annotate BEDTools-Version-2.14.3/src/maskFastaFromBed/maskFastaFromBedMain.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 maskFastaFromBedMain.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 "maskFastaFromBed.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 "maskFastaFromBed"
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 fastaInFile;
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 // defaults for parameters
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool haveFastaIn = 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 softMask = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 char maskChar = 'N';
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46 // check to see if we should print out some help
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 if(argc <= 1) showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 if(PARAMETER_CHECK("-fi", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 haveFastaIn = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 fastaInFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 else if(PARAMETER_CHECK("-fo", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 haveFastaOut = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 fastaOutFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 else if(PARAMETER_CHECK("-bed", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 bedFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 else if(PARAMETER_CHECK("-soft", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 softMask = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 else if(PARAMETER_CHECK("-mc", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 string mask = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 if (mask.size() > 1) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 cerr << "*****ERROR: The mask character (-mc) should be a single character.*****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 maskChar = mask[0];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 cerr << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 if (!haveFastaIn || !haveFastaOut || !haveBed) {
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 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 MaskFastaFromBed *maskFasta = new MaskFastaFromBed(fastaInFile, bedFile, fastaOutFile, softMask, maskChar);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 delete maskFasta;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 cerr << "Summary: Mask a fasta file based on feature coordinates." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -fi <fasta> -out <fasta> -bed <bed/gff/vcf>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 cerr << "Options:" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136 cerr << "\t-fi\tInput FASTA file" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 cerr << "\t-bed\tBED/GFF/VCF file of ranges to mask in -fi" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 cerr << "\t-fo\tOutput FASTA file" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 cerr << "\t-soft\tEnforce \"soft\" masking. That is, instead of masking with Ns," << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140 cerr << "\t\tmask with lower-case bases." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 cerr << "\t-mc\tReplace masking character. That is, instead of masking with Ns, use another character." << endl;
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 }