annotate BEDTools-Version-2.14.3/src/bedToBam/bedToBam.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 bedToBam.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 "lineFileUtilities.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
13 #include "bedFile.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
14 #include "genomeFile.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
15 #include "version.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
16
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
17
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
18 #include "api/BamReader.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
19 #include "api/BamAux.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
20 #include "api/BamWriter.h"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
21 using namespace BamTools;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
22
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
23 #include <vector>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
24 #include <iostream>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
25 #include <fstream>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
26 #include <stdlib.h>
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
27
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
28 using namespace std;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
29
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
30
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
31 // define our program name
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
32 #define PROGRAM_NAME "bedToBam"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
33
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
34 // define our parameter checking macro
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
35 #define PARAMETER_CHECK(param, paramLen, actualLen) (strncmp(argv[i], param, min(actualLen, paramLen))== 0) && (actualLen == paramLen)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
36
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
37
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
38 // function declarations
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
39 void ShowHelp(void);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
40 void ProcessBed(istream &bedInput, BedFile *bed, GenomeFile *genome, bool isBED12, int mapQual, bool uncompressedBam);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
41 void ConvertBedToBam(const BED &bed, BamAlignment &bam, map<string, int> &chromToId, bool isBED12, int mapQual, int lineNum);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
42 void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header, map<string, int> &chromToInt);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
43 int reg2bin(int beg, int end);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
44
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
45
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
46
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
47 int main(int argc, char* argv[]) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
48
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
49 // our configuration variables
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
50 bool showHelp = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
51
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
52 // input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
53 string bedFile = "stdin";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
54 string genomeFile;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
55
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
56 unsigned int mapQual = 255;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
57
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
58 bool haveBed = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
59 bool haveGenome = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
60 bool haveMapQual = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
61 bool isBED12 = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
62 bool uncompressedBam = false;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
63
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
64 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
65 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
66
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
67 if((PARAMETER_CHECK("-h", 2, parameterLength)) ||
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
68 (PARAMETER_CHECK("--help", 5, parameterLength))) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
69 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
70 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
71 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
72
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
73 if(showHelp) ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
74
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
75 // do some parsing (all of these parameters require 2 strings)
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
76 for(int i = 1; i < argc; i++) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
77
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
78 int parameterLength = (int)strlen(argv[i]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
79
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
80 if(PARAMETER_CHECK("-i", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
81 if ((i+1) < argc) {
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("-g", 2, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
87 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
88 haveGenome = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
89 genomeFile = argv[i + 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
90 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
91 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
92 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
93 else if(PARAMETER_CHECK("-mapq", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
94 haveMapQual = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
95 if ((i+1) < argc) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
96 mapQual = atoi(argv[i + 1]);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
97 i++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
98 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
99 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
100 else if(PARAMETER_CHECK("-bed12", 6, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
101 isBED12 = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
102 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
103 else if(PARAMETER_CHECK("-ubam", 5, parameterLength)) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
104 uncompressedBam = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
105 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
106 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
107 cerr << endl << "*****ERROR: Unrecognized parameter: " << argv[i] << " *****" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
108 showHelp = true;
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 // make sure we have an input files
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
113 if (!haveBed ) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
114 cerr << endl << "*****" << endl << "*****ERROR: Need -i (BED) file. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
115 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
116 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
117 if (!haveGenome ) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
118 cerr << endl << "*****" << endl << "*****ERROR: Need -g (genome) file. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
119 showHelp = true;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
120 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
121 if (mapQual < 0 || mapQual > 255) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
122 cerr << endl << "*****" << endl << "*****ERROR: MAPQ must be in range [0,255]. " << endl << "*****" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
123 showHelp = true;
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 if (!showHelp) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
128 BedFile *bed = new BedFile(bedFile);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
129 GenomeFile *genome = new GenomeFile(genomeFile);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
130
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
131 ProcessBed(cin, bed, genome, isBED12, mapQual, uncompressedBam);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
132 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
133 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
134 ShowHelp();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
135 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
136 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
137
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
138
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
139 void ShowHelp(void) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
140
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
141 cerr << endl << "Program: " << PROGRAM_NAME << " (v" << VERSION << ")" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
142
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
143 cerr << "Author: Aaron Quinlan (aaronquinlan@gmail.com)" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
144
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
145 cerr << "Summary: Converts feature records to BAM format." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
146
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
147 cerr << "Usage: " << PROGRAM_NAME << " [OPTIONS] -i <bed/gff/vcf> -g <genome>" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
148
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
149 cerr << "Options: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
