Merge branch 'master' of https://github.com/pezmaster31/bamtools

diff --git a/src/api/CMakeLists.txt b/src/api/CMakeLists.txt
index 66eb35f9f301478976acb9e24f02318a97fe07c6..a9dd0924a2925e7c661f71e6267b02489961b943 100644 (file)
--- a/src/api/CMakeLists.txt
+++ b/src/api/CMakeLists.txt
@@ -13,7 +13,7 @@ add_definitions( -DBAMTOOLS_API_LIBRARY ) # (for proper exporting of library sym
 add_definitions( -fPIC ) # (attempt to force PIC compiling on CentOS, not being set on shared libs by CMake)
 
 # fetch all internal source files
-add_subdirectory ( internal )
+add_subdirectory( internal )
 
 # make list of all API source files
 set( BamToolsAPISources
@@ -44,11 +44,11 @@ set_target_properties( BamTools-static PROPERTIES
                        PREFIX "lib" )
 
 # link libraries automatically with zlib (and Winsock2, if applicable)
-if( _WIN32 )
+if( WIN32 )
     set( APILibs z ws2_32 )
-else( _WIN32 )
+else()
     set( APILibs z )
-endif( _WIN32 )
+endif()
 
 target_link_libraries( BamTools ${APILibs} )
 target_link_libraries( BamTools-static ${APILibs} )

diff --git a/src/api/internal/index/CMakeLists.txt b/src/api/internal/index/CMakeLists.txt
index 1c78cb9a5d840ae0f9c3c361b5587241d7d7577d..d6a7df6ef5492db05292d392b63ebe1778bd7be0 100644 (file)
--- a/src/api/internal/index/CMakeLists.txt
+++ b/src/api/internal/index/CMakeLists.txt
@@ -5,9 +5,9 @@
 # src/api/internal/index
 # ==========================
 
-set ( InternalIndexDir "${InternalDir}/index" )
+set( InternalIndexDir "${InternalDir}/index" )
 
-set ( InternalIndexSources
+set( InternalIndexSources
         ${InternalIndexDir}/BamIndexFactory_p.cpp
         ${InternalIndexDir}/BamStandardIndex_p.cpp
         ${InternalIndexDir}/BamToolsIndex_p.cpp

diff --git a/src/api/internal/io/CMakeLists.txt b/src/api/internal/io/CMakeLists.txt
index d9da416d5443d30abdec9a27af6296be469eb0fb..28153d58465fc7f993bf62b901894fe257417c7b 100644 (file)
--- a/src/api/internal/io/CMakeLists.txt
+++ b/src/api/internal/io/CMakeLists.txt
@@ -5,12 +5,12 @@
 # src/api/internal/io
 # ==========================
 
-set ( InternalIODir "${InternalDir}/io" )
+set( InternalIODir "${InternalDir}/io" )
 
 #--------------------------
 # platform-independent IO
 #--------------------------
-set ( CommonIOSources
+set( CommonIOSources
         ${InternalIODir}/BamDeviceFactory_p.cpp
         ${InternalIODir}/BamFile_p.cpp
         ${InternalIODir}/BamFtp_p.cpp
@@ -30,21 +30,17 @@ set ( CommonIOSources
 #------------------------
 # platform-dependent IO
 #------------------------
-if ( _WIN32 )
-    set ( PlatformIOSources
-            ${InternalIODir}/TcpSocketEngine_win_p.cpp
-    )
-else ( _WIN32 )
-    set ( PlatformIOSources
-            ${InternalIODir}/TcpSocketEngine_unix_p.cpp
-    )
-endif ( _WIN32 )
+if( WIN32 )
+    set( PlatformIOSources ${InternalIODir}/TcpSocketEngine_win_p.cpp )
+else()
+    set( PlatformIOSources ${InternalIODir}/TcpSocketEngine_unix_p.cpp )
+endif()
 
 #---------------------------
 # make build-specific list
 #---------------------------
-set ( InternalIOSources 
-        ${CommonIOSources} 
+set( InternalIOSources
+        ${CommonIOSources}
         ${PlatformIOSources} 
 
         PARENT_SCOPE # <-- leave this last

diff --git a/src/api/internal/sam/CMakeLists.txt b/src/api/internal/sam/CMakeLists.txt
index 4b2bce2bea3b637a8623d33922eef3602859c1b0..2f303bd71a1034119ac23e74d9bc65c316945f38 100644 (file)
--- a/src/api/internal/sam/CMakeLists.txt
+++ b/src/api/internal/sam/CMakeLists.txt
@@ -5,9 +5,9 @@
 # src/api/internal/sam
 # ==========================
 
