|
0
|
1 /*****************************************************************************
|
|
|
2 slopBedMain.cpp
|
|
|
3
|
|
|
4 (c) 2009 - Aaron Quinlan
|
|
|
5 Hall Laboratory
|
|
|
6 Department of Biochemistry and Molecular Genetics
|
|
|
7 University of Virginia
|
|
|
8 aaronquinlan@gmail.com
|
|
|
9
|
|
|
10 Licenced under the GNU General Public License 2.0 license.
|
|
|
11 ******************************************************************************/
|
|
|
12 #include "slopBed.h"
|
|
|
13 #include "version.h"
|
|
|
14
|
|
|
15 using namespace std;
|
|
|
16
|
|
|
17 // define our program name
|
|
|
18 #define PROGRAM_NAME "slopBed"
|
|
|
19
|
|
|
20
|
|
|
21 // define our parameter checking macro
|
|
|
22 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
|
|
|
23
|
|
|
24 // function declarations
|
|
|
25 void ShowHelp(void);
|
|
|
26
|
|
|
27 int main(int argc, char* argv[]) {
|
|
|
28
|
|
|
29 // our configuration variables
|
|
|
30 bool showHelp = false;
|
|
|
31
|
|
|
32 // input files
|
|
|
33 string bedFile = "stdin";
|
|
|
34 string genomeFile;
|
|
|
35
|
|
|
36 bool haveBed = true;
|
|
|
37 bool haveGenome = false;
|
|
|
38 bool haveLeft = false;
|
|
|
39 bool haveRight = false;
|
|
|
40 bool haveBoth = false;
|
|
|
41
|
|
|
42 bool forceStrand = false;
|
|
|
43 float leftSlop = 0.0;
|
|
|
44 float rightSlop = 0.0;
|
|
|
45 bool fractional = false;
|
|
|
46
|
|
|
47 for(int i = 1; i < argc; i++) {
|
|
|
48 int parameterLength = (int)strlen(argv[i]);
|
|
|
49
|
|
|
50 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
|
|
|
51 (PARAMETER_CHECK("--help", 5, parameterLength))) {
|
|
|
52 showHelp = true;
|
|
|
53 }
|
|
|
54 }
|
|
|
55
|
|
|
56 if(showHelp) ShowHelp();
|
|
|
57
|
|
|
58 // do some parsing (all of these parameters require 2 strings)
|
|
|
59 for(int i = 1; i < argc; i++) {
|
|
|
60
|
|
|
61 int parameterLength = (int)strlen(argv[i]);
|
|
|
62
|
|
|
63 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
|
|
|
64 if ((i+1) < argc) {
|
|
|
65 bedFile = argv[i + 1];
|
|
|
66 i++;
|
|
|
67 }
|
|
|
68 }
|
|
|
69 else if(PARAMETER_CHECK("-g", 2, parameterLength)) {
|
|
|
70 if ((i+1) < argc) {
|
|
|
71 haveGenome = true;
|
|
|
72 genomeFile = argv[i + 1];
|
|
|
73 i++;
|
|
|
74 }
|
|
|
75 }
|
|
|
76 else if(PARAMETER_CHECK("-l", 2, parameterLength)) {
|
|
|
77 if ((i+1) < argc) {
|
|
|
78 haveLeft = true;
|
|
|
79 leftSlop = atof(argv[i + 1]);
|
|
|
80 i++;
|
|
|
81 }
|
|
|
82 }
|
|
|
83 else if(PARAMETER_CHECK("-r", 2, parameterLength)) {
|
|
|
84 if ((i+1) < argc) {
|
|
|
85 haveRight = true;
|
|
|
86 rightSlop = atof(argv[i + 1]);
|
|
|
87 i++;
|
|
|
88 }
|
|
|
89 }
|
|
|
90 else if(PARAMETER_CHECK("-b", 2, parameterLength)) {
|
|
|
91 if ((i+1) < argc) {
|
|
|
92 haveBoth = true;
|
|
|
93 leftSlop = atof(argv[i + 1]);
|
|
|
94 rightSlop = atof(argv[i + 1]);
|
|
|
95 i++;
|
|
|
96 }
|
|
|
97 }
|
|
|
98 else if(PARAMETER_CHECK("-s", 2, parameterLength)) {
|
|
|
99 forceStrand = true;
|
|
|
100 }
|
|
|
101 else if(PARAMETER_CHECK("-pct", 4, parameterLength)) {
|
|
|
102 fractional = true;
|
|
|
103 }
|
|
|
104 else {
|
|
|
105 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
|
|
|
106 showHelp = true;
|
|
|
107 }
|
|
|
108 }
|
|
|
109
|
|
|
110 // make sure we have both input files
|
|
|
111 if (!haveBed || !haveGenome) {
|
|
|
112 cerr << endl << "*****" << endl << "*****ERROR: Need both a BED (-i) and a genome (-g) file. " << endl << "*****" << endl;
|
|
|
113 showHelp = true;
|
|
|
114 }
|
|
|
115 if (!haveLeft && !haveRight && !haveBoth) {
|
|
|
116 cerr << endl << "*****" << endl << "*****ERROR: Need -l and -r together or -b alone. " << endl << "*****" << endl;
