ngs_server 0.1

data/ext/bamtools/src/api/SamConstants.h

@@ -0,0 +1,95 @@
+// ***************************************************************************
+// SamConstants.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides constants for SAM header
+// ***************************************************************************
+
+#ifndef SAM_CONSTANTS_H
+#define SAM_CONSTANTS_H
+
+#include <api/api_global.h>
+#include <string>
+
+namespace BamTools {
+namespace Constants {
+
+// basic char constants used in SAM format
+const char SAM_COLON  = ':';
+const char SAM_EQUAL  = '=';
+const char SAM_PERIOD = '.';
+const char SAM_STAR   = '*';
+const char SAM_TAB    = '\t';
+const std::string SAM_DIGITS = "0123456789";
+
+// HD entries
+const std::string SAM_HD_BEGIN_TOKEN    = "@HD";
+const std::string SAM_HD_VERSION_TAG    = "VN";
+const std::string SAM_HD_SORTORDER_TAG  = "SO";
+const std::string SAM_HD_GROUPORDER_TAG = "GO";
+
+// SQ entries
+const std::string SAM_SQ_BEGIN_TOKEN    = "@SQ";
+const std::string SAM_SQ_ASSEMBLYID_TAG = "AS";
+const std::string SAM_SQ_CHECKSUM_TAG   = "M5";
+const std::string SAM_SQ_LENGTH_TAG     = "LN";
+const std::string SAM_SQ_NAME_TAG       = "SN";
+const std::string SAM_SQ_SPECIES_TAG    = "SP";
+const std::string SAM_SQ_URI_TAG        = "UR";
+
+// RG entries
+const std::string SAM_RG_BEGIN_TOKEN             = "@RG";
+const std::string SAM_RG_DESCRIPTION_TAG         = "DS";
+const std::string SAM_RG_FLOWORDER_TAG           = "FO";
+const std::string SAM_RG_ID_TAG                  = "ID";
+const std::string SAM_RG_KEYSEQUENCE_TAG         = "KS";
+const std::string SAM_RG_LIBRARY_TAG             = "LB";
+const std::string SAM_RG_PLATFORMUNIT_TAG        = "PU";
+const std::string SAM_RG_PREDICTEDINSERTSIZE_TAG = "PI";
+const std::string SAM_RG_PRODUCTIONDATE_TAG      = "DT";
+const std::string SAM_RG_PROGRAM_TAG             = "PG";
+const std::string SAM_RG_SAMPLE_TAG              = "SM";
+const std::string SAM_RG_SEQCENTER_TAG           = "CN";
+const std::string SAM_RG_SEQTECHNOLOGY_TAG       = "PL";
+
+// PG entries
+const std::string SAM_PG_BEGIN_TOKEN         = "@PG";
+const std::string SAM_PG_COMMANDLINE_TAG     = "CL";
+const std::string SAM_PG_ID_TAG              = "ID";
+const std::string SAM_PG_NAME_TAG            = "PN";
+const std::string SAM_PG_PREVIOUSPROGRAM_TAG = "PP";
+const std::string SAM_PG_VERSION_TAG         = "VN";
+
+// CO entries
+const std::string SAM_CO_BEGIN_TOKEN = "@CO";
+
+// HD:SO values
+const std::string SAM_HD_SORTORDER_COORDINATE = "coordinate";
+const std::string SAM_HD_SORTORDER_QUERYNAME  = "queryname";
+const std::string SAM_HD_SORTORDER_UNKNOWN    = "unknown";
+const std::string SAM_HD_SORTORDER_UNSORTED   = "unsorted";
+
+// HD:GO values
+const std::string SAM_HD_GROUPORDER_NONE      = "none";
+const std::string SAM_HD_GROUPORDER_QUERY     = "query";
+const std::string SAM_HD_GROUPORDER_REFERENCE = "reference";
+
+// SQ:LN values
+const unsigned int SAM_SQ_LENGTH_MIN = 1;
+const unsigned int SAM_SQ_LENGTH_MAX = 536870911; // 2^29 - 1
+
+// RG:PL values
+const std::string SAM_RG_SEQTECHNOLOGY_CAPILLARY  = "CAPILLARY";
+const std::string SAM_RG_SEQTECHNOLOGY_HELICOS    = "HELICOS";
+const std::string SAM_RG_SEQTECHNOLOGY_ILLUMINA   = "ILLUMINA";
+const std::string SAM_RG_SEQTECHNOLOGY_IONTORRENT = "IONTORRENT";
+const std::string SAM_RG_SEQTECHNOLOGY_LS454      = "LS454";
+const std::string SAM_RG_SEQTECHNOLOGY_PACBIO     = "PACBIO";
+const std::string SAM_RG_SEQTECHNOLOGY_SOLID      = "SOLID";
+
+} // namespace Constants
+} // namespace BamTools
+
+#endif // SAM_CONSTANTS_H