-set ( InternalSamDir "${InternalDir}/sam" )
+set( InternalSamDir "${InternalDir}/sam" )
 
-set ( InternalSamSources
+set( InternalSamSources
         ${InternalSamDir}/SamFormatParser_p.cpp
         ${InternalSamDir}/SamFormatPrinter_p.cpp
         ${InternalSamDir}/SamHeaderValidator_p.cpp

diff --git a/src/api/internal/utils/CMakeLists.txt b/src/api/internal/utils/CMakeLists.txt
index 38a695709dc96b38cd11d1846ead82898afd29d5..4b1e2c2fd323578fd5de10e236df18de8f5b4174 100644 (file)
--- a/src/api/internal/utils/CMakeLists.txt
+++ b/src/api/internal/utils/CMakeLists.txt
@@ -5,9 +5,9 @@
 # src/api/internal/utils
 # ==========================
 
-set ( InternalUtilsDir "${InternalDir}/utils" )
+set( InternalUtilsDir "${InternalDir}/utils" )
 
-set ( InternalUtilsSources
+set( InternalUtilsSources
         ${InternalUtilsDir}/BamException_p.cpp
 
         PARENT_SCOPE # <-- leave this last

diff --git a/src/toolkit/bamtools_coverage.cpp b/src/toolkit/bamtools_coverage.cpp
index c0ecd8f455034c0b3918b5f83be33b637e1f6326..6a4493de9992a5366f12e4997cd12dffeb2adcde 100644 (file)
--- a/src/toolkit/bamtools_coverage.cpp
+++ b/src/toolkit/bamtools_coverage.cpp
@@ -2,7 +2,7 @@
 // bamtools_coverage.cpp (c) 2010 Derek Barnett, Erik Garrison
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 7 April 2011
+// Last modified: 24 July 2013
 // ---------------------------------------------------------------------------
 // Prints coverage data for a single BAM file 
 // ***************************************************************************
@@ -137,6 +137,7 @@ bool CoverageTool::CoverageToolPrivate::Run(void) {
     BamAlignment al;    
     while ( reader.GetNextAlignment(al) ) 
         pileup.AddAlignment(al);
+    pileup.Flush();
     
     // clean up 
     reader.Close();

diff --git a/src/toolkit/bamtools_random.cpp b/src/toolkit/bamtools_random.cpp
index 5282f154f609187b8070b5c3c3f6d617869a1edc..367ac58a475882862c3a678200027b9fb30b1a35 100644 (file)
--- 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: 10 December 2012 (DB)
+// Last modified: 24 July 2013 (DB)
 // ---------------------------------------------------------------------------
 // Grab a random subset of alignments (testing tool)
 // ***************************************************************************
@@ -46,6 +46,7 @@ struct RandomTool::RandomSettings {
     bool HasInput;
     bool HasInputFilelist;
     bool HasOutput;
+    bool HasRandomNumberSeed;
     bool HasRegion;
     bool IsForceCompression;
 
@@ -54,6 +55,7 @@ struct RandomTool::RandomSettings {
     vector<string> InputFiles;
     string InputFilelist;
     string OutputFilename;
+    unsigned int RandomNumberSeed;
     string Region;
     
     // constructor
@@ -62,10 +64,12 @@ struct RandomTool::RandomSettings {
         , HasInput(false)
         , HasInputFilelist(false)
         , HasOutput(false)
+        , HasRandomNumberSeed(false)
         , HasRegion(false)
         , IsForceCompression(false)
         , AlignmentCount(RANDOM_MAX_ALIGNMENT_COUNT)
         , OutputFilename(Options::StandardOut())
+        , RandomNumberSeed(0)
     { }  
 };  
 
@@ -165,7 +169,10 @@ bool RandomTool::RandomToolPrivate::Run(void) {
     }
 
     // seed our random number generator
-    srand( time(NULL) );
+    if ( m_settings->HasRandomNumberSeed )
+        srand( m_settings->RandomNumberSeed );
+    else
+        srand( time(NULL) );
 
     // grab random alignments
     BamAlignment al;
@@ -235,14 +242,17 @@ RandomTool::RandomTool(void)
     
     // 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::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::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);
     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);
     
