// ***************************************************************************
// bamtools_merge.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
-// All rights reserved.
// ---------------------------------------------------------------------------
-// Last modified: 7 September 2010
+// Last modified: 7 April 2011
// ---------------------------------------------------------------------------
-// Merges multiple BAM files into one.
+// Merges multiple BAM files into one
// ***************************************************************************
+#include "bamtools_merge.h"
+
+#include <api/BamMultiReader.h>
+#include <api/BamWriter.h>
+#include <utils/bamtools_options.h>
+#include <utils/bamtools_utilities.h>
+using namespace BamTools;
+
#include <iostream>
#include <string>
#include <vector>
-
-#include "bamtools_merge.h"
-#include "bamtools_options.h"
-#include "bamtools_utilities.h"
-#include "BamMultiReader.h"
-#include "BamWriter.h"
-
using namespace std;
-using namespace BamTools;
// ---------------------------------------------
// MergeSettings implementation
// flags
bool HasInputBamFilename;
+ bool HasInputBamFilelist;
bool HasOutputBamFilename;
bool IsForceCompression;
bool HasRegion;
// filenames
vector<string> InputFiles;
+ string InputFilelist;
// other parameters
string OutputFilename;
// constructor
MergeSettings(void)
: HasInputBamFilename(false)
+ , HasInputBamFilelist(false)
, HasOutputBamFilename(false)
, IsForceCompression(false)
, HasRegion(false)
};
// ---------------------------------------------
-// MergeTool implementation
+// MergeToolPrivate implementation
-MergeTool::MergeTool(void)
- : AbstractTool()
- , m_settings(new MergeSettings)
-{
- // set program details
- Options::SetProgramInfo("bamtools merge", "merges multiple BAM files into one", "[-in <filename> -in <filename> ...] [-out <filename>]");
-
- // set up options
- OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
- Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInputBamFilename, m_settings->InputFiles, IO_Opts);
- Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutputBamFilename, m_settings->OutputFilename, IO_Opts);
- Options::AddOption("-forceCompression", "if results are sent to stdout (like when piping to another tool), default behavior is to leave output uncompressed. Use this flag to override and force compression", m_settings->IsForceCompression, IO_Opts);
-
- OptionGroup* FilterOpts = Options::CreateOptionGroup("Filters");
- Options::AddValueOption("-region", "REGION", "genomic region. See README for more details", "", m_settings->HasRegion, m_settings->Region, FilterOpts);
-}
+struct MergeTool::MergeToolPrivate {
-MergeTool::~MergeTool(void) {
- delete m_settings;
- m_settings = 0;
-}
+ // ctor & dtor
+ public:
+ MergeToolPrivate(MergeTool::MergeSettings* settings)
+ : m_settings(settings)
+ { }
-int MergeTool::Help(void) {
- Options::DisplayHelp();
- return 0;
-}
+ ~MergeToolPrivate(void) { }
-int MergeTool::Run(int argc, char* argv[]) {
-
- // parse command line arguments
- Options::Parse(argc, argv, 1);
-
- // set to default input if none provided
- if ( !m_settings->HasInputBamFilename )
+ // interface
+ public:
+ bool Run(void);
+
+ // data members
+ private:
+ MergeTool::MergeSettings* m_settings;
+};
+
+bool MergeTool::MergeToolPrivate::Run(void) {
+
+ // set to default input if none provided
+ if ( !m_settings->HasInputBamFilename && !m_settings->HasInputBamFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
-
+
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputBamFilelist ) {
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools merge ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+ string line;
+ while ( getline(filelist, line) ) {
+ m_settings->InputFiles.push_back(line);
+ }
+ }
+
// opens the BAM files (by default without checking for indexes)
BamMultiReader reader;
- if ( !reader.Open(m_settings->InputFiles, false, true) ) {
- cerr << "ERROR: Could not open input BAM file(s)... Aborting." << endl;
- return 1;
+ if ( !reader.Open(m_settings->InputFiles) ) {
+ cerr << "bamtools merge ERROR: could not open input BAM file(s)... Aborting." << endl;
+ return false;
}
-
+
// retrieve header & reference dictionary info
std::string mergedHeader = reader.GetHeaderText();
RefVector references = reader.GetReferenceData();
- // open writer
+ // determine compression mode for BamWriter
+ bool writeUncompressed = ( m_settings->OutputFilename == Options::StandardOut() &&
+ !m_settings->IsForceCompression );
+ BamWriter::CompressionMode compressionMode = BamWriter::Compressed;
+ if ( writeUncompressed ) compressionMode = BamWriter::Uncompressed;
+
+ // open BamWriter
BamWriter writer;
- bool writeUncompressed = ( m_settings->OutputFilename == Options::StandardOut() && !m_settings->IsForceCompression );
- if ( !writer.Open(m_settings->OutputFilename, mergedHeader, references, writeUncompressed) ) {
