annotate BEDTools-Version-2.14.3/src/bed12ToBed6/bed12ToBed6.cpp @ 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 bed12ToBed6.cpp
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 #include "lineFileUtilities.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
13 #include "bedFile.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
14 #include "version.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
15
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
16 #include <vector>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
17 #include <iostream>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
18 #include <fstream>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
19 #include <stdlib.h>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
20
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
21 using namespace std;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
22
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
23
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
24 // define our program name
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
25 #define PROGRAM_NAME "bed12ToBed6"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
26
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
27 // define our parameter checking macro
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
28 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
29
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
30
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
31 // function declarations
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
32 void ShowHelp(void);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
33 void DetermineBedInput(BedFile *bed);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34 void ProcessBed(istream &bedInput, BedFile *bed);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 bool addBlockNums = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 int main(int argc, char* argv[]) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 // our configuration variables
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool showHelp = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 // input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 string bedFile = "stdin";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46 bool haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 bedFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 else if(PARAMETER_CHECK("-n", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 addBlockNums = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 // make sure we have an input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 if (!haveBed ) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 cerr << endl << "*****" << endl << "*****ERROR: Need -i (BED) file. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 BedFile *bed = new BedFile(bedFile);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 DetermineBedInput(bed);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 cerr << "Summary: Splits BED12 features into discrete BED6 features." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed12>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 cerr << "\t-n\t" << "Force the score to be the (1-based) block number from the BED12." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 void DetermineBedInput(BedFile *bed) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 // dealing with a proper file
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 if (bed->bedFile != "stdin") {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121 ifstream bedStream(bed->bedFile.c_str(), ios::in);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 if ( !bedStream ) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123 cerr << "Error: The requested bed file (" << bed->bedFile << ") could not be opened. Exiting!" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124 exit (1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 ProcessBed(bedStream, bed);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 // reading from stdin
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 ProcessBed(cin, bed);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 void ProcessBed(istream &bedInput, BedFile *bed) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 // process each BED entry and convert to BAM
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 BED bedEntry, nullBed;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 int lineNum = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140 BedLineStatus bedStatus;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 // open the BED file for reading.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142 bed->Open();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143 while ((bedStatus = bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 if (bedStatus == BED_VALID) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146 bedVector bedBlocks; // vec to store the discrete BED "blocks" from a
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147 splitBedIntoBlocks(bedEntry, lineNum, bedBlocks);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149 for (int i = 0; i < (int) bedBlocks.size(); ++i) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150 if (addBlockNums == false) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 printf ("%s\t%d\t%d\t%s\t%s\t%s\n", bedBlocks[i].chrom.c_str(), bedBlocks[i].start, bedBlocks[i].end, bedBlocks[i].name.c_str(),
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152 bedBlocks[i].score.c_str(), bedBlocks[i].strand.c_str());
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155 if (bedBlocks[i].strand == "+")
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156 printf ("%s\t%d\t%d\t%s\t%d\t%s\n", bedBlocks[i].chrom.c_str(), bedBlocks[i].start, bedBlocks[i].end, bedBlocks[i].name.c_str(),
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 i+1, bedBlocks[i].strand.c_str());
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158 else
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159 printf ("%s\t%d\t%d\t%s\t%d\t%s\n", bedBlocks[i].chrom.c_str(), bedBlocks[i].start, bedBlocks[i].end, bedBlocks[i].name.c_str(),
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160 (int) ((bedBlocks.size()+1)-i), bedBlocks[i].strand.c_str());
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 bedEntry = nullBed;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
164 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
165 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
166 // close up
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
167 bed->Close();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
168 }