     OptionGroup* SettingsOpts = Options::CreateOptionGroup("Settings");
-    Options::AddValueOption("-n", "count", "number of alignments to grab. Note - no duplicate checking is performed", "", m_settings->HasAlignmentCount, m_settings->AlignmentCount, SettingsOpts, RANDOM_MAX_ALIGNMENT_COUNT);
+    Options::AddValueOption("-n", "count", "number of alignments to grab. Note - no duplicate checking is performed", "",
+                            m_settings->HasAlignmentCount, m_settings->AlignmentCount, SettingsOpts, RANDOM_MAX_ALIGNMENT_COUNT);
+    Options::AddValueOption("-seed", "unsigned integer", "random number generator seed (for repeatable results). Current time is used if no seed value is provided.", "",
+                            m_settings->HasRandomNumberSeed, m_settings->RandomNumberSeed, SettingsOpts);
 }
 
 RandomTool::~RandomTool(void) { 

diff --git a/src/toolkit/bamtools_resolve.cpp b/src/toolkit/bamtools_resolve.cpp
index cb42f5b4243322fbc026c6c2b25610f4b02bf0ed..9e5fb8489d1daaf46fb5abc5404fa143c610f611 100644 (file)
--- a/src/toolkit/bamtools_resolve.cpp
+++ b/src/toolkit/bamtools_resolve.cpp
@@ -2,7 +2,7 @@
 // bamtools_resolve.cpp (c) 2011
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 14 October 2011
+// Last modified: 24 July 2013 (DB)
 // ---------------------------------------------------------------------------
 // Resolves paired-end reads (marking the IsProperPair flag as needed).
 // ***************************************************************************
@@ -73,6 +73,20 @@ static const string OPTION_FORCEMARKREADGROUPS  = "ForceMarkReadGroups";
 static const string RG_FIELD_DESCRIPTION =
     "#<name> <medianFL> <minFL> <maxFL> <topModelID> <nextTopModelID> <isAmbiguous?>";
 
+static const string MODEL_DESCRIPTION =
+    "# ------------- Model Types Description ---------------\n"
+    "#\n"
+    "#   ID     Position              Orientation           \n"
+    "#    1   mate1 < mate2   mate1:forward, mate2:forward  \n"
+    "#    2   mate1 < mate2   mate1:forward, mate2:reverse  \n"
+    "#    3   mate1 < mate2   mate1:reverse, mate2:forward  \n"
+    "#    4   mate1 < mate2   mate1:reverse, mate2:reverse  \n"
+    "#    5   mate2 < mate1   mate2:forward, mate1:forward  \n"
+    "#    6   mate2 < mate1   mate2:forward, mate1:reverse  \n"
+    "#    7   mate2 < mate1   mate2:reverse, mate1:forward  \n"
+    "#    8   mate2 < mate1   mate2:reverse, mate1:reverse  \n"
+    "# -----------------------------------------------------\n";
+
 // --------------------------------------------------------------------------
 // unique readname file constants
 // --------------------------------------------------------------------------
@@ -731,9 +745,13 @@ void ResolveTool::StatsFileWriter::WriteHeader(void) {
                   << BAMTOOLS_VERSION_BUILD;
 
     // # bamtools resolve (vX.Y.Z)
+    // #
+    // # MODEL DESCRIPTION - see above for actual text
     // \n
 
     m_stream << COMMENT_CHAR << " bamtools resolve (" << versionStream.str() << ")" << endl
+             << COMMENT_CHAR << endl
+             << MODEL_DESCRIPTION
              << endl;
 }
 

diff --git a/src/toolkit/bamtools_split.cpp b/src/toolkit/bamtools_split.cpp
index e6602a9e590e250d3dd204fea990db579db20db4..6425e9529a63da8cb2f99bcab8ead6b0c29f8352 100644 (file)
--- a/src/toolkit/bamtools_split.cpp
+++ b/src/toolkit/bamtools_split.cpp
@@ -2,7 +2,7 @@
 // bamtools_split.cpp (c) 2010 Derek Barnett, Erik Garrison
 // Marth Lab, Department of Biology, Boston College
 // ---------------------------------------------------------------------------
-// Last modified: 8 December 2011 (DB)
+// Last modified: 24 July 2013 (DB)
 // ---------------------------------------------------------------------------
 // Splits a BAM file on user-specified property, creating a new BAM output
 // file for each value found
@@ -33,6 +33,7 @@ static const string SPLIT_UNMAPPED_TOKEN  = ".UNMAPPED";
 static const string SPLIT_PAIRED_TOKEN    = ".PAIRED_END";
 static const string SPLIT_SINGLE_TOKEN    = ".SINGLE_END";
 static const string SPLIT_REFERENCE_TOKEN = ".REF_";
+static const string SPLIT_TAG_TOKEN       = ".TAG_";
 
 string GetTimestampString(void) {
 
@@ -70,6 +71,7 @@ struct SplitTool::SplitSettings {
     bool HasInputFilename;
     bool HasCustomOutputStub;
     bool HasCustomRefPrefix;
+    bool HasCustomTagPrefix;
     bool IsSplittingMapped;
     bool IsSplittingPaired;
     bool IsSplittingReference;
@@ -78,6 +80,7 @@ struct SplitTool::SplitSettings {
     // string args
     string CustomOutputStub;
     string CustomRefPrefix;
+    string CustomTagPrefix;
     string InputFilename;
     string TagToSplit;
     
