bamtools.cpp

   1 // ***************************************************************************
   2 // bamtools.cpp (c) 2010 Erik Garrison
   3 // Marth Lab, Department of Biology, Boston College
   4 // All rights reserved.
   5 // ---------------------------------------------------------------------------
   6 // Integrates a number of BamTools functionalities into a single executable.
   7 // ***************************************************************************
   8
   9 // Std C/C++ includes
  10 #include <cstdlib>
  11 #include <iostream>
  12 #include <fstream>
  13 #include <string>
  14 using namespace std;
  15
  16 // BamTools includes
  17 #include "BamReader.h"
  18 #include "BamWriter.h"
  19 #include "BamMultiReader.h"
  20 using namespace BamTools;
  21
  22 void usageSummary(string cmdname) {
  23     cerr << "usage: " << cmdname << " <command> [options]" << endl
  24          << "actions:" << endl
  25          << "    index <BAM file>   # generates BAM index <BAM file>.bai" << endl
  26          << "    merge <merged BAM file> [<BAM file> <BAM file> ...]   # merges BAM files into a single file" << endl
  27          << "    dump [<BAM file> <BAM file> ...]   # dumps alignment summaries to stdout" << endl
  28          << "    header [<BAM file> <BAM file> ...]   # prints header, or unified header for BAM file or files" << endl;
  29          //<< "    trim <input BAM file> <input BAM index file> <output BAM file> <reference name> <leftBound> <rightBound>" << endl;
  30 }
  31
  32
  33 void BamMerge(string outputFilename, vector<string> filenames) {
  34
  35     BamMultiReader reader;
  36
  37     reader.Open(filenames, false); // opens the files without checking for indexes
  38
  39     string mergedHeader = reader.GetHeaderText();
  40
  41     RefVector references = reader.GetReferenceData();
  42
  43     // open BamWriter
  44     BamWriter writer;
  45     writer.Open( outputFilename.c_str(), mergedHeader, references);
  46
  47     BamAlignment bAlignment;
  48     while (reader.GetNextAlignment(bAlignment)) {
  49         // write to output file
  50         writer.SaveAlignment(bAlignment);
  51     }
  52
  53     // close output file
  54     writer.Close();
  55     // close input files
  56     reader.Close();
  57
  58 }
  59
  60 void BamCreateIndex(const char* inputFilename) {
  61
  62         // open our BAM reader
  63         BamReader reader;
  64         reader.Open(inputFilename);
  65
  66     // create index file
  67     reader.CreateIndex();
  68
  69         // close our file
  70         reader.Close();
  71
  72 }
  73
  74 // Spit out basic BamAlignment data
  75 void PrintAlignment(const BamAlignment& alignment) {
  76         cout << "---------------------------------" << endl;
  77         cout << "Name: "       << alignment.Name << endl;
  78         cout << "Aligned to: " << alignment.RefID;
  79         cout << ":"            << alignment.Position << endl;
  80         cout << endl;
  81 }
  82
  83 void BamDump(vector<string> files) {
  84
  85         BamMultiReader reader;
  86         reader.Open(files);
  87
  88         BamAlignment bAlignment;
  89         while (reader.GetNextAlignment(bAlignment)) {
  90             PrintAlignment(bAlignment);
  91         }
  92
  93         reader.Close();
  94
  95 }
  96
  97 void BamDumpHeader(vector<string> files) {
  98
  99         BamMultiReader reader;
 100         reader.Open(files, false);
 101
 102     cout << reader.GetHeaderText() << endl;
 103
 104         reader.Close();
 105
 106 }
 107
 108
 109 int main(int argc, char* argv[]) {
 110
 111         // validate argument count
 112         if( argc < 3 ) {
 113         usageSummary(argv[0]);
 114                 exit(1);
 115         }
 116
 117     string command = argv[1];
 118
 119     if (command == "index") {
 120
 121         BamCreateIndex(argv[2]);
 122
 123     } else if (command == "merge") {
 124
 125         vector<string> files;
 126         string outputFile = argv[2];
 127
 128         // check if our output exists, and exit if so
 129         ifstream output(outputFile.c_str());
 130         if (output.good()) {
 131             cerr << "ERROR: output file " << outputFile << " exists, exiting." << endl;
 132             exit(1);
 133         } else {
 134             output.close();
 135         }
 136
 137         for (int i = 3; i<argc; ++i) {
 138             files.push_back(argv[i]);
 139         }
 140         BamMerge(outputFile, files);
 141
 142     } else if (command == "dump") {
 143
 144         vector<string> files;
 145         for (int i = 2; i<argc; ++i) {
 146             files.push_back(argv[i]);
 147         }
 148         BamDump(files);
 149
 150     } else if (command == "header") {
 151
 152         vector<string> files;
 153         for (int i = 2; i<argc; ++i) {
 154             files.push_back(argv[i]);
 155         }
 156         BamDumpHeader(files);
 157
 158     }
 159
 160
 161         return 0;
 162 }