- cerr << "ERROR: Could not open BAM file " << m_settings->OutputFilename << " for writing... Aborting." << endl;
+ writer.SetCompressionMode(compressionMode);
+ if ( !writer.Open(m_settings->OutputFilename, mergedHeader, references) ) {
+ cerr << "bamtools merge ERROR: could not open "
+ << m_settings->OutputFilename << " for writing." << endl;
reader.Close();
- return 1;
+ return false;
}
-
+
// if no region specified, store entire contents of file(s)
if ( !m_settings->HasRegion ) {
BamAlignment al;
while ( reader.GetNextAlignmentCore(al) )
writer.SaveAlignment(al);
}
-
+
// otherwise attempt to use region as constraint
else {
-
+
// if region string parses OK
BamRegion region;
if ( Utilities::ParseRegionString(m_settings->Region, reader, region) ) {
- // attempt to re-open reader with index files
- reader.Close();
- bool openedOK = reader.Open(m_settings->InputFiles, true, true );
-
- // if error
- if ( !openedOK ) {
- cerr << "ERROR: Could not open input BAM file(s)... Aborting." << endl;
- return 1;
- }
-
- // if index data available, we can use SetRegion
- if ( reader.IsIndexLoaded() ) {
-
+ // attempt to find index files
+ reader.LocateIndexes();
+
+ // if index data available for all BAM files, we can use SetRegion
+ if ( reader.HasIndexes() ) {
+
// attempt to use SetRegion(), if failed report error
- if ( !reader.SetRegion(region.LeftRefID, region.LeftPosition, region.RightRefID, region.RightPosition) ) {
- cerr << "ERROR: Region requested, but could not set BamReader region to REGION: " << m_settings->Region << " Aborting." << endl;
+ if ( !reader.SetRegion(region.LeftRefID,
+ region.LeftPosition,
+ region.RightRefID,
+ region.RightPosition) )
+ {
+ cerr << "bamtools merge ERROR: set region failed. Check that REGION describes a valid range"
+ << endl;
reader.Close();
- return 1;
- }
-
+ return false;
+ }
+
// everything checks out, just iterate through specified region, storing alignments
BamAlignment al;
while ( reader.GetNextAlignmentCore(al) )
writer.SaveAlignment(al);
- }
-
+ }
+
// no index data available, we have to iterate through until we
// find overlapping alignments
else {
BamAlignment al;
while ( reader.GetNextAlignmentCore(al) ) {
if ( (al.RefID >= region.LeftRefID) && ( (al.Position + al.Length) >= region.LeftPosition ) &&
- (al.RefID <= region.RightRefID) && ( al.Position <= region.RightPosition) )
+ (al.RefID <= region.RightRefID) && ( al.Position <= region.RightPosition) )
{
writer.SaveAlignment(al);
}
}
}
- }
-
+ }
+
// error parsing REGION string
else {
- cerr << "ERROR: Could not parse REGION - " << m_settings->Region << endl;
- cerr << "Be sure REGION is in valid format (see README) and that coordinates are valid for selected references" << endl;
+ cerr << "bamtools merge ERROR: could not parse REGION - " << m_settings->Region << endl;
+ cerr << "Check that REGION is in valid format (see documentation) and that the coordinates are valid"
+ << endl;
reader.Close();
writer.Close();
- return 1;
+ return false;
}
}
-
+
// clean & exit
reader.Close();
writer.Close();
- return 0;
+ return true;
+}
+
+// ---------------------------------------------
+// MergeTool implementation
+
+MergeTool::MergeTool(void)
+ : AbstractTool()
+ , m_settings(new MergeSettings)
+ , m_impl(0)
+{
+ // set program details
+ Options::SetProgramInfo("bamtools merge", "merges multiple BAM files into one", "[ [-in <filename> -in <filename> ...] | [-list <filelist>] ] [-out <filename> | [-forceCompression]] [-region <REGION>]");
+
+ // set up options
+ OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
+ Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInputBamFilename, m_settings->InputFiles, IO_Opts);
+ Options::AddValueOption("-list", "BAM filename", "the input BAM file list, one line per file", "", m_settings->HasInputBamFilelist, m_settings->InputFilelist, IO_Opts);
+ Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutputBamFilename, m_settings->OutputFilename, IO_Opts);
+ Options::AddOption("-forceCompression", "if results are sent to stdout (like when piping to another tool), default behavior is to leave output uncompressed. Use this flag to override and force compression", m_settings->IsForceCompression, IO_Opts);
+ Options::AddValueOption("-region", "REGION", "genomic region. See README for more details", "", m_settings->HasRegion, m_settings->Region, IO_Opts);
+}
+
+MergeTool::~MergeTool(void) {
+
+ delete m_settings;
+ m_settings = 0;
+
+ delete m_impl;
+ m_impl = 0;
+}
+
+int MergeTool::Help(void) {
+ Options::DisplayHelp();
+ return 0;
+}
+
+int MergeTool::Run(int argc, char* argv[]) {
+
+ // parse command line arguments
+ Options::Parse(argc, argv, 1);
+
+ // initialize MergeTool with settings
+ m_impl = new MergeToolPrivate(m_settings);
+
+ // run MergeTool, return success/fail
+ if ( m_impl->Run() )
+ return 0;
+ else
+ return 1;
}
projects / bamtools.git / blobdiff