@@ -86,12 +89,14 @@ struct SplitTool::SplitSettings {
         : HasInputFilename(false)
         , HasCustomOutputStub(false)
         , HasCustomRefPrefix(false)
+        , HasCustomTagPrefix(false)
         , IsSplittingMapped(false)
         , IsSplittingPaired(false)
         , IsSplittingReference(false)
         , IsSplittingTag(false)
         , CustomOutputStub("")
         , CustomRefPrefix("")
+        , CustomTagPrefix("")
         , InputFilename(Options::StandardIn())
         , TagToSplit("")
     { } 
@@ -454,6 +459,16 @@ bool SplitTool::SplitToolPrivate::SplitTagImpl(BamAlignment& al) {
     WriterMap outputFiles;
     WriterMapIterator writerIter;
 
+    // determine tag prefix
+    string tagPrefix = SPLIT_TAG_TOKEN;
+    if ( m_settings->HasCustomTagPrefix )
+        tagPrefix = m_settings->CustomTagPrefix;
+
+    // make sure prefix starts with '.'
+    const size_t dotFound = tagPrefix.find('.');
+    if ( dotFound != 0 )
+        tagPrefix = string(".") + tagPrefix;
+
     // local variables
     const string tag = m_settings->TagToSplit;
     BamWriter* writer;
@@ -464,7 +479,7 @@ bool SplitTool::SplitToolPrivate::SplitTagImpl(BamAlignment& al) {
     if ( al.GetTag(tag, currentValue) ) {
       
         // open new BamWriter, save first alignment
-        outputFilenameStream << m_outputFilenameStub << ".TAG_" << tag << "_" << currentValue << ".bam";
+        outputFilenameStream << m_outputFilenameStub << tagPrefix << tag << "_" << currentValue << ".bam";
         writer = new BamWriter;
         if ( !writer->Open(outputFilenameStream.str(), m_header, m_references) ) {
             cerr << "bamtools split ERROR: could not open " << outputFilenameStream.str()
@@ -493,7 +508,7 @@ bool SplitTool::SplitToolPrivate::SplitTagImpl(BamAlignment& al) {
         if ( writerIter == outputFiles.end() ) {
         
             // open new BamWriter
-            outputFilenameStream << m_outputFilenameStub << ".TAG_" << tag << "_" << currentValue << ".bam";
+            outputFilenameStream << m_outputFilenameStub << tagPrefix << tag << "_" << currentValue << ".bam";
             writer = new BamWriter;
             if ( !writer->Open(outputFilenameStream.str(), m_header, m_references) ) {
                 cerr << "bamtool split ERROR: could not open " << outputFilenameStream.str()
@@ -542,6 +557,8 @@ SplitTool::SplitTool(void)
     Options::AddValueOption("-in",   "BAM filename",  "the input BAM file",  "", m_settings->HasInputFilename,  m_settings->InputFilename,  IO_Opts, Options::StandardIn());
     Options::AddValueOption("-refPrefix", "string", "custom prefix for splitting by references. Currently files end with REF_<refName>.bam. This option allows you to replace \"REF_\" with a prefix of your choosing.", "",
                             m_settings->HasCustomRefPrefix, m_settings->CustomRefPrefix, IO_Opts);
+    Options::AddValueOption("-tagPrefix", "string", "custom prefix for splitting by tags. Current files end with TAG_<tagname>_<tagvalue>.bam. This option allows you to replace \"TAG_\" with a prefix of your choosing.", "",
+                            m_settings->HasCustomTagPrefix, m_settings->CustomTagPrefix, IO_Opts);
     Options::AddValueOption("-stub", "filename stub", "prefix stub for output BAM files (default behavior is to use input filename, without .bam extension, as stub). If input is stdin and no stub provided, a timestamp is generated as the stub.", "",
                             m_settings->HasCustomOutputStub, m_settings->CustomOutputStub, IO_Opts);