data/ext/bamtools/src/api/SamHeader.cpp

@@ -0,0 +1,184 @@
+// ***************************************************************************
+// SamHeader.cpp (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM header data fields.
+// ***************************************************************************
+
+#include <api/SamConstants.h>
+#include <api/SamHeader.h>
+#include <api/internal/SamFormatParser_p.h>
+#include <api/internal/SamFormatPrinter_p.h>
+#include <api/internal/SamHeaderValidator_p.h>
+using namespace BamTools;
+using namespace BamTools::Internal;
+using namespace std;
+
+/*! \struct BamTools::SamHeader
+    \brief Represents the SAM-formatted text header that is part of the BAM file header.
+
+    Provides direct read/write access to the SAM header data fields.
+
+    \sa \samSpecURL
+*/
+/*! \var SamHeader::Version
+    \brief corresponds to \@HD VN:\<Version\>
+
+    Required for valid SAM header, if @HD record is present.
+*/
+/*! \var SamHeader::SortOrder
+    \brief corresponds to \@HD SO:\<SortOrder\>
+*/
+/*! \var SamHeader::GroupOrder
+    \brief corresponds to \@HD GO:\<GroupOrder\>
+*/
+/*! \var SamHeader::Sequences
+    \brief corresponds to \@SQ entries
+    \sa SamSequence, SamSequenceDictionary
+*/
+/*! \var SamHeader::ReadGroups
+    \brief corresponds to \@RG entries
+    \sa SamReadGroup, SamReadGroupDictionary
+*/
+/*! \var SamHeader::ProgramName
+    \brief corresponds to \@PG ID:\<ProgramName\>
+*/
+/*! \var SamHeader::ProgramVersion
+    \brief corresponds to \@PG VN:\<ProgramVersion\>
+*/
+/*! \var SamHeader::ProgramCommandLine
+    \brief corresponds to \@PG CL:\<ProgramCommandLine\>
+*/
+/*! \var SamHeader::Comments
+    \brief corresponds to \@CO entries
+*/
+
+/*! \fn SamHeader::SamHeader(const std::string& headerText = "")
+    \brief constructor
+*/
+SamHeader::SamHeader(const std::string& headerText)
+    : Version("")
+    , SortOrder(Constants::SAM_HD_SORTORDER_UNKNOWN)
+    , GroupOrder("")
+{
+    SamFormatParser parser(*this);
+    parser.Parse(headerText);
+}
+
+/*! \fn SamHeader::SamHeader(const SamHeader& other)
+    \brief copy constructor
+*/
+SamHeader::SamHeader(const SamHeader& other)
+    : Version(other.Version)
+    , SortOrder(other.SortOrder)
+    , GroupOrder(other.GroupOrder)
+    , Sequences(other.Sequences)
+    , ReadGroups(other.ReadGroups)
+    , Programs(other.Programs)
+{ }
+
+/*! \fn SamHeader::~SamHeader(void)
+    \brief destructor
+*/
+SamHeader::~SamHeader(void) { }
+
+/*! \fn void SamHeader::Clear(void)
+    \brief Clears all header contents.
+*/
+void SamHeader::Clear(void) {
+    Version.clear();
+    SortOrder.clear();
+    GroupOrder.clear();
+    Sequences.Clear();
+    ReadGroups.Clear();
+    Programs.Clear();
+    Comments.clear();
+}
+
+/*! \fn bool SamHeader::HasVersion(void) const
+    \brief Returns \c true if header contains \@HD ID:\<Version\>
+*/
+bool SamHeader::HasVersion(void) const {
+    return (!Version.empty());
+}
+
+/*! \fn bool SamHeader::HasSortOrder(void) const
+    \brief Returns \c true if header contains \@HD SO:\<SortOrder\>
+*/
+bool SamHeader::HasSortOrder(void) const {
+    return (!SortOrder.empty());
+}
+
+/*! \fn bool SamHeader::HasGroupOrder(void) const
+    \brief Returns \c true if header contains \@HD GO:\<GroupOrder\>
+*/
+bool SamHeader::HasGroupOrder(void) const {
+    return (!GroupOrder.empty());
+}
+
+/*! \fn bool SamHeader::HasSequences(void) const
+    \brief Returns \c true if header contains any \@SQ entries
+*/
+bool SamHeader::HasSequences(void) const {
+    return (!Sequences.IsEmpty());
+}
+
+/*! \fn bool SamHeader::HasReadGroups(void) const
+    \brief Returns \c true if header contains any \@RG entries
+*/
+bool SamHeader::HasReadGroups(void) const {
+    return (!ReadGroups.IsEmpty());
+}
+
+/*! \fn bool SamHeader::HasPrograms(void) const
+    \brief Returns \c true if header contains any \@PG entries
+*/
+bool SamHeader::HasPrograms(void) const {
+    return (!Programs.IsEmpty());
+}
+
+/*! \fn bool SamHeader::HasComments(void) const
+    \brief Returns \c true if header contains any \@CO entries
+*/
+bool SamHeader::HasComments(void) const {
+    return (!Comments.empty());
+}
+
+/*! \fn bool SamHeader::IsValid(bool verbose = false) const
+    \brief Checks header contents for required data and proper formatting.
+    \param verbose If set to true, validation errors & warnings will be printed to stderr.
+                   Otherwise, output is suppressed and only validation check occurs.
+    \return \c true if SAM header is well-formed
+*/
+bool SamHeader::IsValid(bool verbose) const {
+    SamHeaderValidator validator(*this);
+    return validator.Validate(verbose);
+}
+
+/*! \fn void SamHeader::SetHeaderText(const std::string& headerText)
+    \brief Replaces header contents with \a headerText.
+    \param headerText SAM formatted-text that will be parsed into data fields
+*/
+void SamHeader::SetHeaderText(const std::string& headerText) {
+
+    // clear prior data
+    Clear();
+
+    // parse header text into data
+    SamFormatParser parser(*this);
+    parser.Parse(headerText);
+}
+
+/*! \fn std::string SamHeader::ToString(void) const
+    \brief Converts data fields to SAM-formatted text.
+
+    Applies any local modifications made since creating this object or calling SetHeaderText().
+
+    \return SAM-formatted header text
+*/
+string SamHeader::ToString(void) const {
+    SamFormatPrinter printer(*this);
+    return printer.ToString();
+}

