ngs_server 0.1

data/ext/bamtools/src/toolkit/bamtools_filter.h

@@ -0,0 +1,37 @@
+// ***************************************************************************
+// bamtools_filter.h (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 28 August 2010
+// ---------------------------------------------------------------------------
+// Filters BAM file(s) according to some user-specified criteria
+// ***************************************************************************
+
+#ifndef BAMTOOLS_FILTER_H
+#define BAMTOOLS_FILTER_H
+
+#include "bamtools_tool.h"
+
+namespace BamTools {
+  
+class FilterTool : public AbstractTool {
+  
+    public:
+        FilterTool(void);
+        ~FilterTool(void);
+  
+    public:
+        int Help(void);
+        int Run(int argc, char* argv[]); 
+        
+    private:
+        struct FilterSettings;
+        FilterSettings* m_settings;
+        
+        struct FilterToolPrivate;
+        FilterToolPrivate* m_impl;
+};
+  
+} // namespace BamTools
+
+#endif // BAMTOOLS_FILTER_H

data/ext/bamtools/src/toolkit/bamtools_header.cpp

@@ -0,0 +1,122 @@
+// ***************************************************************************
+// bamtools_header.cpp (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 7 April 2011
+// ---------------------------------------------------------------------------
+// Prints the SAM-style header from a single BAM file ( or merged header from
+// multiple BAM files) to stdout
+// ***************************************************************************
+
+#include "bamtools_header.h"
+
+#include <api/BamMultiReader.h>
+#include <utils/bamtools_options.h>
+using namespace BamTools;
+
+#include <iostream>
+#include <string>
+#include <vector>
+using namespace std;
+  
+// ---------------------------------------------
+// HeaderSettings implementation
+
+struct HeaderTool::HeaderSettings {
+
+    // flags
+    bool HasInputBamFilename;
+
+    // filenames
+    vector<string> InputFiles;
+    
+    // constructor
+    HeaderSettings(void)
+        : HasInputBamFilename(false)
+    { }
+};  
+
+struct HeaderTool::HeaderToolPrivate {
+
+    // ctor & dtor
+    public:
+        HeaderToolPrivate(HeaderTool::HeaderSettings* settings)
+            : m_settings(settings)
+        { }
+
+        ~HeaderToolPrivate(void) { }
+
+    // interface
+    public:
+        bool Run(void);
+
+    // data members
+    private:
+        HeaderTool::HeaderSettings* m_settings;
+};
+
+bool HeaderTool::HeaderToolPrivate::Run(void) {
+
+    // set to default input if none provided
+    if ( !m_settings->HasInputBamFilename )
+        m_settings->InputFiles.push_back(Options::StandardIn());
+
+    // attemp to open BAM files
+    BamMultiReader reader;
+    if ( !reader.Open(m_settings->InputFiles) ) {
+        cerr << "bamtools header ERROR: could not open BAM file(s) for reading... Aborting." << endl;
+        return false;
+    }
+
+    // dump (merged) header contents to stdout
+    cout << reader.GetHeaderText() << endl;
+
+    // clean up & exit
+    reader.Close();
+    return true;
+}
+
+// ---------------------------------------------
+// HeaderTool implementation
+
+HeaderTool::HeaderTool(void)
+    : AbstractTool()
+    , m_settings(new HeaderSettings)
+    , m_impl(0)
+{
+    // set program details
+    Options::SetProgramInfo("bamtools header", "prints header from BAM file(s)", "[-in <filename> -in <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::StandardIn());
+}
+
+HeaderTool::~HeaderTool(void) {
+
+    delete m_settings;
+    m_settings = 0;
+
+    delete m_impl;
+    m_impl = 0;
+}
+
+int HeaderTool::Help(void) {
+    Options::DisplayHelp();
+    return 0;
+}
+
+int HeaderTool::Run(int argc, char* argv[]) {
+  
+    // parse command line arguments
+    Options::Parse(argc, argv, 1);
+  
+    // initialize HeaderTool with settings
+    m_impl = new HeaderToolPrivate(m_settings);
+
+    // run HeaderTool, return success/fail
+    if ( m_impl->Run() )
+        return 0;
+    else
+        return 1;
+}

