[med-svn] [SCM] bamtools branch, debian, updated. e48c1baa00e4c59ef262fc6b68023232cd519365
Charles Plessy
plessy at debian.org
Mon Apr 1 06:57:34 UTC 2013
The following commit has been merged in the debian branch:
commit faee5bdc0f986ba7a2d2acf4a490116165361897
Author: Charles Plessy <plessy at debian.org>
Date: Mon Apr 1 15:29:03 2013 +0900
Imported Upstream version 2.2.3
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 9e3dc8e..3e81b1a 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -32,7 +32,7 @@ ensure_out_of_source_build( "
# set BamTools version information
set( BamTools_VERSION_MAJOR 2 )
set( BamTools_VERSION_MINOR 2 )
-set( BamTools_VERSION_BUILD 2 )
+set( BamTools_VERSION_BUILD 3 )
# set our library and executable destination dirs
set( EXECUTABLE_OUTPUT_PATH "${CMAKE_SOURCE_DIR}/bin" )
diff --git a/docs/Doxyfile b/docs/Doxyfile
index c2ff078..ff88c61 100644
--- a/docs/Doxyfile
+++ b/docs/Doxyfile
@@ -31,7 +31,7 @@ PROJECT_NAME = BamTools
# This could be handy for archiving the generated documentation or
# if some version control system is used.
-PROJECT_NUMBER = 2.2.2
+PROJECT_NUMBER = 2.2.3
# The OUTPUT_DIRECTORY tag is used to specify the (relative or absolute)
# base path where the generated documentation will be put.
diff --git a/src/api/BamMultiReader.cpp b/src/api/BamMultiReader.cpp
index f61aa26..5c2a065 100644
--- a/src/api/BamMultiReader.cpp
+++ b/src/api/BamMultiReader.cpp
@@ -2,7 +2,7 @@
// BamMultiReader.cpp (c) 2010 Erik Garrison, Derek Barnett
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 25 October 2011 (DB)
+// Last modified: 14 January 2013 (DB)
// ---------------------------------------------------------------------------
// Convenience class for reading multiple BAM files.
//
@@ -23,6 +23,21 @@ using namespace std;
/*! \class BamTools::BamMultiReader
\brief Convenience class for reading multiple BAM files.
*/
+/*! \enum BamMultiReader::MergeOrder
+ \brief Used to describe the merge strategy of the BamMultiReader.
+
+ The merge strategy determines which alignment is 'next' from across
+ all opened BAM files.
+*/
+/*! \var BamMultiReader::MergeOrder BamMultiReader::RoundRobinMerge
+ \brief Merge strategy when BAM files are unsorted, or their sorted status is either unknown or ignored
+*/
+/*! \var BamMultiReader::MergeOrder BamMultiReader::MergeByCoordinate
+ \brief Merge strategy when BAM files are sorted by position ('coordinate')
+*/
+/*! \var BamMultiReader::MergeOrder BamMultiReader::MergeByName
+ \brief Merge strategy when BAM files are sorted by read name ('queryname')
+*/
/*! \fn BamMultiReader::BamMultiReader(void)
\brief constructor
@@ -130,6 +145,16 @@ std::string BamMultiReader::GetHeaderText(void) const {
return d->GetHeaderText();
}
+/*! \fn BamMultiReader::MergeOrder BamMultiReader::GetMergeOrder(void) const
+ \brief Returns curent merge order strategy.
+
+ \returns current merge order enum value
+ \sa BamMultiReader::MergeOrder, SetExplicitMergeOrder()
+*/
+BamMultiReader::MergeOrder BamMultiReader::GetMergeOrder(void) const {
+ return d->GetMergeOrder();
+}
+
/*! \fn bool BamMultiReader::GetNextAlignment(BamAlignment& alignment)
\brief Retrieves next available alignment.
@@ -141,7 +166,7 @@ std::string BamMultiReader::GetHeaderText(void) const {
\param[out] alignment destination for alignment record data
\returns \c true if a valid alignment was found
- \sa GetNextAlignmentCore(), SetRegion(), BamReader::GetNextAlignment()
+ \sa GetNextAlignmentCore(), SetExplicitMergeOrder(), SetRegion(), BamReader::GetNextAlignment()
*/
bool BamMultiReader::GetNextAlignment(BamAlignment& nextAlignment) {
return d->GetNextAlignment(nextAlignment);
@@ -158,7 +183,7 @@ bool BamMultiReader::GetNextAlignment(BamAlignment& nextAlignment) {
\param[out] alignment destination for alignment record data
\returns \c true if a valid alignment was found
- \sa GetNextAlignment(), SetRegion(), BamReader::GetNextAlignmentCore()
+ \sa GetNextAlignment(), SetExplicitMergeOrder(), SetRegion(), BamReader::GetNextAlignmentCore()
*/
bool BamMultiReader::GetNextAlignmentCore(BamAlignment& nextAlignment) {
return d->GetNextAlignmentCore(nextAlignment);
@@ -321,6 +346,35 @@ bool BamMultiReader::Rewind(void) {
return d->Rewind();
}
+/*! \fn void BamMultiReader::SetExplicitMergeOrder(BamMultiReader::MergeOrder order)
+ \brief Sets an explicit merge order, regardless of the BAM files' SO header tag.