|
|
|
117 showHelp = true;
|
|
|
118 }
|
|
|
119 if ((!haveLeft && haveRight) || (haveLeft && !haveRight)) {
|
|
|
120 cerr << endl << "*****" << endl << "*****ERROR: Need both -l and -r. " << endl << "*****" << endl;
|
|
|
121 showHelp = true;
|
|
|
122 }
|
|
|
123 if (forceStrand && (!(haveLeft) || !(haveRight))) {
|
|
|
124 cerr << endl << "*****" << endl << "*****ERROR: Must supply -l and -r with -s. " << endl << "*****" << endl;
|
|
|
125 showHelp = true;
|
|
|
126 }
|
|
|
127
|
|
|
128 if (!showHelp) {
|
|
|
129 BedSlop *bc = new BedSlop(bedFile, genomeFile, forceStrand, leftSlop, rightSlop, fractional);
|
|
|
130 delete bc;
|
|
|
131
|
|
|
132 return 0;
|
|
|
133 }
|
|
|
134 else {
|
|
|
135 ShowHelp();
|
|
|
136 }
|
|
|
137 }
|
|
|
138
|
|
|
139 void ShowHelp(void) {
|
|
|
140
|
|
|
141 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
|
|
|
142
|
|
|
143 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
|
|
|
144
|
|
|
145 cerr << "Summary: Add requested base pairs of \"slop\" to each feature." << endl << endl;
|
|
|
146
|
|
|
147 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome> [-b <int> or (-l and -r)]" << endl << endl;
|
|
|
148
|
|
|
149 cerr << "Options: " << endl;
|
|
|
150 cerr << "\t-b\t" << "Increase the BED/GFF/VCF entry by -b base pairs in each direction." << endl;
|
|
|
151 cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
|
|
|
152
|
|
|
153 cerr << "\t-l\t" << "The number of base pairs to subtract from the start coordinate." << endl;
|
|
|
154 cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
|
|
|
155
|
|
|
156 cerr << "\t-r\t" << "The number of base pairs to add to the end coordinate." << endl;
|
|
|
157 cerr << "\t\t- (Integer) or (Float, e.g. 0.1) if used with -pct." << endl << endl;
|
|
|
158
|
|
|
159 cerr << "\t-s\t" << "Define -l and -r based on strand." << endl;
|
|
|
160 cerr << "\t\tE.g. if used, -l 500 for a negative-stranded feature, " << endl;
|
|
|
161 cerr << "\t\tit will add 500 bp downstream. Default = false." << endl << endl;
|
|
|
162
|
|
|
163 cerr << "\t-pct\t" << "Define -l and -r as a fraction of the feature's length." << endl;
|
|
|
164 cerr << "\t\tE.g. if used on a 1000bp feature, -l 0.50, " << endl;
|
|
|
165 cerr << "\t\twill add 500 bp \"upstream\". Default = false." << endl << endl;
|
|
|
166
|
|
|
167 cerr << "Notes: " << endl;
|
|
|
168 cerr << "\t(1) Starts will be set to 0 if options would force it below 0." << endl;
|
|
|
169 cerr << "\t(2) Ends will be set to the chromosome length if requested slop would" << endl;
|
|
|
170 cerr << "\tforce it above the max chrom length." << endl;
|
|
|
171
|
|
|
172 cerr << "\t(3) The genome file should tab delimited and structured as follows:" << endl;
|
|
|
173 cerr << "\n\t<chromName><TAB><chromSize>" << endl << endl;
|
|
|
174 cerr << "\tFor example, Human (hg19):" << endl;
|
|
|
175 cerr << "\tchr1\t249250621" << endl;
|
|
|
176 cerr << "\tchr2\t243199373" << endl;
|
|
|
177 cerr << "\t..." << endl;
|
|
|
178 cerr << "\tchr18_gl000207_random\t4262" << endl << endl;
|
|
|
179
|
|
|
180
|
|
|
181 cerr << "Tips: " << endl;
|
|
|
182 cerr << "\tOne can use the UCSC Genome Browser's MySQL database to extract" << endl;
|
|
|
183 cerr << "\tchromosome sizes. For example, H. sapiens:" << endl << endl;
|
|
|
184 cerr << "\tmysql --user=genome --host=genome-mysql.cse.ucsc.edu -A -e \\" << endl;
|
|
|
185 cerr << "\t\"select chrom, size from hg19.chromInfo\" > hg19.genome" << endl << endl;
|
|
|
186
|
|
|
187
|
|
|
188 // end the program here
|
|
|
189 exit(1);
|
|
|
190 }
|