150
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
151 cerr << "\t-mapq\t" << "Set the mappinq quality for the BAM records." << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
152 cerr << "\t\t(INT) Default: 255" << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
153
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
154 cerr << "\t-bed12\t" << "The BED file is in BED12 format. The BAM CIGAR" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
155 cerr << "\t\tstring will reflect BED \"blocks\"." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
156
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
157 cerr << "\t-ubam\t" << "Write uncompressed BAM output. Default is to write compressed BAM." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
158
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
159 cerr << "Notes: " << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
160 cerr << "\t(1) BED files must be at least BED4 to be amenable to BAM (needs name field)." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
161
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
162
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
163 // end the program here
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
164 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
165 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
166
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
167
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
168 void ProcessBed(istream &bedInput, BedFile *bed, GenomeFile *genome, bool isBED12, int mapQual, bool uncompressedBam) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
169
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
170 BamWriter *writer = new BamWriter();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
171
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
172 // build a BAM header from the genomeFile
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
173 RefVector refs;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
174 string bamHeader;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
175 map<string, int, std::less<string> > chromToId;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
176 MakeBamHeader(genome->getGenomeFileName(), refs, bamHeader, chromToId);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
177
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
178 // set compression mode
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
179 BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
180 if ( uncompressedBam ) compressionMode = BamWriter::Uncompressed;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
181 writer->SetCompressionMode(compressionMode);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
182 // open a BAM and add the reference headers to the BAM file
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
183 writer->Open("stdout", bamHeader, refs);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
184
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
185
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
186 // process each BED entry and convert to BAM
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
187 BED bedEntry, nullBed;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
188 int lineNum = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
189 BedLineStatus bedStatus;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
190 // open the BED file for reading.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
191 bed->Open();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
192 while ((bedStatus = bed->GetNextBed(bedEntry, lineNum)) != BED_INVALID) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
193 if (bedStatus == BED_VALID) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
194 BamAlignment bamEntry;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
195 if (bed->bedType >= 4) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
196 ConvertBedToBam(bedEntry, bamEntry, chromToId, isBED12, mapQual, lineNum);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
197 writer->SaveAlignment(bamEntry);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
198 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
199 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
200 cerr << "Error: BED entry without name found at line: " << lineNum << ". Exiting!" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
201 exit (1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
202 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
203 bedEntry = nullBed;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
204 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
205 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
206 //close up
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
207 bed->Close();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
208 writer->Close();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
209 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
210
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
211
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
212 void ConvertBedToBam(const BED &bed, BamAlignment &bam, map<string, int, std::less<string> > &chromToId,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
213 bool isBED12, int mapQual, int lineNum) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
214
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
215 bam.Name = bed.name;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
216 bam.Position = bed.start;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
217 bam.Bin = reg2bin(bed.start, bed.end);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
218
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
219 // hard-code the sequence and qualities.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
220 int bedLength = bed.end - bed.start;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
221
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
222 // set dummy seq and qual strings. the input is BED,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
223 // so the sequence is inherently the same as it's
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
224 // reference genome.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
225 // Thanks to James M. Ward for pointing this out.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
226 bam.QueryBases = "";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
227 bam.Qualities = "";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
228
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
229 // chrom and map quality
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
230 bam.RefID = chromToId[bed.chrom];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
231 bam.MapQuality = mapQual;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
232
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
233 // set the BAM FLAG
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
234 bam.AlignmentFlag = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
235 if (bed.strand == "-")
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
236 bam.SetIsReverseStrand(true);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
237
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
238 bam.MatePosition = -1;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
239 bam.InsertSize = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
240 bam.MateRefID = -1;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
241
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
242 bam.CigarData.clear();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
243
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
244 if (isBED12 == false) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
245 CigarOp cOp;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
246 cOp.Type = 'M';
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
247 cOp.Length = bedLength;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
248 bam.CigarData.push_back(cOp);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
249 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
250 // we're being told that the input is BED12.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
251 else{
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
252
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
253 // does it smell like BED12? if so, process it.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
254 if (bed.otherFields.size() == 6) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
255
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
256 // extract the relevant BED fields to convert BED12 to BAM
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
257 // namely: blockCount, blockStarts, blockEnds
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
258 unsigned int blockCount = atoi(bed.otherFields[3].c_str());
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
259
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
260 vector<int> blockSizes, blockStarts;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
261 Tokenize(bed.otherFields[4], blockSizes, ",");
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
262 Tokenize(bed.otherFields[5], blockStarts, ",");
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
263
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
264 // make sure this is a well-formed BED12 entry.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
265 if (blockSizes.size() != blockCount) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
266 cerr << "Error: Number of BED blocks does not match blockCount at line: " << lineNum << ". Exiting!" << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
267 exit (1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
268 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
269 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
270 // does the first block start after the bed.start?