+
+ The default behavior of the BamMultiReader is to check the SO tag in the BAM files'
+ SAM header text to determine the merge strategy". The merge strategy is used to
+ determine from which BAM file the next alignment should come when either
+ GetNextAlignment() or GetNextAlignmentCore() are called. If files share a
+ 'coordinate' or 'queryname' value for this tag, then the merge strategy is
+ selected accordingly. If any of them do not match, or if any fileis marked as
+ 'unsorted', then the merge strategy is simply a round-robin.
+
+ This method allows client code to explicitly override the lookup behavior. This
+ method can be useful when you know, for example, that your BAM files are sorted
+ by coordinate but upstream processes did not set the header tag properly.
+
+ \note This method should \bold not be called while reading alignments via
+ GetNextAlignment() or GetNextAlignmentCore(). For proper results, you should
+ call this method before (or immediately after) opening files, rewinding,
+ jumping, etc. but \bold not once alignment fetching has started. There is
+ nothing in the API to prevent you from doing so, but the results may be
+ unexpected.
+
+ \returns \c true if merge order could be successfully applied
+ \sa BamMultiReader::MergeOrder, GetMergeOrder(), GetNextAlignment(), GetNextAlignmentCore()
+*/
+bool BamMultiReader::SetExplicitMergeOrder(BamMultiReader::MergeOrder order) {
+ return d->SetExplicitMergeOrder(order);
+}
+
/*! \fn bool BamMultiReader::SetRegion(const BamRegion& region)
\brief Sets a target region of interest
diff --git a/src/api/BamMultiReader.h b/src/api/BamMultiReader.h
index e5fc9c9..4f8c133 100644
--- a/src/api/BamMultiReader.h
+++ b/src/api/BamMultiReader.h
@@ -2,7 +2,7 @@
// BamMultiReader.h (c) 2010 Erik Garrison, Derek Barnett
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 25 October 2011 (DB)
+// Last modified: 14 January 2013 (DB)
// ---------------------------------------------------------------------------
// Convenience class for reading multiple BAM files.
// ***************************************************************************
@@ -25,6 +25,14 @@ namespace Internal {
class API_EXPORT BamMultiReader {
+ // enums
+ public:
+ // possible merge order strategies
+ enum MergeOrder { RoundRobinMerge = 0
+ , MergeByCoordinate
+ , MergeByName
+ };
+
// constructor / destructor
public:
BamMultiReader(void);
@@ -43,6 +51,8 @@ class API_EXPORT BamMultiReader {
bool CloseFile(const std::string& filename);
// returns list of filenames for all open BAM files
const std::vector<std::string> Filenames(void) const;
+ // returns curent merge order strategy
+ BamMultiReader::MergeOrder GetMergeOrder(void) const;
// returns true if multireader has any open BAM files
bool HasOpenReaders(void) const;
// performs random-access jump within current BAM files
@@ -53,6 +63,8 @@ class API_EXPORT BamMultiReader {
bool OpenFile(const std::string& filename);
// returns file pointers to beginning of alignments
bool Rewind(void);
+ // sets an explicit merge order, regardless of the BAM files' SO header tag
+ bool SetExplicitMergeOrder(BamMultiReader::MergeOrder order);
// sets the target region of interest
bool SetRegion(const BamRegion& region);
// sets the target region of interest
diff --git a/src/api/CMakeLists.txt b/src/api/CMakeLists.txt
index be46a7a..66eb35f 100644
--- a/src/api/CMakeLists.txt
+++ b/src/api/CMakeLists.txt
@@ -34,7 +34,7 @@ set( BamToolsAPISources
# create main BamTools API shared library
add_library( BamTools SHARED ${BamToolsAPISources} )
set_target_properties( BamTools PROPERTIES
- SOVERSION "2.2.2"
+ SOVERSION "2.2.3"
OUTPUT_NAME "bamtools" )
# create main BamTools API static library
diff --git a/src/api/internal/bam/BamMultiReader_p.cpp b/src/api/internal/bam/BamMultiReader_p.cpp
index d3f2b15..08d445d 100644
--- a/src/api/internal/bam/BamMultiReader_p.cpp
+++ b/src/api/internal/bam/BamMultiReader_p.cpp
@@ -2,7 +2,7 @@
// BamMultiReader_p.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 25 October 2011 (DB)
+// Last modified: 14 January 2013 (DB)
// ---------------------------------------------------------------------------
// Functionality for simultaneously reading multiple BAM files
// *************************************************************************
@@ -25,6 +25,8 @@ using namespace std;
// ctor
BamMultiReaderPrivate::BamMultiReaderPrivate(void)
: m_alignmentCache(0)
+ , m_hasUserMergeOrder(false)
+ , m_mergeOrder(BamMultiReader::RoundRobinMerge)
{ }
// dtor
@@ -115,11 +117,19 @@ bool BamMultiReaderPrivate::CloseFiles(const vector<string>& filenames) {
}
}
- // make sure alignment cache is cleaned up if all readers closed
- if ( m_readers.empty() && m_alignmentCache ) {
- m_alignmentCache->Clear();
- delete m_alignmentCache;
- m_alignmentCache = 0;
+ // make sure we clean up properly if all readers were closed
+ if ( m_readers.empty() ) {
+
+ // clean up merger
