annotate BEDTools-Version-2.14.3/src/slopBed/slopBedMain.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 slopBedMain.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 "slopBed.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 "slopBed"
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 genomeFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36 bool haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 bool haveGenome = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 bool haveLeft = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 bool haveRight = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool haveBoth = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool forceStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 float leftSlop = 0.0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 float rightSlop = 0.0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 bool fractional = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 bedFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 else if(PARAMETER_CHECK("-g", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 haveGenome = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 genomeFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 else if(PARAMETER_CHECK("-l", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 haveLeft = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 leftSlop = atof(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 else if(PARAMETER_CHECK("-r", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 haveRight = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 rightSlop = atof(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 if(PARAMETER_CHECK("-b", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 haveBoth = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 leftSlop = atof(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 rightSlop = atof(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("-s", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 forceStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 else if(PARAMETER_CHECK("-pct", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 fractional = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110 // make sure we have both input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111 if (!haveBed || !haveGenome) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 cerr << endl << "*****" << endl << "*****ERROR: Need both a BED (-i) and a genome (-g) file. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 if (!haveLeft && !haveRight && !haveBoth) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 cerr << endl << "*****" << endl << "*****ERROR: Need -l and -r together or -b alone. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 if ((!haveLeft && haveRight) || (haveLeft && !haveRight)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 cerr << endl << "*****" << endl << "*****ERROR: Need both -l and -r. " << 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 if (forceStrand && (!(haveLeft) || !(haveRight))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124 cerr << endl << "*****" << endl << "*****ERROR: Must supply -l and -r with -s. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 BedSlop *bc = new BedSlop(bedFile, genomeFile, forceStrand, leftSlop, rightSlop, fractional);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 delete bc;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145 cerr << "Summary: Add requested base pairs of \"slop\" to each feature." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome> [-b <int> or (-l and -r)]" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150 cerr << "\t-b\t" << "Increase the BED/GFF/VCF entry by -b base pairs in each direction." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153 cerr << "\t-l\t" << "The number of base pairs to subtract from the start coordinate." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154 cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156 cerr << "\t-r\t" << "The number of base pairs to add to the end coordinate." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159 cerr << "\t-s\t" << "Define -l and -r based on strand." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160 cerr << "\t\tE.g. if used, -l 500 for a negative-stranded feature, " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161 cerr << "\t\tit will add 500 bp downstream. Default = false." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 cerr << "\t-pct\t" << "Define -l and -r as a fraction of the feature's length." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
164 cerr << "\t\tE.g. if used on a 1000bp feature, -l 0.50, " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
165 cerr << "\t\twill add 500 bp \"upstream\". Default = false." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
166
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
167 cerr << "Notes: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
168 cerr << "\t(1) Starts will be set to 0 if options would force it below 0." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
169 cerr << "\t(2) Ends will be set to the chromosome length if requested slop would" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
170 cerr << "\tforce it above the max chrom length." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
171
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
172 cerr << "\t(3) The genome file should tab delimited and structured as follows:" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
173 cerr << "\n\t<chromName><TAB><chromSize>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
174 cerr << "\tFor example, Human (hg19):" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
175 cerr << "\tchr1\t249250621" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
176 cerr << "\tchr2\t243199373" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
177 cerr << "\t..." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
178 cerr << "\tchr18_gl000207_random\t4262" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
179
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
180
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
181 cerr << "Tips: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
182 cerr << "\tOne can use the UCSC Genome Browser's MySQL database to extract" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
183 cerr << "\tchromosome sizes. For example, H. sapiens:" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
184 cerr << "\tmysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -e \\" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
185 cerr << "\t\"select chrom, size from hg19.chromInfo\" > hg19.genome" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
186
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
187
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
188 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
189 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
190 }