annotate BEDTools-Version-2.14.3/src/pairToPair/pairToPairMain.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 pairToPairMain.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 "pairToPair.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
13 #include "version.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
14
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
15 using namespace std;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
16
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
17 // define our program name
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
18 #define PROGRAM_NAME "pairToPair"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
19
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
20 // define our parameter checking macro
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
21 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
22
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
23 // function declarations
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
24 void ShowHelp(void);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
25
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
26 int main(int argc, char* argv[]) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
27
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
28 // our configuration variables
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
29 bool showHelp = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
30
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
31 // input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
32 string bedAFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
33 string bedBFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35 // input arguments
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36 float overlapFraction = 1E-9;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 int slop = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 string searchType = "both";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 // flags to track parameters
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 bool haveBedA = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool haveBedB = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 bool haveSearchType = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 bool haveFraction = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 bool ignoreStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46 bool requireDifferentNames = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 bool haveSlop = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 bool strandedSlop = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49 // check to see if we should print out some help
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 if(argc <= 1) showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 if(PARAMETER_CHECK("-a", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 haveBedA = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 bedAFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 else if(PARAMETER_CHECK("-b", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 haveBedB = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 bedBFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 else if(PARAMETER_CHECK("-type", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 haveSearchType = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 searchType = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 else if(PARAMETER_CHECK("-f", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 haveFraction = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 overlapFraction = atof(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 else if(PARAMETER_CHECK("-slop", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 haveSlop = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 slop = atoi(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 else if(PARAMETER_CHECK("-ss", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 strandedSlop = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 else if(PARAMETER_CHECK("-rdn", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 requireDifferentNames = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109 else if(PARAMETER_CHECK("-is", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110 ignoreStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 // make sure we have both input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 if (!haveBedA || !haveBedB) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121 cerr << endl << "*****" << endl << "*****ERROR: Need -a and -b files. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125 if (haveSearchType && (searchType != "neither") && (searchType != "both") && (searchType != "either") && (searchType != "notboth")) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 cerr << endl << "*****" << endl << "*****ERROR: Request \"both\",\"neither\",\"either\",or \"notboth\"" << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 if (strandedSlop == true && haveSlop == false) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 cerr << endl << "*****" << endl << "*****ERROR: Need a -slop value if requesting -ss." << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 PairToPair *bi = new PairToPair(bedAFile, bedBFile, overlapFraction, searchType,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 ignoreStrand, requireDifferentNames, slop, strandedSlop);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 delete bi;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153 cerr << "Summary: Report overlaps between two paired-end BED files (BEDPE)." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <BEDPE> -b <BEDPE>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158 cerr << "\t-f\t" << "Minimum overlap required as fraction of A (e.g. 0.05)." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159 cerr << "\t\tDefault is 1E-9 (effectively 1bp)." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161 cerr << "\t-type \t" << "Approach to reporting overlaps between A and B." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162 cerr << "\t\tneither\tReport overlaps if neither end of A overlaps B." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 cerr << "\t\teither\tReport overlaps if either ends of A overlap B." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
164 cerr << "\t\tboth\tReport overlaps if both ends of A overlap B." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
165 cerr << "\t\tnotboth\tReport overlaps if one or neither of ends of A overlap B." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
166
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
167 cerr << "\t\t- Default = both." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
168
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
169 cerr << "\t-slop \t" << "The amount of slop (in b.p.). to be added to each footprint." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
170 cerr << "\t\t*Note*: Slop is subtracted from start1 and start2 and added to end1 and end2." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
171
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
172 cerr << "\t-ss\t" << "Add slop based to each BEDPE footprint based on strand." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
173 cerr << "\t\t- If strand is \"+\", slop is only added to the end coordinates." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
174 cerr << "\t\t- If strand is \"-\", slop is only added to the start coordinates." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
175 cerr << "\t\t- By default, slop is added in both directions." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
176
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
177 cerr << "\t-is\t" << "Ignore strands when searching for overlaps." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
178 cerr << "\t\t- By default, strands are enforced." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
179
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
180 cerr << "\t-rdn\t" << "Require the hits to have different names (i.e. avoid self-hits)." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
181 cerr << "\t\t- By default, same names are allowed." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
182
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
183
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
184 cerr << "Refer to the BEDTools manual for BEDPE format." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
185
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
186 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
187 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
188 }