annotate BEDTools-Version-2.14.3/src/annotateBed/annotateMain.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 annotateMain.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 "annotateBed.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 the version
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
18 #define PROGRAM_NAME "annotateBed"
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 file
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
32 string mainFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
33
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34 // parm flags
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35 bool sameStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36 bool diffStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 bool haveBed = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 bool haveFiles = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 bool haveTitles = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool reportCounts = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 bool reportBoth = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 // list of annotation files / names
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 vector<string> inputFiles;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 vector<string> inputTitles;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 // check to see if we should print out some help
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 if(argc <= 1) showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 mainFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 else if(PARAMETER_CHECK("-files", 6, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 haveFiles = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 i = i+1;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 string file = argv[i];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 while (file[0] != '-' && i < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 inputFiles.push_back(file);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 if (i < argc)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 file = argv[i];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 i--;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 else if(PARAMETER_CHECK("-names", 6, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 haveTitles = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 i = i+1;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 string title = argv[i];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 while (title[0] != '-' && i < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 inputTitles.push_back(title);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95 if (i < argc)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 title = argv[i];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 i--;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 else if(PARAMETER_CHECK("-counts", 7, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 reportCounts = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 else if(PARAMETER_CHECK("-both", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 reportBoth = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 else if (PARAMETER_CHECK("-s", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 sameStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110 else if (PARAMETER_CHECK("-S", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111 diffStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 showHelp = true;
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 (!haveBed || !haveFiles) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121 cerr << endl << "*****" << endl << "*****ERROR: Need -i and -files 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 if (sameStrand && diffStrand) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125 cerr << endl << "*****" << endl << "*****ERROR: Request either -s OR -S, not both." << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 BedAnnotate *ba = new BedAnnotate(mainFile, inputFiles, inputTitles, sameStrand, diffStrand, reportCounts, reportBoth);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 ba->AnnotateBed();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 delete ba;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146 cerr << "Summary: Annotates the depth & breadth of coverage of features from multiple files" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147 cerr << "\t on the intervals in -i." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -files FILE1 FILE2 .. FILEn" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153 cerr << "\t-names\t" << "A list of names (one / file) to describe each file in -i." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154 cerr << "\t\tThese names will be printed as a header line." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156 cerr << "\t-counts\t" << "Report the count of features in each file that overlap -i." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 cerr << "\t\t- Default is to report the fraction of -i covered by each file." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159 cerr << "\t-both\t" << "Report the counts followed by the % coverage." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160 cerr << "\t\t- Default is to report the fraction of -i covered by each file." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162 cerr << "\t-s\t" << "Require same strandedness. That is, only counts overlaps" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 cerr << "\t\ton the _same_ strand." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
164 cerr << "\t\t- By default, overlaps are counted without respect to strand." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
165
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
166 cerr << "\t-S\t" << "Require different strandedness. That is, only count overlaps" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
167 cerr << "\t\ton the _opposite_ strand." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
168 cerr << "\t\t- By default, overlaps are counted without respect to strand." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
169 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
170 }