+ if ( m_alignmentCache ) {
+ m_alignmentCache->Clear();
+ delete m_alignmentCache;
+ m_alignmentCache = 0;
+ }
+
+ // reset merge flags
+ m_hasUserMergeOrder = false;
+ m_mergeOrder = BamMultiReader::RoundRobinMerge;
}
// return whether all readers closed OK
@@ -161,21 +171,46 @@ bool BamMultiReaderPrivate::CreateIndexes(const BamIndex::IndexType& type) {
return true;
}
-IMultiMerger* BamMultiReaderPrivate::CreateAlignmentCache(void) const {
+IMultiMerger* BamMultiReaderPrivate::CreateAlignmentCache(void) {
+
+ // if no merge order set explicitly, use SAM header to lookup proper order
+ if ( !m_hasUserMergeOrder ) {
+
+ // fetch SamHeader from BAM files
+ SamHeader header = GetHeader();
+
+ // if BAM files are sorted by position
+ if ( header.SortOrder == Constants::SAM_HD_SORTORDER_COORDINATE )
+ m_mergeOrder = BamMultiReader::MergeByCoordinate;
- // fetch SamHeader
- SamHeader header = GetHeader();
+ // if BAM files are sorted by read name
+ if ( header.SortOrder == Constants::SAM_HD_SORTORDER_QUERYNAME )
+ m_mergeOrder = BamMultiReader::MergeByName;
- // if BAM files are sorted by position
- if ( header.SortOrder == Constants::SAM_HD_SORTORDER_COORDINATE )
- return new MultiMerger<Algorithms::Sort::ByPosition>();
+ // otherwise, sorting is either "unknown" or marked as "unsorted"
+ else
+ m_mergeOrder = BamMultiReader::RoundRobinMerge;
+ }
+
+ // use current merge order to create proper 'multi-merger'
+ switch ( m_mergeOrder ) {
+
+ // merge BAM files by position
+ case BamMultiReader::MergeByCoordinate :
+ return new MultiMerger<Algorithms::Sort::ByPosition>();
+
+ // merge BAM files by read name
+ case BamMultiReader::MergeByName :
+ return new MultiMerger<Algorithms::Sort::ByName>();
- // if BAM files are sorted by read name
- if ( header.SortOrder == Constants::SAM_HD_SORTORDER_QUERYNAME )
- return new MultiMerger<Algorithms::Sort::ByName>();
+ // sorting is "unknown", "unsorted" or "ignored"... so use unsorted merger
+ case BamMultiReader::RoundRobinMerge :
+ return new MultiMerger<Algorithms::Sort::Unsorted>();
- // otherwise "unknown" or "unsorted", use unsorted merger and just read in
- return new MultiMerger<Algorithms::Sort::Unsorted>();
+ // unknown merge order, can't create merger
+ default:
+ return 0;
+ }
}
const vector<string> BamMultiReaderPrivate::Filenames(void) const {
@@ -248,6 +283,10 @@ string BamMultiReaderPrivate::GetHeaderText(void) const {
return mergedHeader.ToString();
}
+BamMultiReader::MergeOrder BamMultiReaderPrivate::GetMergeOrder(void) const {
+ return m_mergeOrder;
+}
+
// get next alignment among all files
bool BamMultiReaderPrivate::GetNextAlignment(BamAlignment& al) {
return PopNextCachedAlignment(al, true);
@@ -622,6 +661,40 @@ void BamMultiReaderPrivate::SaveNextAlignment(BamReader* reader, BamAlignment* a
m_alignmentCache->Add( MergeItem(reader, alignment) );
}
+bool BamMultiReaderPrivate::SetExplicitMergeOrder(BamMultiReader::MergeOrder order) {
+
+ // set new merge flags
+ m_hasUserMergeOrder = true;
+ m_mergeOrder = order;
+
+ // remove any existing merger (storing any existing data sitting in the cache)
+ vector<MergeItem> currentCacheData;
+ if ( m_alignmentCache ) {
+ while ( !m_alignmentCache->IsEmpty() )
+ currentCacheData.push_back( m_alignmentCache->TakeFirst() );
+ delete m_alignmentCache;
+ m_alignmentCache = 0;
+ }
+
+ // create new cache using the new merge flags
+ m_alignmentCache = CreateAlignmentCache();
+ if ( m_alignmentCache == 0 ) {
+ SetErrorString("BamMultiReader::SetExplicitMergeOrder", "requested order is unrecognized");
+ return false;
+ }
+
+ // push current data onto new cache
+ vector<MergeItem>::const_iterator readerIter = currentCacheData.begin();
+ vector<MergeItem>::const_iterator readerEnd = currentCacheData.end();
+ for ( ; readerIter != readerEnd; ++readerIter ) {
+ const MergeItem& item = (*readerIter);
+ m_alignmentCache->Add(item);
+ }
+
+ // return success
+ return true;
+}
+
void BamMultiReaderPrivate::SetErrorString(const string& where, const string& what) const {
static const string SEPARATOR = ": ";
m_errorString = where + SEPARATOR + what;
diff --git a/src/api/internal/bam/BamMultiReader_p.h b/src/api/internal/bam/BamMultiReader_p.h
index 9d7c39a..3a7a0b2 100644
--- a/src/api/internal/bam/BamMultiReader_p.h
+++ b/src/api/internal/bam/BamMultiReader_p.h
@@ -2,7 +2,7 @@
// BamMultiReader_p.h (c) 2010 Derek Barnett
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 25 October 2011 (DB)
+// Last modified: 14 January 2013 (DB)
// ---------------------------------------------------------------------------
// Functionality for simultaneously reading multiple BAM files
// *************************************************************************
@@ -54,9 +54,11 @@ class BamMultiReaderPrivate {
bool SetRegion(const BamRegion& region);
// access alignment data
+ BamMultiReader::MergeOrder GetMergeOrder(void) const;
bool GetNextAlignment(BamAlignment& al);
