annotate BEDTools-Version-2.14.3/src/shuffleBed/shuffleBedMain.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 shuffleBedMain.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 "shuffleBed.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 "shuffleBed"
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 string excludeFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35 string includeFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36 string genomeFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 bool haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 bool haveGenome = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool haveExclude = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 bool haveInclude = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool haveSeed = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 float overlapFraction = 0.0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 int seed = -1;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 bool sameChrom = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46
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("-g", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 haveGenome = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 genomeFile = 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("-excl", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 haveExclude = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 excludeFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 else if(PARAMETER_CHECK("-incl", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 haveInclude = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 includeFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 else if(PARAMETER_CHECK("-seed", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 haveSeed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 seed = atoi(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 else if(PARAMETER_CHECK("-chrom", 6, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 sameChrom = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 else if(PARAMETER_CHECK("-f", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 overlapFraction = atof(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << 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
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 // make sure we have both input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 if (!haveBed || !haveGenome) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 cerr << endl << "*****" << endl << "*****ERROR: Need both a BED (-i) and a genome (-g) file. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 if (haveInclude && haveExclude) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 cerr << endl << "*****" << endl << "*****ERROR: Cannot use -incl and -excl together." << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125 BedShuffle *bc = new BedShuffle(bedFile, genomeFile, excludeFile, includeFile,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 haveSeed, haveExclude, haveInclude, sameChrom,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127 overlapFraction, seed);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 delete bc;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142 cerr << "Summary: Randomly permute the locations of a feature file among a genome." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147 cerr << "\t-excl\t" << "A BED/GFF/VCF file of coordinates in which features in -i" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148 cerr << "\t\tshould not be placed (e.g. gaps.bed)." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150 cerr << "\t-incl\t" << "Instead of randomly placing features in a genome, the -incl" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 cerr << "\t\toptions defines a BED/GFF/VCF file of coordinates in which " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152 cerr << "\t\tfeatures in -i should be randomly placed (e.g. genes.bed). " << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154 cerr << "\t-chrom\t" << "Keep features in -i on the same chromosome."<< endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155 cerr << "\t\t- By default, the chrom and position are randomly chosen." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 cerr << "\t-seed\t" << "Supply an integer seed for the shuffling." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158 cerr << "\t\t- By default, the seed is chosen automatically." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159 cerr << "\t\t- (INTEGER)" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161 cerr << "\t-f\t" << "Maximum overlap (as a fraction of the -i feature) with an -excl" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162 cerr << "\t\tfeature that is tolerated before searching for a new, " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 cerr << "\t\trandomized locus. For example, -f 0.10 allows up to 10%" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
164 cerr << "\t\tof a randomized feature to overlap with a given feature" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
165 cerr << "\t\tin the -excl file. **Cannot be used with -incl file.**" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
166 cerr << "\t\t- Default is 1E-9 (i.e., 1bp)." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
167 cerr << "\t\t- FLOAT (e.g. 0.50)" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
168
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
169 cerr << "Notes: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
170 cerr << "\t(1) The genome file should tab delimited and structured as follows:" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
171 cerr << "\t <chromName><TAB><chromSize>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
172 cerr << "\tFor example, Human (hg19):" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
173 cerr << "\tchr1\t249250621" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
174 cerr << "\tchr2\t243199373" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
175 cerr << "\t..." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
176 cerr << "\tchr18_gl000207_random\t4262" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
177
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
178
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
179 cerr << "Tips: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
180 cerr << "\tOne can use the UCSC Genome Browser's MySQL database to extract" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
181 cerr << "\tchromosome sizes. For example, H. sapiens:" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
182 cerr << "\tmysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -e \\" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
183 cerr << "\t\"select chrom, size from hg19.chromInfo\" > hg19.genome" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
184
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 }