data/ext/bamtools/src/api/SamHeader.h

@@ -0,0 +1,68 @@
+// ***************************************************************************
+// SamHeader.h (c) 2010 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 18 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM header data fields.
+// ***************************************************************************
+
+#ifndef SAM_HEADER_H
+#define SAM_HEADER_H
+
+#include <api/api_global.h>
+#include <api/SamProgramChain.h>
+#include <api/SamReadGroupDictionary.h>
+#include <api/SamSequenceDictionary.h>
+#include <string>
+#include <vector>
+
+namespace BamTools {
+
+struct API_EXPORT SamHeader {
+
+    // ctor & dtor
+    SamHeader(const std::string& headerText = "");
+    SamHeader(const SamHeader& other);
+    ~SamHeader(void);
+
+    // query/modify entire SamHeader
+    void Clear(void);                                   // clears all header contents
+    bool IsValid(bool verbose = false) const;           // returns true if SAM header is well-formed
+    void SetHeaderText(const std::string& headerText);  // replaces data fields with contents of SAM-formatted text
+    std::string ToString(void) const;                   // returns the printable, SAM-formatted header text
+
+    // convenience query methods
+    bool HasVersion(void) const;            // returns true if header contains format version entry
+    bool HasSortOrder(void) const;          // returns true if header contains sort order entry
+    bool HasGroupOrder(void) const;         // returns true if header contains group order entry
+    bool HasSequences(void) const;          // returns true if header contains any sequence entries
+    bool HasReadGroups(void) const;         // returns true if header contains any read group entries
+    bool HasPrograms(void) const;           // returns true if header contains any program record entries
+    bool HasComments(void) const;           // returns true if header contains comments
+
+    // --------------
+    // data members
+    // --------------
+
+    // header metadata (@HD line)
+    std::string Version;                    // VN:<Version>  *Required for valid SAM header, if @HD record is present*
+    std::string SortOrder;                  // SO:<SortOrder>
+    std::string GroupOrder;                 // GO:<GroupOrder>
+
+    // header sequences (@SQ entries)
+    SamSequenceDictionary Sequences;
+
+    // header read groups (@RG entries)
+    SamReadGroupDictionary ReadGroups;
+
+    // header program data (@PG entries)
+    SamProgramChain Programs;
+
+    // header comments (@CO entries)
+    std::vector<std::string> Comments;
+};
+
+} // namespace BamTools
+
+#endif // SAM_HEADER_H

data/ext/bamtools/src/api/SamProgram.cpp

@@ -0,0 +1,139 @@
+// ***************************************************************************
+// SamProgram.cpp (c) 2011 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM header program records.
+// ***************************************************************************
+
+#include <api/SamProgram.h>
+using namespace BamTools;
+using namespace std;
+
+/*! \struct BamTools::SamProgram
+    \brief Represents a SAM program record.
+
