annotate BEDTools-Version-2.14.3/src/fastaFromBed/fastaFromBedMain.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 fastaFromBedMain.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 "fastaFromBed.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 "fastaFromBed"
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 fastaDbFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34 string bedFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36 // output files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 string fastaOutFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 // checks for existence of parameters
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool haveFastaDb = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 bool haveBed = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool haveFastaOut = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 bool useNameOnly = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 bool useFasta = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 bool useStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 // check to see if we should print out some help
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48 if(argc <= 1) showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 if(PARAMETER_CHECK("-fi", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 haveFastaDb = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 fastaDbFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 else if(PARAMETER_CHECK("-fo", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 haveFastaOut = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 fastaOutFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 else if(PARAMETER_CHECK("-bed", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 bedFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 else if(PARAMETER_CHECK("-name", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 useNameOnly = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 else if(PARAMETER_CHECK("-tab", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 useFasta = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 else if(PARAMETER_CHECK("-s", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 useStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 cerr << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 if (!haveFastaDb || !haveFastaOut || !haveBed) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 Bed2Fa *b2f = new Bed2Fa(useNameOnly, fastaDbFile, bedFile, fastaOutFile, useFasta, useStrand);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109 delete b2f;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124 cerr << "Summary: Extract DNA sequences into a fasta file based on feature coordinates." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -fi <fasta> -bed <bed/gff/vcf> -fo <fasta> " << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 cerr << "\t-fi\tInput FASTA file" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 cerr << "\t-bed\tBED/GFF/VCF file of ranges to extract from -fi" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 cerr << "\t-fo\tOutput file (can be FASTA or TAB-delimited)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 cerr << "\t-name\tUse the name field for the FASTA header" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 cerr << "\t-tab\tWrite output in TAB delimited format." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 cerr << "\t\t- Default is FASTA format." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 cerr << "\t-s\tForce strandedness. If the feature occupies the antisense strand," << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 cerr << "\t\tthe sequence will be reverse complemented." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 cerr << "\t\t- By default, strand information is ignored." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146 }