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
271 // if so, we need to do some "splicing"
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
272 if (blockStarts[0] > 0) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
273 CigarOp cOp;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
274 cOp.Length = blockStarts[0];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
275 cOp.Type = 'N';
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
276 bam.CigarData.push_back(cOp);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
277 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
278 // handle the "middle" blocks
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
279 for (unsigned int i = 0; i < blockCount - 1; ++i) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
280 CigarOp cOp;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
281 cOp.Length = blockSizes[i];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
282 cOp.Type = 'M';
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
283 bam.CigarData.push_back(cOp);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
284
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
285 if (blockStarts[i+1] > (blockStarts[i] + blockSizes[i])) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
286 CigarOp cOp;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
287 cOp.Length = (blockStarts[i+1] - (blockStarts[i] + blockSizes[i]));
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
288 cOp.Type = 'N';
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
289 bam.CigarData.push_back(cOp);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
290 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
291 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
292 // handle the last block.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
293 CigarOp cOp;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
294 cOp.Length = blockSizes[blockCount - 1];
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
295 cOp.Type = 'M';
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
296 bam.CigarData.push_back(cOp);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
297 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
298 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
299 // it doesn't smell like BED12. complain.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
300 else {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
301 cerr << "You've indicated that the input file is in BED12 format, yet the relevant fields cannot be found. Exiting." << endl << endl;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
302 exit(1);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
303 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
304 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
305 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
306
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
307
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
308 void MakeBamHeader(const string &genomeFile, RefVector &refs, string &header,
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
309 map<string, int, std::less<string> > &chromToId) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
310
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
311 // make a genome map of the genome file.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
312 GenomeFile genome(genomeFile);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
313
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
314 header += "@HD\tVN:1.0\tSO:unsorted\n";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
315 header += "@PG\tID:BEDTools_bedToBam\tVN:V";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
316 header += VERSION;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
317 header += "\n";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
318
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
319 int chromId = 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
320 vector<string> chromList = genome.getChromList();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
321 sort(chromList.begin(), chromList.end());
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
322
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
323 // create a BAM header (@SQ) entry for each chrom in the BEDTools genome file.
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
324 vector<string>::const_iterator genomeItr = chromList.begin();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
325 vector<string>::const_iterator genomeEnd = chromList.end();
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
326 for (; genomeItr != genomeEnd; ++genomeItr) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
327 chromToId[*genomeItr] = chromId;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
328 chromId++;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
329
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
330 // add to the header text
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
331 int size = genome.getChromSize(*genomeItr);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
332 string chromLine = "@SQ\tSN:" + *genomeItr + "\tAS:" + genomeFile + "\tLN:" + ToString(size) + "\n";
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
333 header += chromLine;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
334
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
335 // create a chrom entry and add it to
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
336 // the RefVector
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
337 RefData chrom;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
338 chrom.RefName = *genomeItr;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
339 chrom.RefLength = size;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
340 refs.push_back(chrom);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
341 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
342 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
343
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
344
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
345 /* Taken directly from the SAMTools spec
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
346 calculate bin given an alignment in [beg,end) (zero-based, half-close, half-open) */
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
347 int reg2bin(int beg, int end) {
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
348 --end;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
349 if (beg>>14 == end>>14) return ((1<<15)-1)/7 + (beg>>14);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
350 if (beg>>17 == end>>17) return ((1<<12)-1)/7 + (beg>>17);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
351 if (beg>>20 == end>>20) return ((1<<9)-1)/7 + (beg>>20);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
352 if (beg>>23 == end>>23) return ((1<<6)-1)/7 + (beg>>23);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
353 if (beg>>26 == end>>26) return ((1<<3)-1)/7 + (beg>>26);
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
354 return 0;
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
355 }
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
356
dfcd8b6c1bda Uploaded
aaronquinlan
parents:
diff changeset
357