+    Provides direct read/write access to the SAM header program records.
+
+    \sa \samSpecURL
+*/
+/*! \var SamProgram::CommandLine
+    \brief corresponds to \@PG CL:\<CommandLine\>
+*/
+/*! \var SamProgram::ID
+    \brief corresponds to \@PG ID:\<ID\>
+
+    Required for valid SAM header.
+*/
+/*! \var SamProgram::Name
+    \brief corresponds to \@PG PN:\<Name\>
+*/
+/*! \var SamProgram::PreviousProgramID
+    \brief corresponds to \@PG PP:\<PreviousProgramID\>
+*/
+/*! \var SamProgram::Version
+    \brief corresponds to \@PG VN:\<Version\>
+*/
+/*! \var SamProgram::NextProgramID
+    \internal
+    Holds ID of the "next" program record in a SamProgramChain
+*/
+
+/*! \fn SamProgram::SamProgram(void)
+    \brief default constructor
+*/
+SamProgram::SamProgram(void)
+    : CommandLine("")
+    , ID("")
+    , Name("")
+    , PreviousProgramID("")
+    , Version("")
+    , NextProgramID("")
+{ }
+
+/*! \fn SamProgram::SamProgram(const std::string& id)
+    \brief constructs program record with \a id
+
+    \param id desired program record ID
+*/
+SamProgram::SamProgram(const std::string& id)
+    : CommandLine("")
+    , ID(id)
+    , Name("")
+    , PreviousProgramID("")
+    , Version("")
+    , NextProgramID("")
+{ }
+
+/*! \fn SamProgram::SamProgram(const SamProgram& other)
+    \brief copy constructor
+*/
+SamProgram::SamProgram(const SamProgram& other)
+    : CommandLine(other.CommandLine)
+    , ID(other.ID)
+    , Name(other.Name)
+    , PreviousProgramID(other.PreviousProgramID)
+    , Version(other.Version)
+    , NextProgramID(other.NextProgramID)
+{ }
+
+/*! \fn SamProgram::~SamProgram(void)
+    \brief destructor
+*/
+SamProgram::~SamProgram(void) { }
+
+/*! \fn void SamProgram::Clear(void)
+    \brief Clears all data fields.
+*/
+void SamProgram::Clear(void) {
+    CommandLine.clear();
+    ID.clear();
+    Name.clear();
+    PreviousProgramID.clear();
+    Version.clear();
+    NextProgramID.clear();
+}
+
+/*! \fn bool SamProgram::HasCommandLine(void) const
+    \brief Returns \c true if program record contains \@PG: CL:\<CommandLine\>
+*/
+bool SamProgram::HasCommandLine(void) const {
+    return (!CommandLine.empty());
+}
+
+/*! \fn bool SamProgram::HasID(void) const
+    \brief Returns \c true if program record contains \@PG: ID:\<ID\>
+*/
+bool SamProgram::HasID(void) const {
+    return (!ID.empty());
+}
+
+/*! \fn bool SamProgram::HasName(void) const
+    \brief Returns \c true if program record contains \@PG: PN:\<Name\>
+*/
+bool SamProgram::HasName(void) const {
+    return (!Name.empty());
+}
+
+/*! \fn bool SamProgram::HasNextProgramID(void) const
+    \internal
+    \return true if program has a "next" record in a SamProgramChain
+*/
+bool SamProgram::HasNextProgramID(void) const {
+    return (!NextProgramID.empty());
+}
+
+/*! \fn bool SamProgram::HasPreviousProgramID(void) const
+    \brief Returns \c true if program record contains \@PG: PP:\<PreviousProgramID\>
+*/
+bool SamProgram::HasPreviousProgramID(void) const {
+    return (!PreviousProgramID.empty());
+}
+
+/*! \fn bool SamProgram::HasVersion(void) const
+    \brief Returns \c true if program record contains \@PG: VN:\<Version\>
+*/
+bool SamProgram::HasVersion(void) const {
+    return (!Version.empty());
+}

