|
0
|
1 /*****************************************************************************
|
|
|
2 windowMain.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 "windowBed.h"
|
|
|
13 #include "version.h"
|
|
|
14
|
|
|
15 using namespace std;
|
|
|
16
|
|
|
17 // define the version
|
|
|
18 #define PROGRAM_NAME "windowBed"
|
|
|
19
|
|
|
20 // define our parameter checking macro
|
|
|
21 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
|
|
|
22
|
|
|
23 // function declarations
|
|
|
24 void ShowHelp(void);
|
|
|
25
|
|
|
26
|
|
|
27 int main(int argc, char* argv[]) {
|
|
|
28
|
|
|
29 // our configuration variables
|
|
|
30 bool showHelp = false;
|
|
|
31
|
|
|
32 // input files
|
|
|
33 string bedAFile;
|
|
|
34 string bedBFile;
|
|
|
35
|
|
|
36 // input arguments
|
|
|
37 int leftSlop = 1000;
|
|
|
38 int rightSlop = 1000;
|
|
|
39
|
|
|
40 bool haveBedA = false;
|
|
|
41 bool haveBedB = false;
|
|
|
42 bool noHit = false;
|
|
|
43 bool anyHit = false;
|
|
|
44 bool writeCount = false;
|
|
|
45 bool haveSlop = false;
|
|
|
46 bool haveLeft = false;
|
|
|
47 bool haveRight = false;
|
|
|
48 bool strandWindows = false;
|
|
|
49 bool matchOnSameStrand = false;
|
|
|
50 bool matchOnDiffStrand = false;
|
|
|
51 bool inputIsBam = false;
|
|
|
52 bool outputIsBam = true;
|
|
|
53 bool uncompressedBam = false;
|
|
|
54
|
|
|
55 // check to see if we should print out some help
|
|
|
56 if(argc <= 1) showHelp = true;
|
|
|
57
|
|
|
58 for(int i = 1; i < argc; i++) {
|
|
|
59 int parameterLength = (int)strlen(argv[i]);
|
|
|
60
|
|
|
61 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
|
|
|
62 (PARAMETER_CHECK("--help", 5, parameterLength))) {
|
|
|
63 showHelp = true;
|
|
|
64 }
|
|
|
65 }
|
|
|
66
|
|
|
67 if(showHelp) ShowHelp();
|
|
|
68
|
|
|
69 // do some parsing (all of these parameters require 2 strings)
|
|
|
70 for(int i = 1; i < argc; i++) {
|
|
|
71
|
|
|
72 int parameterLength = (int)strlen(argv[i]);
|
|
|
73
|
|
|
74 if(PARAMETER_CHECK("-a", 2, parameterLength)) {
|
|
|
75 if ((i+1) < argc) {
|
|
|
76 haveBedA = true;
|
|
|
77 bedAFile = argv[i + 1];
|
|
|
78 i++;
|
|
|
79 }
|
|
|
80 }
|
|
|
81 else if(PARAMETER_CHECK("-abam", 5, parameterLength)) {
|
|
|
82 if ((i+1) < argc) {
|
|
|
83 haveBedA = true;
|
|
|
84 inputIsBam = true;
|
|
|
85 bedAFile = argv[i + 1];
|
|
|
86 i++;
|
|
|
87 }
|
|
|
88 }
|
|
|
89 else if(PARAMETER_CHECK("-b", 2, parameterLength)) {
|
|
|
90 if ((i+1) < argc) {
|
|
|
91 haveBedB = true;
|
|
|
92 bedBFile = argv[i + 1];
|
|
|
93 i++;
|
|
|
94 }
|
|
|
95 }
|
|
|
96 else if(PARAMETER_CHECK("-bed", 4, parameterLength)) {
|
|
|
97 outputIsBam = false;
|
|
|
98 }
|
|
|
99 else if(PARAMETER_CHECK("-u", 2, parameterLength)) {
|
|
|
100 anyHit = true;
|
|
|
101 }
|
|
|
102 else if(PARAMETER_CHECK("-c", 2, parameterLength)) {
|
|
|
103 writeCount = true;
|
|
|
104 }
|
|
|
105 else if (PARAMETER_CHECK("-v", 2, parameterLength)) {
|
|
|
106 noHit = true;
|
|
|
107 }
|
|
|
108 else if (PARAMETER_CHECK("-sw", 3, parameterLength)) {
|
|
|
109 strandWindows = true;
|
|
|
110 }
|
|
|
111 else if (PARAMETER_CHECK("-sm", 3, parameterLength)) {
|
|
|
112 matchOnSameStrand = true;
|
|
|
113 }
|
|
|
114 else if (PARAMETER_CHECK("-Sm", 3, parameterLength)) {
|
|
|
115 matchOnDiffStrand = true;
|
|
|
116 }
|
|
|