bool GetNextAlignmentCore(BamAlignment& al);
bool HasOpenReaders(void);
+ bool SetExplicitMergeOrder(BamMultiReader::MergeOrder order);
// access auxiliary data
SamHeader GetHeader(void) const;
@@ -78,7 +80,7 @@ class BamMultiReaderPrivate {
public:
bool CloseFiles(const std::vector<std::string>& filenames);
- IMultiMerger* CreateAlignmentCache(void) const;
+ IMultiMerger* CreateAlignmentCache(void);
bool PopNextCachedAlignment(BamAlignment& al, const bool needCharData);
bool RewindReaders(void);
void SaveNextAlignment(BamReader* reader, BamAlignment* alignment);
@@ -90,6 +92,10 @@ class BamMultiReaderPrivate {
public:
std::vector<MergeItem> m_readers;
IMultiMerger* m_alignmentCache;
+
+ bool m_hasUserMergeOrder;
+ BamMultiReader::MergeOrder m_mergeOrder;
+
mutable std::string m_errorString;
};
diff --git a/src/api/internal/bam/BamWriter_p.cpp b/src/api/internal/bam/BamWriter_p.cpp
index 1bb086f..637bb7a 100644
--- a/src/api/internal/bam/BamWriter_p.cpp
+++ b/src/api/internal/bam/BamWriter_p.cpp
@@ -284,7 +284,7 @@ void BamWriterPrivate::WriteAlignment(const BamAlignment& al) {
// write the base qualities
char* pBaseQualities = new char[queryLength]();
- if ( al.Qualities.empty() || al.Qualities[0] == '*' || al.Qualities[0] == (char)0xFF )
+ if ( al.Qualities.empty() || ( al.Qualities.size() == 1 && al.Qualities[0] == '*' ) || al.Qualities[0] == (char)0xFF )
memset(pBaseQualities, 0xFF, queryLength); // if missing or '*', fill with invalid qual
else {
for ( size_t i = 0; i < queryLength; ++i )
diff --git a/src/toolkit/CMakeLists.txt b/src/toolkit/CMakeLists.txt
index ba5b640..1be9efb 100644
--- a/src/toolkit/CMakeLists.txt
+++ b/src/toolkit/CMakeLists.txt
@@ -31,7 +31,7 @@ add_executable( bamtools_cmd
# set BamTools application properties
set_target_properties( bamtools_cmd PROPERTIES
- VERSION 2.2.2
+ VERSION 2.2.3
OUTPUT_NAME "bamtools"
)
# make version info available in application
diff --git a/src/toolkit/bamtools_convert.cpp b/src/toolkit/bamtools_convert.cpp
index 0e1743f..b0aae07 100644
--- a/src/toolkit/bamtools_convert.cpp
+++ b/src/toolkit/bamtools_convert.cpp
@@ -2,7 +2,7 @@
// bamtools_convert.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 11 November 2012
+// Last modified: 10 December 2012
// ---------------------------------------------------------------------------
// Converts between BAM and a number of other formats
// ***************************************************************************
@@ -76,6 +76,7 @@ struct ConvertTool::ConvertSettings {
// flag
bool HasInput;
+ bool HasInputFilelist;
bool HasOutput;
bool HasFormat;
bool HasRegion;
@@ -87,6 +88,7 @@ struct ConvertTool::ConvertSettings {
// options
vector<string> InputFiles;
+ string InputFilelist;
string OutputFilename;
string Format;
string Region;
@@ -97,6 +99,7 @@ struct ConvertTool::ConvertSettings {
// constructor
ConvertSettings(void)
: HasInput(false)
+ , HasInputFilelist(false)
, HasOutput(false)
, HasFormat(false)
, HasRegion(false)
@@ -151,9 +154,23 @@ bool ConvertTool::ConvertToolPrivate::Run(void) {
// initialize conversion input/output
// set to default input if none provided
- if ( !m_settings->HasInput )
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools convert ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+
+ string line;
+ while ( getline(filelist, line) )
+ m_settings->InputFiles.push_back(line);
+ }
+
// open input files
BamMultiReader reader;
if ( !reader.Open(m_settings->InputFiles) ) {
@@ -703,11 +720,13 @@ ConvertTool::ConvertTool(void)
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools convert", "converts BAM to a number of other formats", "-format <FORMAT> [-in <filename> -in <filename> ...] [-out <filename>] [-region <REGION>] [format-specific options]");
+ Options::SetProgramInfo("bamtools convert", "converts BAM to a number of other formats",
+ "-format <FORMAT> [-in <filename> -in <filename> ... | -list <filelist>] [-out <filename>] [-region <REGION>] [format-specific options]");
// set up options
OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
+ Options::AddValueOption("-list", "filename", "the input BAM file list, one line per file", "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutput, m_settings->OutputFilename, IO_Opts, Options::StandardOut());
Options::AddValueOption("-format", "FORMAT", "the output file format - see README for recognized formats", "", m_settings->HasFormat, m_settings->Format, IO_Opts);
Options::AddValueOption("-region", "REGION", "genomic region. Index file is recommended for better performance, and is used automatically if it exists. See \'bamtools help index\' for more details on creating one", "", m_settings->HasRegion, m_settings->Region, IO_Opts);
diff --git a/src/toolkit/bamtools_count.cpp b/src/toolkit/bamtools_count.cpp
index 3593f4d..5a7c0a7 100644
--- a/src/toolkit/bamtools_count.cpp
+++ b/src/toolkit/bamtools_count.cpp
@@ -2,7 +2,7 @@
// bamtools_count.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 7 April 2011
