added (=consumed), the next interval will be read from the file
75 using 'LoadNextBedgraphItem'
76 */
77 void AddInterval(int index);
78
79 /*
80 Loads the next interval from BedGraph file 'index'.
81 Stores it in 'current_bedgraph_item' vector.
82 */
83 void LoadNextBedgraphItem(int index);
84
85 /*
86 Scans the 'current_bedgraph_item' vector,
87 find the 'first' chromosome to use (different BedGraph files can start with different chromosomes).
88 */
89 std::string DetermineNextChrom();
90
91 /*
92 Returns 'true' if ALL intervals from ALL BedGraph files were used
93 */
94 bool AllFilesDone();
95
96 /*
97 Extract the next coordinate from the queue, and updates the current coverage information.
98 If multiple interval share the same coordinate values, all of them are handled.
99 If an END coordinate is consumed, the next interval (from the corresponding file) is read.
100 */
101 CHRPOS ConsumeNextCoordinate();
102
103 /*
104 Updates the coverage information based on the given item.
105 Item can be a START coordinate or an END coordiante.
106 */
107 void UpdateInformation(const IntervalItem &item);
108
109 /*
110 prints chrom/start/end and the current depth coverage values of all the files.
111 */
112 void PrintCoverage(CHRPOS start, CHRPOS end);
113
114 /*
115 prints chrom/start/end and the ZERO depth coverage values of all the files.
116 */
117 void PrintEmptyCoverage(CHRPOS start, CHRPOS end);
118
119 void DebugPrintQueue();
120 };
121
122
123 #endif