117 else if (PARAMETER_CHECK("-w", 2, parameterLength)) {
|
|
|
118 if ((i+1) < argc) {
|
|
|
119 haveSlop = true;
|
|
|
120 leftSlop = atoi(argv[i + 1]);
|
|
|
121 rightSlop = leftSlop;
|
|
|
122 i++;
|
|
|
123 }
|
|
|
124 }
|
|
|
125 else if (PARAMETER_CHECK("-l", 2, parameterLength)) {
|
|
|
126 if ((i+1) < argc) {
|
|
|
127 haveLeft = true;
|
|
|
128 leftSlop = atoi(argv[i + 1]);
|
|
|
129 i++;
|
|
|
130 }
|
|
|
131 }
|
|
|
132 else if (PARAMETER_CHECK("-r", 2, parameterLength)) {
|
|
|
133 if ((i+1) < argc) {
|
|
|
134 haveRight = true;
|
|
|
135 rightSlop = atoi(argv[i + 1]);
|
|
|
136 i++;
|
|
|
137 }
|
|
|
138 }
|
|
|
139 else if(PARAMETER_CHECK("-ubam", 5, parameterLength)) {
|
|
|
140 uncompressedBam = true;
|
|
|
141 }
|
|
|
142 else {
|
|
|
143 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
|
|
|
144 showHelp = true;
|
|
|
145 }
|
|
|
146 }
|
|
|
147
|
|
|
148 // make sure we have both input files
|
|
|
149 if (!haveBedA || !haveBedB) {
|
|
|
150 cerr << endl << "*****" << endl << "*****ERROR: Need -a and -b files. " << endl << "*****" << endl;
|
|
|
151 showHelp = true;
|
|
|
152 }
|
|
|
153
|
|
|
154 if (anyHit && noHit) {
|
|
|
155 cerr << endl << "*****" << endl << "*****ERROR: Request either -u OR -v, not both." << endl << "*****" << endl;
|
|
|
156 showHelp = true;
|
|
|
157 }
|
|
|
158
|
|
|
159 if (anyHit && writeCount) {
|
|
|
160 cerr << endl << "*****" << endl << "*****ERROR: Request either -u OR -c, not both." << endl << "*****" << endl;
|
|
|
161 showHelp = true;
|
|
|
162 }
|
|
|
163
|
|
|
164 if (haveLeft && (leftSlop < 0)) {
|
|
|
165 cerr << endl << "*****" << endl << "*****ERROR: Upstream window (-l) must be positive." << endl << "*****" << endl;
|
|
|
166 showHelp = true;
|
|
|
167 }
|
|
|
168
|
|
|
169 if (haveRight && (rightSlop < 0)) {
|
|
|
170 cerr << endl << "*****" << endl << "*****ERROR: Downstream window (-r) must be positive." << endl << "*****" << endl;
|
|
|
171 showHelp = true;
|
|
|
172 }
|
|
|
173
|
|
|
174 if (haveSlop && (haveLeft || haveRight)) {
|
|
|
175 cerr << endl << "*****" << endl << "*****ERROR: Cannot choose -w with -l or -r. Either specify -l and -r or specify solely -w" << endl << "*****" << endl;
|
|
|
176 showHelp = true;
|
|
|
177 }
|
|
|
178
|
|
|
179 if ((haveLeft && !haveRight) || (haveRight && !haveLeft)) {
|
|
|
180 cerr << endl << "*****" << endl << "*****ERROR: Please specify both -l and -r." << endl << "*****" << endl;
|
|
|
181 showHelp = true;
|
|
|
182 }
|
|
|
183
|
|
|
184 if (matchOnSameStrand && matchOnDiffStrand) {
|
|
|
185 cerr << endl << "*****" << endl << "*****ERROR: Use either -sm or -Sm, not both." << endl << "*****" << endl;
|
|
|
186 showHelp = true;
|
|
|
187 }
|
|
|
188
|
|
|
189 if (!showHelp) {
|
|
|
190 BedWindow *bi = new BedWindow(bedAFile, bedBFile, leftSlop, rightSlop, anyHit,
|
|
|
191 noHit, writeCount, strandWindows, matchOnSameStrand, matchOnDiffStrand,
|
|
|
192 inputIsBam, outputIsBam, uncompressedBam);
|
|
|
193 delete bi;
|
|
|
194 return 0;
|
|
|
195 }
|
|
|
196 else {
|
|
|
197 ShowHelp();
|
|
|
198 }
|
|
|
199 }
|
|
|
200
|
|
|
201
|
|
|
202 void ShowHelp(void) {
|
|
|
203
|
|
|
204 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
|
|
|
205
|
|
|
206 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
|
|
|
207
|
|
|
208 cerr << "Summary: Examines a \"window\" around each feature in A and" << endl;