+// Last modified: 10 December 2012
// ---------------------------------------------------------------------------
// Prints alignment count for BAM file(s)
// ***************************************************************************
@@ -15,6 +15,7 @@
#include <utils/bamtools_utilities.h>
using namespace BamTools;
+#include <fstream>
#include <iostream>
#include <string>
#include <vector>
@@ -27,15 +28,18 @@ struct CountTool::CountSettings {
// flags
bool HasInput;
+ bool HasInputFilelist;
bool HasRegion;
// filenames
vector<string> InputFiles;
+ string InputFilelist;
string Region;
// constructor
CountSettings(void)
: HasInput(false)
+ , HasInputFilelist(false)
, HasRegion(false)
{ }
};
@@ -64,10 +68,24 @@ struct CountTool::CountToolPrivate {
bool CountTool::CountToolPrivate::Run(void) {
- // if no '-in' args supplied, default to stdin
- if ( !m_settings->HasInput )
+ // set to default input if none provided
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools count ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+
+ string line;
+ while ( getline(filelist, line) )
+ m_settings->InputFiles.push_back(line);
+ }
+
// open reader without index
BamMultiReader reader;
if ( !reader.Open(m_settings->InputFiles) ) {
@@ -150,12 +168,16 @@ CountTool::CountTool(void)
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools count", "prints number of alignments in BAM file(s)", "[-in <filename> -in <filename> ...] [-region <REGION>]");
+ Options::SetProgramInfo("bamtools count", "prints number of alignments in BAM file(s)",
+ "[-in <filename> -in <filename> ... | -list <filelist>] [-region <REGION>]");
// set up options
OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
- Options::AddValueOption("-region", "REGION", "genomic region. Index file is recommended for better performance, and is used automatically if it exists. See \'bamtools help index\' for more details on creating one", "", m_settings->HasRegion, m_settings->Region, IO_Opts);
+ Options::AddValueOption("-list", "filename", "the input BAM file list, one line per file", "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
+ Options::AddValueOption("-region", "REGION",
+ "genomic region. Index file is recommended for better performance, and is used automatically if it exists. See \'bamtools help index\' for more details on creating one",
+ "", m_settings->HasRegion, m_settings->Region, IO_Opts);
}
CountTool::~CountTool(void) {
diff --git a/src/toolkit/bamtools_filter.cpp b/src/toolkit/bamtools_filter.cpp
index 8af9cb9..1189981 100644
--- a/src/toolkit/bamtools_filter.cpp
+++ b/src/toolkit/bamtools_filter.cpp
@@ -2,7 +2,7 @@
// bamtools_filter.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 14 October 2011
+// Last modified: 10 December 2012
// ---------------------------------------------------------------------------
// Filters BAM file(s) according to some user-specified criteria
// ***************************************************************************
@@ -20,6 +20,7 @@ using namespace BamTools;
using namespace Json;
#include <cstdio>
+#include <fstream>
#include <iostream>
#include <sstream>
#include <string>
@@ -236,14 +237,16 @@ struct FilterTool::FilterSettings {
// IO opts
// flags
- bool HasInputBamFilename;
- bool HasOutputBamFilename;
+ bool HasInput;
+ bool HasInputFilelist;
+ bool HasOutput;
bool HasRegion;
- bool HasScriptFilename;
+ bool HasScript;
bool IsForceCompression;
// filenames
vector<string> InputFiles;
+ string InputFilelist;
string OutputFilename;
string Region;
string ScriptFilename;
@@ -302,10 +305,11 @@ struct FilterTool::FilterSettings {
// constructor
FilterSettings(void)
- : HasInputBamFilename(false)
- , HasOutputBamFilename(false)
+ : HasInput(false)
+ , HasInputFilelist(false)
+ , HasOutput(false)
, HasRegion(false)
- , HasScriptFilename(false)
+ , HasScript(false)
, IsForceCompression(false)
, OutputFilename(Options::StandardOut())
, HasAlignmentFlagFilter(false)
@@ -463,11 +467,11 @@ bool FilterTool::FilterToolPrivate::AddPropertyTokensToFilter(const string& filt
m_filterEngine.setProperty(filterName, propertyName, stringValue, type);
}
- else if ( propertyName == TAG_PROPERTY ) {
- // this will be stored directly as the TAG:VALUE token
- // (VALUE may contain compare ops, will be parsed out later)
- m_filterEngine.setProperty(filterName, propertyName, token, PropertyFilterValue::EXACT);
- }
+ else if ( propertyName == TAG_PROPERTY ) {
+ // this will be stored directly as the TAG:VALUE token
+ // (VALUE may contain compare ops, will be parsed out later)
+ m_filterEngine.setProperty(filterName, propertyName, token, PropertyFilterValue::EXACT);
+ }
// else unknown property
else {
@@ -500,7 +504,8 @@ const string FilterTool::FilterToolPrivate::GetScriptContents(void) {
// peek ahead, make sure there is data available
char ch = fgetc(inFile);
ungetc(ch, inFile);
- if( feof(inFile) ) break;
+ if( feof(inFile) )
