annotate BEDTools-Version-2.14.3/src/mergeBed/mergeMain.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 mergeMain.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 "mergeBed.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 "mergeBed"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
19
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
20
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
21 // define our parameter checking macro
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
22 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
23
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
24 // function declarations
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
25 void ShowHelp(void);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
26
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
27 int main(int argc, char* argv[]) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
28
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
29 // our configuration variables
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
30 bool showHelp = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
31
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
32 // input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
33 string bedFile = "stdin";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34 int maxDistance = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35 string scoreOp = "";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 // input arguments
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 bool haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 bool numEntries = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool haveMaxDistance = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 bool forceStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool reportNames = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 bool reportScores = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63 bedFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 else if(PARAMETER_CHECK("-n", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 numEntries = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 else if(PARAMETER_CHECK("-d", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 haveMaxDistance = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 maxDistance = atoi(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 else if (PARAMETER_CHECK("-s", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 forceStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 else if (PARAMETER_CHECK("-nms", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 reportNames = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 else if (PARAMETER_CHECK("-scores", 7, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 reportScores = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 scoreOp = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 showHelp = true;
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 // make sure we have both input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 if (!haveBed) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 cerr << endl << "*****" << endl << "*****ERROR: Need -i BED file. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 if (reportNames && numEntries) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 cerr << endl << "*****" << endl << "*****ERROR: Request either -n OR -nms, not both." << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 if ((reportScores == true) && (scoreOp != "sum") && (scoreOp != "max") && (scoreOp != "min") && (scoreOp != "mean") &&
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 (scoreOp != "mode") && (scoreOp != "median") && (scoreOp != "antimode") && (scoreOp != "collapse"))
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 cerr << endl << "*****" << endl << "*****ERROR: Invalid scoreOp selection \"" << scoreOp << endl << "\" *****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 BedMerge *bm = new BedMerge(bedFile, numEntries, maxDistance, forceStrand, reportNames, reportScores, scoreOp);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 delete bm;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 cerr << "Summary: Merges overlapping BED/GFF/VCF entries into a single interval." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 cerr << "\t-s\t" << "Force strandedness. That is, only merge features" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 cerr << "\t\tthat are the same strand." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 cerr << "\t\t- By default, merging is done without respect to strand." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 cerr << "\t-n\t" << "Report the number of BED entries that were merged." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 cerr << "\t\t- Note: \"1\" is reported if no merging occurred." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 cerr << "\t-d\t" << "Maximum distance between features allowed for features" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142 cerr << "\t\tto be merged." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143 cerr << "\t\t- Def. 0. That is, overlapping & book-ended features are merged." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 cerr << "\t\t- (INTEGER)" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146 cerr << "\t-nms\t" << "Report the names of the merged features separated by semicolons." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148 cerr << "\t-scores\t" << "Report the scores of the merged features. Specify one of " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149 cerr << "\t\tthe following options for reporting scores:" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150 cerr << "\t\t sum, min, max," << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 cerr << "\t\t mean, median, mode, antimode," << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152 cerr << "\t\t collapse (i.e., print a semicolon-separated list)," << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153 cerr << "\t\t- (INTEGER)" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155 cerr << "Notes: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156 cerr << "\t(1) All output, regardless of input type (e.g., GFF or VCF)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 cerr << "\t will in BED format with zero-based starts" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 }