annotate BEDTools-Version-2.14.3/src/nucBed/nucBedMain.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 nucBedMain.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 "nucBed.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 "nucBed"
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 string pattern;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37 // checks for existence of parameters
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 bool haveFastaDb = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 bool haveBed = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 bool printSeq = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 bool hasPattern = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 bool forceStrand = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44 // check to see if we should print out some help
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45 if(argc <= 1) showHelp = true;
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("-fi", 3, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 haveFastaDb = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66 fastaDbFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 else if(PARAMETER_CHECK("-bed", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72 haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 bedFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77 else if(PARAMETER_CHECK("-seq", 4, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 printSeq = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 else if(PARAMETER_CHECK("-s", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 forceStrand = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
82 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
83 else if(PARAMETER_CHECK("-pattern", 8, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
84 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
85 hasPattern = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
86 pattern = 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 {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 cerr << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 if (!haveFastaDb || !haveBed) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 NucBed *nuc = new NucBed(fastaDbFile, bedFile, printSeq, hasPattern, pattern, forceStrand);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 delete nuc;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
109 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
110 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
111
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
112 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 cerr << "Summary: Profiles the nucleotide content of intervals in a fasta file." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -fi <fasta> -bed <bed/gff/vcf>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123 cerr << "\t-fi\tInput FASTA file" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
124 cerr << "\t-bed\tBED/GFF/VCF file of ranges to extract from -fi" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
125 cerr << "\t-s\tProfile the sequence according to strand." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
126 cerr << "\t-seq\tPrint the extracted sequence" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
127 cerr << "\t-pattern\tReport the number of times a user-defined sequence is observed (case-insensitive)." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130 cerr << "Output format: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 cerr << "\tThe following information will be reported after each original BED entry:" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 cerr << "\t 1) %AT content" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 cerr << "\t 2) %GC content" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 cerr << "\t 3) Number of As observed" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 cerr << "\t 4) Number of Cs observed" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136 cerr << "\t 5) Number of Gs observed" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137 cerr << "\t 6) Number of Ts observed" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138 cerr << "\t 7) Number of Ns observed" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 cerr << "\t 8) Number of other bases observed" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140 cerr << "\t 9) The length of the explored sequence/interval." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 cerr << "\t 10) The sequence extracted from the FASTA file. (optional, if -seq is used)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142 cerr << "\t 11) The number of times a user defined pattern was observed. (optional, if -pattern is used.)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147 }