// Marth Lab, Department of Biology, Boston College
// All rights reserved.
// ---------------------------------------------------------------------------
-// Last modified: 21 March 2011 (DB)
+// Last modified: 7 April 2011 (DB)
// ---------------------------------------------------------------------------
-// Sorts an input BAM file (default by position) and stores in a new BAM file.
+// Sorts an input BAM file
// ***************************************************************************
#include "bamtools_sort.h"
namespace BamTools {
- // defaults
- //
- // ** These defaults should be tweaked & 'optimized' per testing ** //
- //
- // I say 'optimized' because each system will naturally perform
- // differently. We will attempt to determine a sensible
- // compromise that should perform well on average.
- const unsigned int SORT_DEFAULT_MAX_BUFFER_COUNT = 500000; // max numberOfAlignments for buffer
- const unsigned int SORT_DEFAULT_MAX_BUFFER_MEMORY = 1024; // Mb
-
- // -----------------------------------
- // comparison objects (for sorting)
-
- struct SortLessThanPosition {
- bool operator() (const BamAlignment& lhs, const BamAlignment& rhs) {
- if ( lhs.RefID != rhs.RefID )
- return lhs.RefID < rhs.RefID;
- else
- return lhs.Position < rhs.Position;
- }
- };
-
- struct SortLessThanName {
- bool operator() (const BamAlignment& lhs, const BamAlignment& rhs) {
- return lhs.Name < rhs.Name;
- }
- };
-
-} // namespace BamTools
+// defaults
+//
+// ** These defaults should be tweaked & 'optimized' per testing ** //
+//
+// I say 'optimized' because each system will naturally perform
+// differently. We will attempt to determine a sensible
+// compromise that should perform well on average.
+const unsigned int SORT_DEFAULT_MAX_BUFFER_COUNT = 500000; // max numberOfAlignments for buffer
+const unsigned int SORT_DEFAULT_MAX_BUFFER_MEMORY = 1024; // Mb
+
+// -----------------------------------
+// comparison objects (for sorting)
+
+struct SortLessThanPosition {
+ bool operator() (const BamAlignment& lhs, const BamAlignment& rhs) {
+ if ( lhs.RefID != rhs.RefID )
+ return lhs.RefID < rhs.RefID;
+ else
+ return lhs.Position < rhs.Position;
+ }
+};
-// ---------------------------------------------
-// SortToolPrivate declaration
-class SortTool::SortToolPrivate {
-
- // ctor & dtor
- public:
- SortToolPrivate(SortTool::SortSettings* settings);
- ~SortToolPrivate(void);
-
- // 'public' interface
- public:
- bool Run(void);
-
- // internal methods
- private:
- void ClearBuffer(vector<BamAlignment>& buffer);
- bool GenerateSortedRuns(void);
- bool HandleBufferContents(vector<BamAlignment>& buffer);
- bool MergeSortedRuns(void);
- bool WriteTempFile(const vector<BamAlignment>& buffer, const string& tempFilename);
- void SortBuffer(vector<BamAlignment>& buffer);
-
- // data members
- private:
- SortTool::SortSettings* m_settings;
- string m_tempFilenameStub;
- int m_numberOfRuns;
- string m_headerText;
- RefVector m_references;
- vector<string> m_tempFilenames;
+struct SortLessThanName {
+ bool operator() (const BamAlignment& lhs, const BamAlignment& rhs) {
+ return lhs.Name < rhs.Name;
+ }
};
+
+} // namespace BamTools
// ---------------------------------------------
// SortSettings implementation
// filenames
string InputBamFilename;
string OutputBamFilename;
-
+
// parameters
unsigned int MaxBufferCount;
unsigned int MaxBufferMemory;
-
+
// constructor
SortSettings(void)
: HasInputBamFilename(false)
, OutputBamFilename(Options::StandardOut())
, MaxBufferCount(SORT_DEFAULT_MAX_BUFFER_COUNT)
, MaxBufferMemory(SORT_DEFAULT_MAX_BUFFER_MEMORY)
- { }
-};
+ { }
+};
// ---------------------------------------------
-// SortTool implementation
-
-SortTool::SortTool(void)
- : AbstractTool()
- , m_settings(new SortSettings)
- , m_impl(0)
-{
- // set program details
- Options::SetProgramInfo("bamtools sort", "sorts a BAM file", "[-in <filename>] [-out <filename>] [sortOptions]");
-
- // set up options
- OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
- Options::AddValueOption("-in", "BAM filename", "the input BAM file", "", m_settings->HasInputBamFilename, m_settings->InputBamFilename, IO_Opts, Options::StandardIn());
- Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutputBamFilename, m_settings->OutputBamFilename, IO_Opts, Options::StandardOut());
-
- OptionGroup* SortOpts = Options::CreateOptionGroup("Sorting Methods");
- Options::AddOption("-byname", "sort by alignment name", m_settings->IsSortingByName, SortOpts);
-
- OptionGroup* MemOpts = Options::CreateOptionGroup("Memory Settings");
- Options::AddValueOption("-n", "count", "max number of alignments per tempfile", "", m_settings->HasMaxBufferCount, m_settings->MaxBufferCount, MemOpts, SORT_DEFAULT_MAX_BUFFER_COUNT);
- Options::AddValueOption("-mem", "Mb", "max memory to use", "", m_settings->HasMaxBufferMemory, m_settings->MaxBufferMemory, MemOpts, SORT_DEFAULT_MAX_BUFFER_MEMORY);
-}
-
-SortTool::~SortTool(void) {
-
- delete m_settings;
- m_settings = 0;
-
- delete m_impl;
- m_impl = 0;
-}
-
-int SortTool::Help(void) {
- Options::DisplayHelp();
- return 0;
-}
+// SortToolPrivate implementation
-int SortTool::Run(int argc, char* argv[]) {
-
- // parse command line arguments
- Options::Parse(argc, argv, 1);
-
- // run internal SortTool implementation, return success/fail
- m_impl = new SortToolPrivate(m_settings);
-
- if ( m_impl->Run() ) return 0;
- else return 1;
-}
+class SortTool::SortToolPrivate {
+
+ // ctor & dtor
+ public:
+ SortToolPrivate(SortTool::SortSettings* settings);
+ ~SortToolPrivate(void) { }
+
+ // 'public' interface
+ public:
+ bool Run(void);
+
+ // internal methods
+ private:
+ void ClearBuffer(vector<BamAlignment>& buffer);
+ bool GenerateSortedRuns(void);
+ bool HandleBufferContents(vector<BamAlignment>& buffer);
+ bool MergeSortedRuns(void);
+ bool WriteTempFile(const vector<BamAlignment>& buffer, const string& tempFilename);
+ void SortBuffer(vector<BamAlignment>& buffer);
+
+ // data members
+ private:
+ SortTool::SortSettings* m_settings;
+ string m_tempFilenameStub;
+ int m_numberOfRuns;
+ string m_headerText;
+ RefVector m_references;
+ vector<string> m_tempFilenames;
+};
-// ---------------------------------------------
-// SortToolPrivate implementation
// constructor
SortTool::SortToolPrivate::SortToolPrivate(SortTool::SortSettings* settings)
}
}
-// destructor
-SortTool::SortToolPrivate::~SortToolPrivate(void) { }
-
// generates mutiple sorted temp BAM files from single unsorted BAM file
bool SortTool::SortToolPrivate::GenerateSortedRuns(void) {
tempWriter.Close();
return true;
}
+
+// ---------------------------------------------
+// SortTool implementation
+
+SortTool::SortTool(void)
+ : AbstractTool()
+ , m_settings(new SortSettings)
+ , m_impl(0)
+{
+ // set program details
+ Options::SetProgramInfo("bamtools sort", "sorts a BAM file", "[-in <filename>] [-out <filename>] [sortOptions]");
+
+ // set up options
+ OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
+ Options::AddValueOption("-in", "BAM filename", "the input BAM file", "", m_settings->HasInputBamFilename, m_settings->InputBamFilename, IO_Opts, Options::StandardIn());
+ Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutputBamFilename, m_settings->OutputBamFilename, IO_Opts, Options::StandardOut());
+
+ OptionGroup* SortOpts = Options::CreateOptionGroup("Sorting Methods");
+ Options::AddOption("-byname", "sort by alignment name", m_settings->IsSortingByName, SortOpts);
+
+ OptionGroup* MemOpts = Options::CreateOptionGroup("Memory Settings");
+ Options::AddValueOption("-n", "count", "max number of alignments per tempfile", "", m_settings->HasMaxBufferCount, m_settings->MaxBufferCount, MemOpts, SORT_DEFAULT_MAX_BUFFER_COUNT);
+ Options::AddValueOption("-mem", "Mb", "max memory to use", "", m_settings->HasMaxBufferMemory, m_settings->MaxBufferMemory, MemOpts, SORT_DEFAULT_MAX_BUFFER_MEMORY);
+}
+
+SortTool::~SortTool(void) {
+
+ delete m_settings;
+ m_settings = 0;
+
+ delete m_impl;
+ m_impl = 0;
+}
+
+int SortTool::Help(void) {
+ Options::DisplayHelp();
+ return 0;
+}
+
+int SortTool::Run(int argc, char* argv[]) {
+
+ // parse command line arguments
+ Options::Parse(argc, argv, 1);
+
+ // initialize SortTool with settings
+ m_impl = new SortToolPrivate(m_settings);
+
+ // run SortTool, return success/fail
+ if ( m_impl->Run() )
+ return 0;
+ else
+ return 1;
+}
projects / bamtools.git / blobdiff