+ break;
// read next block of data
if ( fgets(buffer, 1024, inFile) == 0 ) {
@@ -682,12 +687,27 @@ bool FilterTool::FilterToolPrivate::ParseScript(void) {
bool FilterTool::FilterToolPrivate::Run(void) {
// set to default input if none provided
- if ( !m_settings->HasInputBamFilename )
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools filter ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+
+ string line;
+ while ( getline(filelist, line) )
+ m_settings->InputFiles.push_back(line);
+ }
+
// initialize defined properties & user-specified filters
// quit if failed
- if ( !SetupFilters() ) return false;
+ if ( !SetupFilters() )
+ return false;
// open reader without index
BamMultiReader reader;
@@ -786,7 +806,7 @@ bool FilterTool::FilterToolPrivate::SetupFilters(void) {
InitProperties();
// parse script for filter rules, if given
- if ( m_settings->HasScriptFilename )
+ if ( m_settings->HasScript )
return ParseScript();
// otherwise check command line for filters
@@ -804,9 +824,10 @@ FilterTool::FilterTool(void)
// ----------------------------------
// set program details
- const string usage = "[-in <filename> -in <filename> ...] "
+ const string usage = "[-in <filename> -in <filename> ... | -list <filelist>] "
"[-out <filename> | [-forceCompression]] [-region <REGION>] "
"[ [-script <filename] | [filterOptions] ]";
+
Options::SetProgramInfo("bamtools filter", "filters BAM file(s)", usage );
// ----------------------------------
@@ -815,6 +836,7 @@ FilterTool::FilterTool(void)
OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
const string inDesc = "the input BAM file(s)";
+ const string listDesc = "the input BAM file list, one line per file";
const string outDesc = "the output BAM file";
const string regionDesc = "only read data from this genomic region (see documentation for more details)";
const string scriptDesc = "the filter script file (see documentation for more details)";
@@ -822,10 +844,11 @@ FilterTool::FilterTool(void)
"default behavior is to leave output uncompressed. Use this flag to "
"override and force compression";
- Options::AddValueOption("-in", "BAM filename", inDesc, "", m_settings->HasInputBamFilename, m_settings->InputFiles, IO_Opts, Options::StandardIn());
- Options::AddValueOption("-out", "BAM filename", outDesc, "", m_settings->HasOutputBamFilename, m_settings->OutputFilename, IO_Opts, Options::StandardOut());
- Options::AddValueOption("-region", "REGION", regionDesc, "", m_settings->HasRegion, m_settings->Region, IO_Opts);
- Options::AddValueOption("-script", "filename", scriptDesc, "", m_settings->HasScriptFilename, m_settings->ScriptFilename, IO_Opts);
+ Options::AddValueOption("-in", "BAM filename", inDesc, "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
+ Options::AddValueOption("-list", "filename", listDesc, "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
+ Options::AddValueOption("-out", "BAM filename", outDesc, "", m_settings->HasOutput, m_settings->OutputFilename, IO_Opts, Options::StandardOut());
+ Options::AddValueOption("-region", "REGION", regionDesc, "", m_settings->HasRegion, m_settings->Region, IO_Opts);
+ Options::AddValueOption("-script", "filename", scriptDesc, "", m_settings->HasScript, m_settings->ScriptFilename, IO_Opts);
Options::AddOption("-forceCompression",forceDesc, m_settings->IsForceCompression, IO_Opts);
// ----------------------------------
diff --git a/src/toolkit/bamtools_header.cpp b/src/toolkit/bamtools_header.cpp
index a08c632..534bb14 100644
--- a/src/toolkit/bamtools_header.cpp
+++ b/src/toolkit/bamtools_header.cpp
@@ -2,7 +2,7 @@
// bamtools_header.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 7 April 2011
+// Last modified: 10 December 2012
// ---------------------------------------------------------------------------
// Prints the SAM-style header from a single BAM file ( or merged header from
// multiple BAM files) to stdout
@@ -14,6 +14,7 @@
#include <utils/bamtools_options.h>
using namespace BamTools;
+#include <fstream>
#include <iostream>
#include <string>
#include <vector>
@@ -25,14 +26,17 @@ using namespace std;
struct HeaderTool::HeaderSettings {
// flags
- bool HasInputBamFilename;
+ bool HasInput;
+ bool HasInputFilelist;
// filenames
vector<string> InputFiles;
+ string InputFilelist;
// constructor
HeaderSettings(void)
- : HasInputBamFilename(false)
+ : HasInput(false)
+ , HasInputFilelist(false)
{ }
};
@@ -58,9 +62,23 @@ struct HeaderTool::HeaderToolPrivate {
bool HeaderTool::HeaderToolPrivate::Run(void) {
// set to default input if none provided
- if ( !m_settings->HasInputBamFilename )
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools header ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+
+ string line;
+ while ( getline(filelist, line) )
+ m_settings->InputFiles.push_back(line);
+ }
+
// attemp to open BAM files
BamMultiReader reader;
if ( !reader.Open(m_settings->InputFiles) ) {
@@ -85,11 +103,12 @@ HeaderTool::HeaderTool(void)