|
|
|
209 cerr << "\t reports all features in B that overlap the window. For each" << endl;
|
|
|
210 cerr << "\t overlap the entire entry in A and B are reported." << endl << endl;
|
|
|
211
|
|
|
212 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -a <bed/gff/vcf> -b <bed/gff/vcf>" << endl << endl;
|
|
|
213
|
|
|
214 cerr << "Options: " << endl;
|
|
|
215
|
|
|
216 cerr << "\t-abam\t" << "The A input file is in BAM format. Output will be BAM as well." << endl << endl;
|
|
|
217
|
|
|
218 cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
|
|
|
219
|
|
|
220 cerr << "\t-bed\t" << "When using BAM input (-abam), write output as BED. The default" << endl;
|
|
|
221 cerr << "\t\tis to write output in BAM when using -abam." << endl << endl;
|
|
|
222
|
|
|
223 cerr << "\t-w\t" << "Base pairs added upstream and downstream of each entry" << endl;
|
|
|
224 cerr << "\t\tin A when searching for overlaps in B." << endl;
|
|
|
225 cerr << "\t\t- Creates symterical \"windows\" around A." << endl;
|
|
|
226 cerr << "\t\t- Default is 1000 bp." << endl;
|
|
|
227 cerr << "\t\t- (INTEGER)" << endl << endl;
|
|
|
228
|
|
|
229 cerr << "\t-l\t" << "Base pairs added upstream (left of) of each entry" << endl;
|
|
|
230 cerr << "\t\tin A when searching for overlaps in B." << endl;
|
|
|
231 cerr << "\t\t- Allows one to define assymterical \"windows\"." << endl;
|
|
|
232 cerr << "\t\t- Default is 1000 bp." << endl;
|
|
|
233 cerr << "\t\t- (INTEGER)" << endl << endl;
|
|
|
234
|
|
|
235 cerr << "\t-r\t" << "Base pairs added downstream (right of) of each entry" << endl;
|
|
|
236 cerr << "\t\tin A when searching for overlaps in B." << endl;
|
|
|
237 cerr << "\t\t- Allows one to define assymterical \"windows\"." << endl;
|
|
|
238 cerr << "\t\t- Default is 1000 bp." << endl;
|
|
|
239 cerr << "\t\t- (INTEGER)" << endl << endl;
|
|
|
240
|
|
|
241 cerr << "\t-sw\t" << "Define -l and -r based on strand. For example if used, -l 500" << endl;
|
|
|
242 cerr << "\t\tfor a negative-stranded feature will add 500 bp downstream." << endl;
|
|
|
243 cerr << "\t\t- Default = disabled." << endl << endl;
|
|
|
244
|
|
|
245 cerr << "\t-sm\t" << "Only report hits in B that overlap A on the _same_ strand." << endl;
|
|
|
246 cerr << "\t\t- By default, overlaps are reported without respect to strand." << endl << endl;
|
|
|
247
|
|
|
248 cerr << "\t-Sm\t" << "Only report hits in B that overlap A on the _opposite_ strand." << endl;
|
|
|
249 cerr << "\t\t- By default, overlaps are reported without respect to strand." << endl << endl;
|
|
|
250
|
|
|
251 cerr << "\t-u\t" << "Write the original A entry _once_ if _any_ overlaps found in B." << endl;
|
|
|
252 cerr << "\t\t- In other words, just report the fact >=1 hit was found." << endl << endl;
|
|
|
253
|
|
|
254 cerr << "\t-c\t" << "For each entry in A, report the number of overlaps with B." << endl;
|
|
|
255 cerr << "\t\t- Reports 0 for A entries that have no overlap with B." << endl;
|
|
|
256 cerr << "\t\t- Overlaps restricted by -f." << endl << endl;
|
|
|
257
|
|
|
258 cerr << "\t-v\t" << "Only report those entries in A that have _no overlaps_ with B." << endl;
|
|
|
259 cerr << "\t\t- Similar to \"grep -v.\"" << endl << endl;
|
|
|
260
|
|
|
261 // end the program here
|
|
|
262 exit(1);
|
|
|
263 }
|
