ms_spectral_lib_file.hpp

/*
##############################################################################
# File: ms_nist_msp.hpp                                                      #
# Mascot Parser toolkit                                                      #
# Encapsulates spectral library files (e.g. msp, sptext)                     #
#                                                                            #
##############################################################################
# COPYRIGHT NOTICE                                                           #
# Copyright 2015 Matrix Science Limited  All Rights Reserved.                #
#                                                                            #
##############################################################################
#    $Source: parser/inc/ms_spectral_lib_file.hpp $
#    $Author: dcreasy@matrixscience.com $ 
#      $Date: 2019-10-23 11:32:40 +0100 $ 
#  $Revision: e1e932b0e014aa84a197578948d47baeca33fd07 | MSPARSER_REL_3_0_0-2024-09-24-0-g93ebaeb4f4 $
##############################################################################
*/
 
#ifndef MS_SPECTRAL_LIB_FILE
#define MS_SPECTRAL_LIB_FILE
 
 
// Includes from the standard template library
 
#include <map>
#include <set>
#include <string>
 
 
namespace matrix_science {
  
 
    class MS_MASCOTRESFILE_API ms_spectral_lib_file : public ms_errors
    {
    public:
        ms_spectral_lib_file(const char * fileName, const char * regexForAccession, const char * cdbFileName = 0);
 
#ifndef SWIG
        ms_spectral_lib_file(const char * fileName, const char * regexForAccession, const char * cdbFileName, const std::map<std::string, std::string> & modificationAliases);
#endif
 
        ~ms_spectral_lib_file();
        
        int getNumEntries() const;
        
        int getNumResidues() const;
 
        std::vector<int> findEntries(const char * sequence, const char * checksum = 0, const char * accession = 0, const char * mods=0) const;
        
        ms_spectral_lib_entry getEntryFromNumber(const int number) const;
 
        std::vector<std::string> getEntryFromNumberAsText(const int number) const;
 
 
        std::string getAccessionFromNumber(const int number) const;
 
        std::string getChecksumFromNumber(const int number) const;
 
        std::string getSequenceFromNumber(const int number) const;
 
        std::string getModsFromNumber(const int number) const;
 
        double getPrecursorMZFromNumber(const int number) const;
 
        int getPrecursorChargeFromNumber(const int number) const;
 
        std::vector<std::string> getAllMods() const;
 
        ms_spectral_lib::FILE_FORMAT getFormat() const;
 
        bool saveAs(const char * fileName, 
                    const bool   replaceProteinName  = true, 
                    ms_spectral_lib::FILE_FORMAT fileFormat = ms_spectral_lib::FORMAT_NIST_MSP,
                    const int    startNumber = 1,
                    const int    endNumber = -1,
                    const ms_spectral_lib_entry::WHAT_TO_ANNOTATE whatToAnnotate = ms_spectral_lib_entry::ANNOTATE_REPLACE_QUESTION_MARKS,
                    const double annotateTol = 0.6,
                    const char * annotateTolu = "Da",
                    const ms_umod_configfile * unimod = 0) const;
 
        std::string getStatsInformation() const;
 
        std::string getFileName() const;
 
        long getQmatch(double minMz, double maxMz) const;
 
        bool verifyThatModsAreInUnimod(const ms_umod_configfile & unimod);
 
    private:
        //Do not copy this object
        ms_spectral_lib_file(const ms_spectral_lib_file & src);
        ms_spectral_lib_file& operator=(const ms_spectral_lib_file & right);
 
        struct sequentialIndex_t {
            std::string accession;
            std::string checksum;
            std::string sequence;
            std::string mods;
            double      precursorMZ;
            int         precursorCharge;
            OFFSET64_T  fileOffset;
        };
        void initialise(const char * fileName, const char * regexForAccession, const char * cdbFileName, const std::map<std::string, std::string> & aliases);
        bool readFile(int recordNumber) const;
        void getline(std::string & str) const;
        ms_errs::err getEntryAsText(const int number, std::vector<std::string> & lines, std::string & errorMsg) const;
 
        void saveIndexes(const  ms_spectral_lib_entry & msp, 
                         const  OFFSET64_T          fileOffset);
        OFFSET64_T getFileOffsetFromId(int number) const;
        const char * getSpectrumDelimiter() const;
 
        int applyModAliases(const std::map<std::string, std::string> & aliases);
        void replaceModAliases(std::string & comment, const ms_spectral_lib_entry & spect) const;
 
        // the following are always valid - with or without a cdb file
                std::string  fileName_;
        mutable int          numEntries_;  // May be less than the total in the file if it's being read on demand
        mutable int          numResidues_; // May be less than the total in the file if it's being read on demand
                ms_parserule parseRule_;
        mutable bool         wholeFileRead_;
                FILE       * ifs_;
        mutable ms_spectral_lib::FILE_FORMAT  fileFormat_;
 
        // the following are valid when creating or using the cdb file
        ms_tinycdb * cdb_;
 
        // the following are only valid when no cdb specified *or* when *first creating* the cdb file
        mutable OFFSET64_T offsetToNextRecord_;
        typedef std::map<std::string, std::vector<int> > strLookup_t;  // e.g. sequence => vector of spectra
        std::vector<strLookup_t> lookup_;
        std::vector<sequentialIndex_t> sequentialIndex_;
        typedef std::map<std::string, std::string> modMap_t;  // nameInMsp => nameInUnimod (or empty if no alias)
        mutable modMap_t modsMap_;
 
        typedef std::multiset<double> precursors_t;
        mutable precursors_t precursors_;
 
    }; // class ms_spectral_lib_file
 // end of tools_group
} // namespace matrix_science
 
#endif // MS_SPECTRAL_LIB_FILE
 
/*------------------------------- End of File -------------------------------*/