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools header", "prints header from BAM file(s)", "[-in <filename> -in <filename> ...] ");
+ Options::SetProgramInfo("bamtools header", "prints header from BAM file(s)", "[-in <filename> -in <filename> ... | -list <filelist>]");
// 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::StandardIn());
+ Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
+ Options::AddValueOption("-list", "filename", "the input BAM file list, one line per file", "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
}
HeaderTool::~HeaderTool(void) {
diff --git a/src/toolkit/bamtools_merge.cpp b/src/toolkit/bamtools_merge.cpp
index 1e8312f..6a33d12 100644
--- a/src/toolkit/bamtools_merge.cpp
+++ b/src/toolkit/bamtools_merge.cpp
@@ -2,7 +2,7 @@
// bamtools_merge.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 7 April 2011
+// Last modified: 10 December 2012
// ---------------------------------------------------------------------------
// Merges multiple BAM files into one
// ***************************************************************************
@@ -15,6 +15,7 @@
#include <utils/bamtools_utilities.h>
using namespace BamTools;
+#include <fstream>
#include <iostream>
#include <string>
#include <vector>
@@ -26,13 +27,15 @@ using namespace std;
struct MergeTool::MergeSettings {
// flags
- bool HasInputBamFilename;
- bool HasOutputBamFilename;
+ bool HasInput;
+ bool HasInputFilelist;
+ bool HasOutput;
bool IsForceCompression;
bool HasRegion;
// filenames
vector<string> InputFiles;
+ string InputFilelist;
// other parameters
string OutputFilename;
@@ -40,8 +43,9 @@ struct MergeTool::MergeSettings {
// constructor
MergeSettings(void)
- : HasInputBamFilename(false)
- , HasOutputBamFilename(false)
+ : HasInput(false)
+ , HasInputFilelist(false)
+ , HasOutput(false)
, IsForceCompression(false)
, HasRegion(false)
, OutputFilename(Options::StandardOut())
@@ -73,9 +77,23 @@ struct MergeTool::MergeToolPrivate {
bool MergeTool::MergeToolPrivate::Run(void) {
// set to default input if none provided
- if ( !m_settings->HasInputBamFilename )
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ 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) ) {
@@ -181,12 +199,14 @@ MergeTool::MergeTool(void)
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools merge", "merges multiple BAM files into one", "[-in <filename> -in <filename> ...] [-out <filename> | [-forceCompression]] [-region <REGION>]");
+ 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("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutputBamFilename, m_settings->OutputFilename, IO_Opts);
+ Options::AddValueOption("-in", "BAM filename", "the input BAM file(s)", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts);
+ Options::AddValueOption("-list", "filename", "the input BAM file list, one line per file", "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
+ Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutput, 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);
}
diff --git a/src/toolkit/bamtools_random.cpp b/src/toolkit/bamtools_random.cpp
index e28ea70..5282f15 100644
--- a/src/toolkit/bamtools_random.cpp
+++ b/src/toolkit/bamtools_random.cpp
@@ -2,7 +2,7 @@
// bamtools_random.cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 7 April 2011 (DB)
+// Last modified: 10 December 2012 (DB)
// ---------------------------------------------------------------------------
// Grab a random subset of alignments (testing tool)
// ***************************************************************************
@@ -17,6 +17,7 @@ using namespace BamTools;
#include <ctime>
#include <cstdlib>
+#include <fstream>
#include <iostream>
#include <string>
#include <vector>
@@ -43,6 +44,7 @@ struct RandomTool::RandomSettings {
// flags
bool HasAlignmentCount;
bool HasInput;
+ bool HasInputFilelist;
bool HasOutput;
bool HasRegion;
bool IsForceCompression;
@@ -50,6 +52,7 @@ struct RandomTool::RandomSettings {
// parameters
unsigned int AlignmentCount;
vector<string> InputFiles;
+ string InputFilelist;
string OutputFilename;
string Region;
@@ -57,6 +60,7 @@ struct RandomTool::RandomSettings {
RandomSettings(void)
: HasAlignmentCount(false)
, HasInput(false)
+ , HasInputFilelist(false)
, HasOutput(false)
, HasRegion(false)
, IsForceCompression(false)
@@ -90,9 +94,23 @@ struct RandomTool::RandomToolPrivate {
bool RandomTool::RandomToolPrivate::Run(void) {
// set to default stdin if no input files provided
- if ( !m_settings->HasInput )
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools random ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+
+ string line;
+ while ( getline(filelist, line) )
+ m_settings->InputFiles.push_back(line);
+ }
+
// open our reader
BamMultiReader reader;
if ( !reader.Open(m_settings->InputFiles) ) {
@@ -212,11 +230,13 @@ RandomTool::RandomTool(void)