data/ext/bamtools/src/api/SamProgram.h

@@ -0,0 +1,61 @@
+// ***************************************************************************
+// SamProgram.h (c) 2011 Derek Barnett
+// Marth Lab, Department of Biology, Boston College
+// ---------------------------------------------------------------------------
+// Last modified: 19 April 2011 (DB)
+// ---------------------------------------------------------------------------
+// Provides direct read/write access to the SAM header program records.
+// ***************************************************************************
+
+#ifndef SAM_PROGRAM_H
+#define SAM_PROGRAM_H
+
+#include "api/api_global.h"
+#include <string>
+
+namespace BamTools {
+
+class SamProgramChain;
+
+struct API_EXPORT SamProgram {
+
+    // ctor & dtor
+    SamProgram(void);
+    SamProgram(const std::string& id);
+    SamProgram(const SamProgram& other);
+    ~SamProgram(void);
+
+    // query/modify entire program record
+    void Clear(void);                           // clears all data fields
+
+    // convenience query methods
+    bool HasCommandLine(void) const;            // returns true if program record has a command line entry
+    bool HasID(void) const;                     // returns true if program record has an ID
+    bool HasName(void) const;                   // returns true if program record has a name
+    bool HasPreviousProgramID(void) const;      // returns true if program record has a 'previous program ID'
+    bool HasVersion(void) const;                // returns true if program record has a version
+
+    // data members
+    std::string CommandLine;                    // CL:<CommandLine>
+    std::string ID;                             // ID:<ID>          *Required for valid SAM header*
+    std::string Name;                           // PN:<Name>
+    std::string PreviousProgramID;              // PP:<PreviousProgramID>
+    std::string Version;                        // VN:<Version>
+
+    // internal (non-standard) methods & fields
+    private:
+        bool HasNextProgramID(void) const;
+        std::string NextProgramID;
+        friend class BamTools::SamProgramChain;
+};
+
+/*! \fn bool operator==(const SamProgram& lhs, const SamProgram& rhs)
+    \brief tests equality by comparing program IDs
+*/
+API_EXPORT inline bool operator==(const SamProgram& lhs, const SamProgram& rhs) {
+    return lhs.ID == rhs.ID;
+}
+
+} // namespace BamTools
+
+#endif // SAM_PROGRAM_H