data/ext/bamtools/src/toolkit/bamtools_header.h

@@ -0,0 +1,38 @@
+// ***************************************************************************
+// bamtools_header.h (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 7 April 2011
+// ---------------------------------------------------------------------------
+// Prints the SAM-style header from a single BAM file ( or merged header from
+// multiple BAM files) to stdout
+// ***************************************************************************
+
+#ifndef BAMTOOLS_HEADER_H
+#define BAMTOOLS_HEADER_H
+
+#include "bamtools_tool.h"
+
+namespace BamTools {
+  
+class HeaderTool : public AbstractTool {
+  
+    public:
+        HeaderTool(void);
+        ~HeaderTool(void);
+  
+    public:
+        int Help(void);
+        int Run(int argc, char* argv[]); 
+        
+    private:
+        struct HeaderSettings;
+        HeaderSettings* m_settings;
+
+        struct HeaderToolPrivate;
+        HeaderToolPrivate* m_impl;
+};
+  
+} // namespace BamTools
+
+#endif // BAMTOOLS_HEADER_H

data/ext/bamtools/src/toolkit/bamtools_index.cpp

@@ -0,0 +1,126 @@
+// ***************************************************************************
+// bamtools_index.cpp (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 7 April 2011
+// ---------------------------------------------------------------------------
+// Creates a BAM index file
+// ***************************************************************************
+
+#include "bamtools_index.h"
+
+#include <api/BamReader.h>
+#include <utils/bamtools_options.h>
+using namespace BamTools;
+
+#include <iostream>
+#include <string>
+using namespace std;
+
+// ---------------------------------------------
+// IndexSettings implementation
+
+struct IndexTool::IndexSettings {
+
+    // flags
+    bool HasInputBamFilename;
+    bool IsUsingBamtoolsIndex;
+
+    // filenames
+    string InputBamFilename;
+    
+    // constructor
+    IndexSettings(void)
+        : HasInputBamFilename(false)
+        , IsUsingBamtoolsIndex(false)
+        , InputBamFilename(Options::StandardIn())
+    { }
+};  
+
+// ---------------------------------------------
+// IndexToolPrivate implementation
+
+struct IndexTool::IndexToolPrivate {
+
+    // ctor & dtor
+    public:
+        IndexToolPrivate(IndexTool::IndexSettings* settings)
+            : m_settings(settings)
+        { }
+
+        ~IndexToolPrivate(void) { }
+
+    // interface
+    public:
+        bool Run(void);
+
+    // data members
+    private:
+        IndexTool::IndexSettings* m_settings;
+};
+
+bool IndexTool::IndexToolPrivate::Run(void) {
+
+    // open our BAM reader
+    BamReader reader;
+    if ( !reader.Open(m_settings->InputBamFilename) ) {
+        cerr << "bamtools index ERROR: could not open BAM file: "
+             << m_settings->InputBamFilename << endl;
+        return false;
+    }
+
+    // create index for BAM file
+    const BamIndex::IndexType type = ( m_settings->IsUsingBamtoolsIndex ? BamIndex::BAMTOOLS
+                                                                        : BamIndex::STANDARD );
+    reader.CreateIndex(type);
+
+    // clean & exit
+    reader.Close();
+    return true;
+}
+
+// ---------------------------------------------
+// IndexTool implementation
+
+IndexTool::IndexTool(void)
+    : AbstractTool()
+    , m_settings(new IndexSettings)
+    , m_impl(0)
+{
+    // set program details
+    Options::SetProgramInfo("bamtools index", "creates index for BAM file", "[-in <filename>] [-bti]");
+    
+    // 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::AddOption("-bti", "create (non-standard) BamTools index file (*.bti). Default behavior is to create standard BAM index (*.bai)", m_settings->IsUsingBamtoolsIndex, IO_Opts);
+}
+
+IndexTool::~IndexTool(void) {
+
+    delete m_settings;
+    m_settings = 0;
+
+    delete m_impl;
+    m_impl = 0;
+}
+
+int IndexTool::Help(void) {
+    Options::DisplayHelp();
+    return 0;
+}
+
+int IndexTool::Run(int argc, char* argv[]) {
+  
+    // parse command line arguments
+    Options::Parse(argc, argv, 1);
+    
+    // initialize IndexTool with settings
+    m_impl = new IndexToolPrivate(m_settings);
+
+    // run IndexTool, return success/fail
+    if ( m_impl->Run() )
+        return 0;
+    else
+        return 1;
+}