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools random", "grab a random subset of alignments", "[-in <filename> -in <filename> ...] [-out <filename>] [-forceCompression] [-n] [-region <REGION>]");
+ Options::SetProgramInfo("bamtools random", "grab a random subset of alignments",
+ "[-in <filename> -in <filename> ... | -list <filelist>] [-out <filename>] [-forceCompression] [-n] [-region <REGION>]");
// set up options
OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
Options::AddValueOption("-in", "BAM filename", "the input BAM file", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
+ Options::AddValueOption("-list", "filename", "the input BAM file list, one line per file", "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
Options::AddValueOption("-out", "BAM filename", "the output BAM file", "", m_settings->HasOutput, m_settings->OutputFilename, IO_Opts, Options::StandardOut());
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", "only pull random alignments from within this genomic region. Index file is recommended for better performance, and is used automatically if it exists. See \'bamtools help index\' for more details on creating one", "", m_settings->HasRegion, m_settings->Region, IO_Opts);
diff --git a/src/toolkit/bamtools_stats.cpp b/src/toolkit/bamtools_stats.cpp
index eb57a95..0035913 100644
--- a/src/toolkit/bamtools_stats.cpp
+++ b/src/toolkit/bamtools_stats.cpp
@@ -2,7 +2,7 @@
// bamtools_cpp (c) 2010 Derek Barnett, Erik Garrison
// Marth Lab, Department of Biology, Boston College
// ---------------------------------------------------------------------------
-// Last modified: 7 April 2011
+// Last modified: 10 December 2012
// ---------------------------------------------------------------------------
// Prints general alignment statistics for BAM file(s).
// ***************************************************************************
@@ -15,6 +15,7 @@ using namespace BamTools;
#include <cmath>
#include <algorithm>
+#include <fstream>
#include <functional>
#include <iostream>
#include <numeric>
@@ -29,14 +30,17 @@ struct StatsTool::StatsSettings {
// flags
bool HasInput;
+ bool HasInputFilelist;
bool IsShowingInsertSizeSummary;
// filenames
vector<string> InputFiles;
+ string InputFilelist;
// constructor
StatsSettings(void)
: HasInput(false)
+ , HasInputFilelist(false)
, IsShowingInsertSizeSummary(false)
{ }
};
@@ -102,7 +106,8 @@ StatsTool::StatsToolPrivate::StatsToolPrivate(StatsTool::StatsSettings* settings
bool StatsTool::StatsToolPrivate::CalculateMedian(vector<int>& data, double& median) {
// skip if data empty
- if ( data.empty() ) return false;
+ if ( data.empty() )
+ return false;
// find middle element
size_t middleIndex = data.size() / 2;
@@ -202,7 +207,8 @@ void StatsTool::StatsToolPrivate::ProcessAlignment(const BamAlignment& al) {
}
// check for explicit proper pair flag
- if ( al.IsProperPair() ) ++m_numProperPair;
+ if ( al.IsProperPair() )
+ ++m_numProperPair;
// store insert size for first mate
if ( m_settings->IsShowingInsertSizeSummary && al.IsFirstMate() && (al.InsertSize != 0) ) {
@@ -215,9 +221,23 @@ void StatsTool::StatsToolPrivate::ProcessAlignment(const BamAlignment& al) {
bool StatsTool::StatsToolPrivate::Run() {
// set to default input if none provided
- if ( !m_settings->HasInput )
+ if ( !m_settings->HasInput && !m_settings->HasInputFilelist )
m_settings->InputFiles.push_back(Options::StandardIn());
+ // add files in the filelist to the input file list
+ if ( m_settings->HasInputFilelist ) {
+
+ ifstream filelist(m_settings->InputFilelist.c_str(), ios::in);
+ if ( !filelist.is_open() ) {
+ cerr << "bamtools stats ERROR: could not open input BAM file list... Aborting." << endl;
+ return false;
+ }
+
+ string line;
+ while ( getline(filelist, line) )
+ m_settings->InputFiles.push_back(line);
+ }
+
// open the BAM files
BamMultiReader reader;
if ( !reader.Open(m_settings->InputFiles) ) {
@@ -246,11 +266,12 @@ StatsTool::StatsTool(void)
, m_impl(0)
{
// set program details
- Options::SetProgramInfo("bamtools stats", "prints general alignment statistics", "[-in <filename> -in <filename> ...] [statsOptions]");
+ Options::SetProgramInfo("bamtools stats", "prints general alignment statistics", "[-in <filename> -in <filename> ... | -list <filelist>] [statsOptions]");
// set up options
OptionGroup* IO_Opts = Options::CreateOptionGroup("Input & Output");
Options::AddValueOption("-in", "BAM filename", "the input BAM file", "", m_settings->HasInput, m_settings->InputFiles, IO_Opts, Options::StandardIn());
+ Options::AddValueOption("-list", "filename", "the input BAM file list, one line per file", "", m_settings->HasInputFilelist, m_settings->InputFilelist, IO_Opts);
OptionGroup* AdditionalOpts = Options::CreateOptionGroup("Additional Stats");
Options::AddOption("-insert", "summarize insert size data", m_settings->IsShowingInsertSizeSummary, AdditionalOpts);
--
debian packaging for bamtools
More information about the debian-med-commit
mailing list