data/ext/bamtools/src/toolkit/bamtools_index.h

@@ -0,0 +1,37 @@
+// ***************************************************************************
+// bamtools_index.h (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 7 April 2011
+// ---------------------------------------------------------------------------
+// Creates a BAM index file
+// ***************************************************************************
+
+#ifndef BAMTOOLS_INDEX_H
+#define BAMTOOLS_INDEX_H
+
+#include "bamtools_tool.h"
+
+namespace BamTools {
+  
+class IndexTool : public AbstractTool {
+  
+    public:
+        IndexTool(void);
+        ~IndexTool(void);
+  
+    public:
+        int Help(void);
+        int Run(int argc, char* argv[]); 
+        
+    private:
+        struct IndexSettings;
+        IndexSettings* m_settings;
+
+        struct IndexToolPrivate;
+        IndexToolPrivate* m_impl;
+};
+  
+} // namespace BamTools
+
+#endif // BAMTOOLS_INDEX_H

data/ext/bamtools/src/toolkit/bamtools_merge.cpp

@@ -0,0 +1,221 @@
+// ***************************************************************************
+// bamtools_merge.cpp (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 7 April 2011
+// ---------------------------------------------------------------------------
+// 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>
+using namespace std;
+
+// ---------------------------------------------
+// MergeSettings implementation
+
+struct MergeTool::MergeSettings {
+
+    // flags
+    bool HasInputBamFilename;
+    bool HasOutputBamFilename;
+    bool IsForceCompression;
+    bool HasRegion;
+    
+    // filenames
+    vector<string> InputFiles;
+    
+    // other parameters
+    string OutputFilename;
+    string Region;
+    
+    // constructor
+    MergeSettings(void)
+        : HasInputBamFilename(false)
+        , HasOutputBamFilename(false)
+        , IsForceCompression(false)
+        , HasRegion(false)
+        , OutputFilename(Options::StandardOut())
+    { }
+};  
+
+// ---------------------------------------------
+// MergeToolPrivate implementation
+
+struct MergeTool::MergeToolPrivate {
+
+    // ctor & dtor
+    public:
+        MergeToolPrivate(MergeTool::MergeSettings* settings)
+            : m_settings(settings)
+        { }
+
+        ~MergeToolPrivate(void) { }
+
+    // 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->InputFiles.push_back(Options::StandardIn());
+
+    // opens the BAM files (by default without checking for indexes)
+    BamMultiReader reader;
+    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();
+
+    // 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;
+    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 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 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 << "bamtools merge ERROR: set region failed. Check that REGION describes a valid range"
+                         << endl;
+                    reader.Close();
+                    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) )
+                    {
+                        writer.SaveAlignment(al);
+                    }
+                }
+            }
+        }
+
+        // error parsing REGION string
+        else {
+            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 false;
+        }
+    }
+
+    // clean & exit
+    reader.Close();
+    writer.Close();
+    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> ...] [-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::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;
+}

data/ext/bamtools/src/toolkit/bamtools_merge.h

@@ -0,0 +1,37 @@
+// ***************************************************************************
+// bamtools_merge.h (c) 2010 Derek Barnett, Erik Garrison
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 7 April 2011
+// ---------------------------------------------------------------------------
+// Merges multiple BAM files into one
+// ***************************************************************************
+
+#ifndef BAMTOOLS_MERGE_H
+#define BAMTOOLS_MERGE_H
+
+#include "bamtools_tool.h"
+
+namespace BamTools {
+  
+class MergeTool : public AbstractTool {
+  
+    public:
+        MergeTool(void);
+        ~MergeTool(void);
+  
+    public:
+        int Help(void);
+        int Run(int argc, char* argv[]); 
+        
+    private:
+        struct MergeSettings;
+        MergeSettings* m_settings;
+
+        struct MergeToolPrivate;
+        MergeToolPrivate* m_impl;
+};
+  
+} // namespace BamTools
+
+#endif // BAMTOOLS_MERGE_H